[llvm] [CostModel/RISCV] Fix costs of [l]lrint, [l]lround (PR #145714)

[Ramkumar Ramachandra via llvm-commits]

https://github.com/artagnon created https://github.com/llvm/llvm-project/pull/145714

None

>From b2000119c737ae1c5fcba68e89f706f1cffb7226 Mon Sep 17 00:00:00 2001
From: Ramkumar Ramachandra <ramkumar.ramachandra at codasip.com>
Date: Wed, 25 Jun 2025 12:21:12 +0100
Subject: [PATCH 1/2] [CostModel] Pre-commit tests for [l]lround

---
 llvm/test/Analysis/CostModel/RISCV/fround.ll | 194 +++++++++++++++++++
 1 file changed, 194 insertions(+)

diff --git a/llvm/test/Analysis/CostModel/RISCV/fround.ll b/llvm/test/Analysis/CostModel/RISCV/fround.ll
index a0818d487d151..94c09389cadec 100644
--- a/llvm/test/Analysis/CostModel/RISCV/fround.ll
+++ b/llvm/test/Analysis/CostModel/RISCV/fround.ll
@@ -606,6 +606,200 @@ define void @llrint_fp16() {
   ret void
 }
 
+define void @lround() {
+; CHECK-LABEL: 'lround'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = call i32 @llvm.lround.i32.bf16(bfloat poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %2 = call <2 x i32> @llvm.lround.v2i32.v2bf16(<2 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %3 = call <4 x i32> @llvm.lround.v4i32.v4bf16(<4 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %4 = call <8 x i32> @llvm.lround.v8i32.v8bf16(<8 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %5 = call <16 x i32> @llvm.lround.v16i32.v16bf16(<16 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %6 = call <vscale x 1 x i32> @llvm.lround.nxv1i32.nxv1bf16(<vscale x 1 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %7 = call <vscale x 2 x i32> @llvm.lround.nxv2i32.nxv2bf16(<vscale x 2 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %8 = call <vscale x 4 x i32> @llvm.lround.nxv4i32.nxv4bf16(<vscale x 4 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %9 = call <vscale x 8 x i32> @llvm.lround.nxv8i32.nxv8bf16(<vscale x 8 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %10 = call <vscale x 16 x i32> @llvm.lround.nxv16i32.nxv16bf16(<vscale x 16 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %11 = call i32 @llvm.lround.i32.f32(float poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %12 = call <2 x i32> @llvm.lround.v2i32.v2f32(<2 x float> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %13 = call <4 x i32> @llvm.lround.v4i32.v4f32(<4 x float> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %14 = call <8 x i32> @llvm.lround.v8i32.v8f32(<8 x float> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %15 = call <16 x i32> @llvm.lround.v16i32.v16f32(<16 x float> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %16 = call <vscale x 1 x i32> @llvm.lround.nxv1i32.nxv1f32(<vscale x 1 x float> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %17 = call <vscale x 2 x i32> @llvm.lround.nxv2i32.nxv2f32(<vscale x 2 x float> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %18 = call <vscale x 4 x i32> @llvm.lround.nxv4i32.nxv4f32(<vscale x 4 x float> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %19 = call <vscale x 8 x i32> @llvm.lround.nxv8i32.nxv8f32(<vscale x 8 x float> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %20 = call <vscale x 16 x i32> @llvm.lround.nxv16i32.nxv16f32(<vscale x 16 x float> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %21 = call i32 @llvm.lround.i32.f64(double poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %22 = call <2 x i32> @llvm.lround.v2i32.v2f64(<2 x double> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %23 = call <4 x i32> @llvm.lround.v4i32.v4f64(<4 x double> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %24 = call <8 x i32> @llvm.lround.v8i32.v8f64(<8 x double> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %25 = call <16 x i32> @llvm.lround.v16i32.v16f64(<16 x double> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %26 = call <vscale x 1 x i32> @llvm.lround.nxv1i32.nxv1f64(<vscale x 1 x double> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %27 = call <vscale x 2 x i32> @llvm.lround.nxv2i32.nxv2f64(<vscale x 2 x double> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %28 = call <vscale x 4 x i32> @llvm.lround.nxv4i32.nxv4f64(<vscale x 4 x double> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %29 = call <vscale x 8 x i32> @llvm.lround.nxv8i32.nxv8f64(<vscale x 8 x double> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %30 = call <vscale x 16 x i32> @llvm.lround.nxv16i32.nxv16f64(<vscale x 16 x double> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %31 = call i64 @llvm.lround.i64.bf16(bfloat poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %32 = call <2 x i64> @llvm.lround.v2i64.v2bf16(<2 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %33 = call <4 x i64> @llvm.lround.v4i64.v4bf16(<4 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %34 = call <8 x i64> @llvm.lround.v8i64.v8bf16(<8 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %35 = call <16 x i64> @llvm.lround.v16i64.v16bf16(<16 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %36 = call <vscale x 1 x i64> @llvm.lround.nxv1i64.nxv1bf16(<vscale x 1 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %37 = call <vscale x 2 x i64> @llvm.lround.nxv2i64.nxv2bf16(<vscale x 2 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %38 = call <vscale x 4 x i64> @llvm.lround.nxv4i64.nxv4bf16(<vscale x 4 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %39 = call <vscale x 8 x i64> @llvm.lround.nxv8i64.nxv8bf16(<vscale x 8 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %40 = call <vscale x 16 x i64> @llvm.lround.nxv16i64.nxv16bf16(<vscale x 16 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %41 = call i64 @llvm.lround.i64.f32(float poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %42 = call <2 x i64> @llvm.lround.v2i64.v2f32(<2 x float> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %43 = call <4 x i64> @llvm.lround.v4i64.v4f32(<4 x float> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %44 = call <8 x i64> @llvm.lround.v8i64.v8f32(<8 x float> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %45 = call <16 x i64> @llvm.lround.v16i64.v16f32(<16 x float> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %46 = call <vscale x 1 x i64> @llvm.lround.nxv1i64.nxv1f32(<vscale x 1 x float> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %47 = call <vscale x 2 x i64> @llvm.lround.nxv2i64.nxv2f32(<vscale x 2 x float> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %48 = call <vscale x 4 x i64> @llvm.lround.nxv4i64.nxv4f32(<vscale x 4 x float> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %49 = call <vscale x 8 x i64> @llvm.lround.nxv8i64.nxv8f32(<vscale x 8 x float> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %50 = call <vscale x 16 x i64> @llvm.lround.nxv16i64.nxv16f32(<vscale x 16 x float> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %51 = call i64 @llvm.lround.i64.f64(double poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %52 = call <2 x i64> @llvm.lround.v2i64.v2f64(<2 x double> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %53 = call <4 x i64> @llvm.lround.v4i64.v4f64(<4 x double> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %54 = call <8 x i64> @llvm.lround.v8i64.v8f64(<8 x double> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %55 = call <16 x i64> @llvm.lround.v16i64.v16f64(<16 x double> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %56 = call <vscale x 1 x i64> @llvm.lround.nxv1i64.nxv1f64(<vscale x 1 x double> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %57 = call <vscale x 2 x i64> @llvm.lround.nxv2i64.nxv2f64(<vscale x 2 x double> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %58 = call <vscale x 4 x i64> @llvm.lround.nxv4i64.nxv4f64(<vscale x 4 x double> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %59 = call <vscale x 8 x i64> @llvm.lround.nxv8i64.nxv8f64(<vscale x 8 x double> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %60 = call <vscale x 16 x i64> @llvm.lround.nxv16i64.nxv16f64(<vscale x 16 x double> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
+  call i32 @llvm.lround.i32.bf16(bfloat poison)
+  call <2 x i32> @llvm.lround.v2i32.v2bf16(<2 x bfloat> poison)
+  call <4 x i32> @llvm.lround.v4i32.v4bf16(<4 x bfloat> poison)
+  call <8 x i32> @llvm.lround.v8i32.v8bf16(<8 x bfloat> poison)
+  call <16 x i32> @llvm.lround.v16i32.v16bf16(<16 x bfloat> poison)
+  call <vscale x 1 x i32> @llvm.lround.nxv1i32.nxv1bf16(<vscale x 1 x bfloat> poison)
+  call <vscale x 2 x i32> @llvm.lround.nxv2i32.nxv2bf16(<vscale x 2 x bfloat> poison)
+  call <vscale x 4 x i32> @llvm.lround.nxv4i32.nxv4bf16(<vscale x 4 x bfloat> poison)
+  call <vscale x 8 x i32> @llvm.lround.nxv8i32.nxv8bf16(<vscale x 8 x bfloat> poison)
+  call <vscale x 16 x i32> @llvm.lround.nxv16i32.nxv16bf16(<vscale x 16 x bfloat> poison)
+  call i32 @llvm.lround.i32.f32(float poison)
+  call <2 x i32> @llvm.lround.v2i32.v2f32(<2 x float> poison)
+  call <4 x i32> @llvm.lround.v4i32.v4f32(<4 x float> poison)
+  call <8 x i32> @llvm.lround.v8i32.v8f32(<8 x float> poison)
+  call <16 x i32> @llvm.lround.v16i32.v16f32(<16 x float> poison)
+  call <vscale x 1 x i32> @llvm.lround.nxv1i32.nxv1f32(<vscale x 1 x float> poison)
+  call <vscale x 2 x i32> @llvm.lround.nxv2i32.nxv2f32(<vscale x 2 x float> poison)
+  call <vscale x 4 x i32> @llvm.lround.nxv4i32.nxv4f32(<vscale x 4 x float> poison)
+  call <vscale x 8 x i32> @llvm.lround.nxv8i32.nxv8f32(<vscale x 8 x float> poison)
+  call <vscale x 16 x i32> @llvm.lround.nxv16i32.nxv16f32(<vscale x 16 x float> poison)
+  call i32 @llvm.lround.i32.f64(double poison)
+  call <2 x i32> @llvm.lround.v2i32.v2f64(<2 x double> poison)
+  call <4 x i32> @llvm.lround.v4i32.v4f64(<4 x double> poison)
+  call <8 x i32> @llvm.lround.v8i32.v8f64(<8 x double> poison)
+  call <16 x i32> @llvm.lround.v16i32.v16f64(<16 x double> poison)
+  call <vscale x 1 x i32> @llvm.lround.nxv1i32.nxv1f64(<vscale x 1 x double> poison)
+  call <vscale x 2 x i32> @llvm.lround.nxv2i32.nxv2f64(<vscale x 2 x double> poison)
+  call <vscale x 4 x i32> @llvm.lround.nxv4i32.nxv4f64(<vscale x 4 x double> poison)
+  call <vscale x 8 x i32> @llvm.lround.nxv8i32.nxv8f64(<vscale x 8 x double> poison)
+  call <vscale x 16 x i32> @llvm.lround.nxv16i32.nxv16f64(<vscale x 16 x double> poison)
+  call i64 @llvm.lround.i64.bf16(bfloat poison)
+  call <2 x i64> @llvm.lround.v2i64.v2bf16(<2 x bfloat> poison)
+  call <4 x i64> @llvm.lround.v4i64.v4bf16(<4 x bfloat> poison)
+  call <8 x i64> @llvm.lround.v8i64.v8bf16(<8 x bfloat> poison)
+  call <16 x i64> @llvm.lround.v16i64.v16bf16(<16 x bfloat> poison)
+  call <vscale x 1 x i64> @llvm.lround.nxv1i64.nxv1bf16(<vscale x 1 x bfloat> poison)
+  call <vscale x 2 x i64> @llvm.lround.nxv2i64.nxv2bf16(<vscale x 2 x bfloat> poison)
+  call <vscale x 4 x i64> @llvm.lround.nxv4i64.nxv4bf16(<vscale x 4 x bfloat> poison)
+  call <vscale x 8 x i64> @llvm.lround.nxv8i64.nxv8bf16(<vscale x 8 x bfloat> poison)
+  call <vscale x 16 x i64> @llvm.lround.nxv16i64.nxv16bf16(<vscale x 16 x bfloat> poison)
+  call i64 @llvm.lround.i64.f32(float poison)
+  call <2 x i64> @llvm.lround.v2i64.v2f32(<2 x float> poison)
+  call <4 x i64> @llvm.lround.v4i64.v4f32(<4 x float> poison)
+  call <8 x i64> @llvm.lround.v8i64.v8f32(<8 x float> poison)
+  call <16 x i64> @llvm.lround.v16i64.v16f32(<16 x float> poison)
+  call <vscale x 1 x i64> @llvm.lround.nxv1i64.nxv1f32(<vscale x 1 x float> poison)
+  call <vscale x 2 x i64> @llvm.lround.nxv2i64.nxv2f32(<vscale x 2 x float> poison)
+  call <vscale x 4 x i64> @llvm.lround.nxv4i64.nxv4f32(<vscale x 4 x float> poison)
+  call <vscale x 8 x i64> @llvm.lround.nxv8i64.nxv8f32(<vscale x 8 x float> poison)
+  call <vscale x 16 x i64> @llvm.lround.nxv16i64.nxv16f32(<vscale x 16 x float> poison)
+  call i64 @llvm.lround.i64.f64(double poison)
+  call <2 x i64> @llvm.lround.v2i64.v2f64(<2 x double> poison)
+  call <4 x i64> @llvm.lround.v4i64.v4f64(<4 x double> poison)
+  call <8 x i64> @llvm.lround.v8i64.v8f64(<8 x double> poison)
+  call <16 x i64> @llvm.lround.v16i64.v16f64(<16 x double> poison)
+  call <vscale x 1 x i64> @llvm.lround.nxv1i64.nxv1f64(<vscale x 1 x double> poison)
+  call <vscale x 2 x i64> @llvm.lround.nxv2i64.nxv2f64(<vscale x 2 x double> poison)
+  call <vscale x 4 x i64> @llvm.lround.nxv4i64.nxv4f64(<vscale x 4 x double> poison)
+  call <vscale x 8 x i64> @llvm.lround.nxv8i64.nxv8f64(<vscale x 8 x double> poison)
+  call <vscale x 16 x i64> @llvm.lround.nxv16i64.nxv16f64(<vscale x 16 x double> poison)
+  ret void
+}
+
+define void @llround() {
+; CHECK-LABEL: 'llround'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = call i64 @llvm.llround.i64.bf16(bfloat poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %2 = call <2 x i64> @llvm.llround.v2i64.v2bf16(<2 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %3 = call <4 x i64> @llvm.llround.v4i64.v4bf16(<4 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %4 = call <8 x i64> @llvm.llround.v8i64.v8bf16(<8 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %5 = call <16 x i64> @llvm.llround.v16i64.v16bf16(<16 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %6 = call <vscale x 1 x i64> @llvm.llround.nxv1i64.nxv1bf16(<vscale x 1 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %7 = call <vscale x 2 x i64> @llvm.llround.nxv2i64.nxv2bf16(<vscale x 2 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %8 = call <vscale x 4 x i64> @llvm.llround.nxv4i64.nxv4bf16(<vscale x 4 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %9 = call <vscale x 8 x i64> @llvm.llround.nxv8i64.nxv8bf16(<vscale x 8 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %10 = call <vscale x 16 x i64> @llvm.llround.nxv16i64.nxv16bf16(<vscale x 16 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %11 = call i64 @llvm.llround.i64.f32(float poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %12 = call <2 x i64> @llvm.llround.v2i64.v2f32(<2 x float> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %13 = call <4 x i64> @llvm.llround.v4i64.v4f32(<4 x float> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %14 = call <8 x i64> @llvm.llround.v8i64.v8f32(<8 x float> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %15 = call <16 x i64> @llvm.llround.v16i64.v16f32(<16 x float> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %16 = call <vscale x 1 x i64> @llvm.llround.nxv1i64.nxv1f32(<vscale x 1 x float> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %17 = call <vscale x 2 x i64> @llvm.llround.nxv2i64.nxv2f32(<vscale x 2 x float> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %18 = call <vscale x 4 x i64> @llvm.llround.nxv4i64.nxv4f32(<vscale x 4 x float> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %19 = call <vscale x 8 x i64> @llvm.llround.nxv8i64.nxv8f32(<vscale x 8 x float> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %20 = call <vscale x 16 x i64> @llvm.llround.nxv16i64.nxv16f32(<vscale x 16 x float> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %21 = call i64 @llvm.llround.i64.f64(double poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %22 = call <2 x i64> @llvm.llround.v2i64.v2f64(<2 x double> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %23 = call <4 x i64> @llvm.llround.v4i64.v4f64(<4 x double> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %24 = call <8 x i64> @llvm.llround.v8i64.v8f64(<8 x double> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %25 = call <16 x i64> @llvm.llround.v16i64.v16f64(<16 x double> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %26 = call <vscale x 1 x i64> @llvm.llround.nxv1i64.nxv1f64(<vscale x 1 x double> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %27 = call <vscale x 2 x i64> @llvm.llround.nxv2i64.nxv2f64(<vscale x 2 x double> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %28 = call <vscale x 4 x i64> @llvm.llround.nxv4i64.nxv4f64(<vscale x 4 x double> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %29 = call <vscale x 8 x i64> @llvm.llround.nxv8i64.nxv8f64(<vscale x 8 x double> poison)
+; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %30 = call <vscale x 16 x i64> @llvm.llround.nxv16i64.nxv16f64(<vscale x 16 x double> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
+  call i64 @llvm.llround.i64.bf16(bfloat poison)
+  call <2 x i64> @llvm.llround.v2i64.v2bf16(<2 x bfloat> poison)
+  call <4 x i64> @llvm.llround.v4i64.v4bf16(<4 x bfloat> poison)
+  call <8 x i64> @llvm.llround.v8i64.v8bf16(<8 x bfloat> poison)
+  call <16 x i64> @llvm.llround.v16i64.v16bf16(<16 x bfloat> poison)
+  call <vscale x 1 x i64> @llvm.llround.nxv1i64.nxv1bf16(<vscale x 1 x bfloat> poison)
+  call <vscale x 2 x i64> @llvm.llround.nxv2i64.nxv2bf16(<vscale x 2 x bfloat> poison)
+  call <vscale x 4 x i64> @llvm.llround.nxv4i64.nxv4bf16(<vscale x 4 x bfloat> poison)
+  call <vscale x 8 x i64> @llvm.llround.nxv8i64.nxv8bf16(<vscale x 8 x bfloat> poison)
+  call <vscale x 16 x i64> @llvm.llround.nxv16i64.nxv16bf16(<vscale x 16 x bfloat> poison)
+  call i64 @llvm.llround.i64.f32(float poison)
+  call <2 x i64> @llvm.llround.v2i64.v2f32(<2 x float> poison)
+  call <4 x i64> @llvm.llround.v4i64.v4f32(<4 x float> poison)
+  call <8 x i64> @llvm.llround.v8i64.v8f32(<8 x float> poison)
+  call <16 x i64> @llvm.llround.v16i64.v16f32(<16 x float> poison)
+  call <vscale x 1 x i64> @llvm.llround.nxv1i64.nxv1f32(<vscale x 1 x float> poison)
+  call <vscale x 2 x i64> @llvm.llround.nxv2i64.nxv2f32(<vscale x 2 x float> poison)
+  call <vscale x 4 x i64> @llvm.llround.nxv4i64.nxv4f32(<vscale x 4 x float> poison)
+  call <vscale x 8 x i64> @llvm.llround.nxv8i64.nxv8f32(<vscale x 8 x float> poison)
+  call <vscale x 16 x i64> @llvm.llround.nxv16i64.nxv16f32(<vscale x 16 x float> poison)
+  call i64 @llvm.llround.i64.f64(double poison)
+  call <2 x i64> @llvm.llround.v2i64.v2f64(<2 x double> poison)
+  call <4 x i64> @llvm.llround.v4i64.v4f64(<4 x double> poison)
+  call <8 x i64> @llvm.llround.v8i64.v8f64(<8 x double> poison)
+  call <16 x i64> @llvm.llround.v16i64.v16f64(<16 x double> poison)
+  call <vscale x 1 x i64> @llvm.llround.nxv1i64.nxv1f64(<vscale x 1 x double> poison)
+  call <vscale x 2 x i64> @llvm.llround.nxv2i64.nxv2f64(<vscale x 2 x double> poison)
+  call <vscale x 4 x i64> @llvm.llround.nxv4i64.nxv4f64(<vscale x 4 x double> poison)
+  call <vscale x 8 x i64> @llvm.llround.nxv8i64.nxv8f64(<vscale x 8 x double> poison)
+  call <vscale x 16 x i64> @llvm.llround.nxv16i64.nxv16f64(<vscale x 16 x double> poison)
+  ret void
+}
+
 define void @nearbyint() {
 ; CHECK-LABEL: 'nearbyint'
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %1 = call bfloat @llvm.nearbyint.bf16(bfloat undef)

>From 6fc7e25288450835f8766471a51de3d4ae4fdd8a Mon Sep 17 00:00:00 2001
From: Ramkumar Ramachandra <ramkumar.ramachandra at codasip.com>
Date: Wed, 25 Jun 2025 14:59:59 +0100
Subject: [PATCH 2/2] [CostModel/RISCV] Fix costs of [l]lrint, [l]lround

---
 llvm/include/llvm/CodeGen/BasicTTIImpl.h      |    6 +
 .../Target/RISCV/RISCVTargetTransformInfo.cpp |   31 +-
 llvm/test/Analysis/CostModel/RISCV/fround.ll  |  124 +-
 .../RISCV/rvv/fixed-vectors-llround.ll        | 1030 +++++++++++++++++
 .../CodeGen/RISCV/rvv/fixed-vectors-lround.ll |  926 +++++++++++++++
 5 files changed, 2047 insertions(+), 70 deletions(-)
 create mode 100644 llvm/test/CodeGen/RISCV/rvv/fixed-vectors-llround.ll
 create mode 100644 llvm/test/CodeGen/RISCV/rvv/fixed-vectors-lround.ll

diff --git a/llvm/include/llvm/CodeGen/BasicTTIImpl.h b/llvm/include/llvm/CodeGen/BasicTTIImpl.h
index cabecbec175b3..a19f93903e15a 100644
--- a/llvm/include/llvm/CodeGen/BasicTTIImpl.h
+++ b/llvm/include/llvm/CodeGen/BasicTTIImpl.h
@@ -2290,6 +2290,12 @@ class BasicTTIImplBase : public TargetTransformInfoImplCRTPBase<T> {
     case Intrinsic::llrint:
       ISD = ISD::LLRINT;
       break;
+    case Intrinsic::lround:
+      ISD = ISD::LROUND;
+      break;
+    case Intrinsic::llround:
+      ISD = ISD::LLROUND;
+      break;
     case Intrinsic::round:
       ISD = ISD::FROUND;
       break;
diff --git a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
index 67a51c12b508e..9fb5896f3df40 100644
--- a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
+++ b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
@@ -1191,9 +1191,6 @@ static const CostTblEntry VectorIntrinsicCostTable[]{
     {Intrinsic::roundeven, MVT::f64, 9},
     {Intrinsic::rint, MVT::f32, 7},
     {Intrinsic::rint, MVT::f64, 7},
-    {Intrinsic::lrint, MVT::i32, 1},
-    {Intrinsic::lrint, MVT::i64, 1},
-    {Intrinsic::llrint, MVT::i64, 1},
     {Intrinsic::nearbyint, MVT::f32, 9},
     {Intrinsic::nearbyint, MVT::f64, 9},
     {Intrinsic::bswap, MVT::i16, 3},
@@ -1262,11 +1259,29 @@ RISCVTTIImpl::getIntrinsicInstrCost(const IntrinsicCostAttributes &ICA,
   switch (ICA.getID()) {
   case Intrinsic::lrint:
   case Intrinsic::llrint:
-    // We can't currently lower half or bfloat vector lrint/llrint.
-    if (auto *VecTy = dyn_cast<VectorType>(ICA.getArgTypes()[0]);
-        VecTy && VecTy->getElementType()->is16bitFPTy())
-      return InstructionCost::getInvalid();
-    [[fallthrough]];
+  case Intrinsic::lround:
+  case Intrinsic::llround: {
+    auto LT = getTypeLegalizationCost(RetTy);
+    switch (ICA.getID()) {
+    case Intrinsic::lrint:
+    case Intrinsic::llrint:
+      // We can't currently lower half or bfloat vector lrint/llrint.
+      if (auto *VecTy = dyn_cast<VectorType>(ICA.getArgTypes()[0]);
+          VecTy && VecTy->getElementType()->is16bitFPTy())
+        return InstructionCost::getInvalid();
+      break;
+    case Intrinsic::lround:
+    case Intrinsic::llround:
+      // We can't currently lower scalable-vector lround/llround.
+      if (LT.second.isScalableVector())
+        return InstructionCost::getInvalid();
+      break;
+    }
+    if (ST->hasVInstructions() && LT.second.isVector())
+      return LT.first *
+             getRISCVInstructionCost(RISCV::VFCVT_X_F_V, LT.second, CostKind);
+    break;
+  }
   case Intrinsic::ceil:
   case Intrinsic::floor:
   case Intrinsic::trunc:
diff --git a/llvm/test/Analysis/CostModel/RISCV/fround.ll b/llvm/test/Analysis/CostModel/RISCV/fround.ll
index 94c09389cadec..2edceb6425ead 100644
--- a/llvm/test/Analysis/CostModel/RISCV/fround.ll
+++ b/llvm/test/Analysis/CostModel/RISCV/fround.ll
@@ -436,23 +436,23 @@ define void @lrint() {
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %10 = call <vscale x 16 x i64> @llvm.lrint.nxv16i64.nxv16bf16(<vscale x 16 x bfloat> undef)
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %11 = call i64 @llvm.lrint.i64.f32(float undef)
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %12 = call <2 x i64> @llvm.lrint.v2i64.v2f32(<2 x float> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %13 = call <4 x i64> @llvm.lrint.v4i64.v4f32(<4 x float> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %14 = call <8 x i64> @llvm.lrint.v8i64.v8f32(<8 x float> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %15 = call <16 x i64> @llvm.lrint.v16i64.v16f32(<16 x float> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %13 = call <4 x i64> @llvm.lrint.v4i64.v4f32(<4 x float> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %14 = call <8 x i64> @llvm.lrint.v8i64.v8f32(<8 x float> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %15 = call <16 x i64> @llvm.lrint.v16i64.v16f32(<16 x float> undef)
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %16 = call <vscale x 1 x i64> @llvm.lrint.nxv1i64.nxv1f32(<vscale x 1 x float> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %17 = call <vscale x 2 x i64> @llvm.lrint.nxv2i64.nxv2f32(<vscale x 2 x float> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %18 = call <vscale x 4 x i64> @llvm.lrint.nxv4i64.nxv4f32(<vscale x 4 x float> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %19 = call <vscale x 8 x i64> @llvm.lrint.nxv8i64.nxv8f32(<vscale x 8 x float> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %20 = call <vscale x 16 x i64> @llvm.lrint.nxv16i64.nxv16f32(<vscale x 16 x float> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %17 = call <vscale x 2 x i64> @llvm.lrint.nxv2i64.nxv2f32(<vscale x 2 x float> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %18 = call <vscale x 4 x i64> @llvm.lrint.nxv4i64.nxv4f32(<vscale x 4 x float> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %19 = call <vscale x 8 x i64> @llvm.lrint.nxv8i64.nxv8f32(<vscale x 8 x float> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %20 = call <vscale x 16 x i64> @llvm.lrint.nxv16i64.nxv16f32(<vscale x 16 x float> undef)
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %21 = call i64 @llvm.lrint.i64.f64(double undef)
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %22 = call <2 x i64> @llvm.lrint.v2i64.v2f64(<2 x double> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %23 = call <4 x i64> @llvm.lrint.v4i64.v4f64(<4 x double> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %24 = call <8 x i64> @llvm.lrint.v8i64.v8f64(<8 x double> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %25 = call <16 x i64> @llvm.lrint.v16i64.v16f64(<16 x double> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %23 = call <4 x i64> @llvm.lrint.v4i64.v4f64(<4 x double> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %24 = call <8 x i64> @llvm.lrint.v8i64.v8f64(<8 x double> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %25 = call <16 x i64> @llvm.lrint.v16i64.v16f64(<16 x double> undef)
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %26 = call <vscale x 1 x i64> @llvm.lrint.nxv1i64.nxv1f64(<vscale x 1 x double> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %27 = call <vscale x 2 x i64> @llvm.lrint.nxv2i64.nxv2f64(<vscale x 2 x double> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %28 = call <vscale x 4 x i64> @llvm.lrint.nxv4i64.nxv4f64(<vscale x 4 x double> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %29 = call <vscale x 8 x i64> @llvm.lrint.nxv8i64.nxv8f64(<vscale x 8 x double> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %27 = call <vscale x 2 x i64> @llvm.lrint.nxv2i64.nxv2f64(<vscale x 2 x double> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %28 = call <vscale x 4 x i64> @llvm.lrint.nxv4i64.nxv4f64(<vscale x 4 x double> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %29 = call <vscale x 8 x i64> @llvm.lrint.nxv8i64.nxv8f64(<vscale x 8 x double> undef)
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
   call i64 @llvm.lrint.i64.bf16(bfloat undef)
@@ -528,23 +528,23 @@ define void @llrint() {
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %10 = call <vscale x 16 x i64> @llvm.llrint.nxv16i64.nxv16bf16(<vscale x 16 x bfloat> undef)
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %11 = call i64 @llvm.llrint.i64.f32(float undef)
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %12 = call <2 x i64> @llvm.llrint.v2i64.v2f32(<2 x float> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %13 = call <4 x i64> @llvm.llrint.v4i64.v4f32(<4 x float> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %14 = call <8 x i64> @llvm.llrint.v8i64.v8f32(<8 x float> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %15 = call <16 x i64> @llvm.llrint.v16i64.v16f32(<16 x float> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %13 = call <4 x i64> @llvm.llrint.v4i64.v4f32(<4 x float> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %14 = call <8 x i64> @llvm.llrint.v8i64.v8f32(<8 x float> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %15 = call <16 x i64> @llvm.llrint.v16i64.v16f32(<16 x float> undef)
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %16 = call <vscale x 1 x i64> @llvm.llrint.nxv1i64.nxv1f32(<vscale x 1 x float> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %17 = call <vscale x 2 x i64> @llvm.llrint.nxv2i64.nxv2f32(<vscale x 2 x float> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %18 = call <vscale x 4 x i64> @llvm.llrint.nxv4i64.nxv4f32(<vscale x 4 x float> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %19 = call <vscale x 8 x i64> @llvm.llrint.nxv8i64.nxv8f32(<vscale x 8 x float> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %20 = call <vscale x 16 x i64> @llvm.llrint.nxv16i64.nxv16f32(<vscale x 16 x float> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %17 = call <vscale x 2 x i64> @llvm.llrint.nxv2i64.nxv2f32(<vscale x 2 x float> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %18 = call <vscale x 4 x i64> @llvm.llrint.nxv4i64.nxv4f32(<vscale x 4 x float> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %19 = call <vscale x 8 x i64> @llvm.llrint.nxv8i64.nxv8f32(<vscale x 8 x float> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %20 = call <vscale x 16 x i64> @llvm.llrint.nxv16i64.nxv16f32(<vscale x 16 x float> undef)
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %21 = call i64 @llvm.llrint.i64.f64(double undef)
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %22 = call <2 x i64> @llvm.llrint.v2i64.v2f64(<2 x double> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %23 = call <4 x i64> @llvm.llrint.v4i64.v4f64(<4 x double> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %24 = call <8 x i64> @llvm.llrint.v8i64.v8f64(<8 x double> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %25 = call <16 x i64> @llvm.llrint.v16i64.v16f64(<16 x double> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %23 = call <4 x i64> @llvm.llrint.v4i64.v4f64(<4 x double> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %24 = call <8 x i64> @llvm.llrint.v8i64.v8f64(<8 x double> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %25 = call <16 x i64> @llvm.llrint.v16i64.v16f64(<16 x double> undef)
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %26 = call <vscale x 1 x i64> @llvm.llrint.nxv1i64.nxv1f64(<vscale x 1 x double> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %27 = call <vscale x 2 x i64> @llvm.llrint.nxv2i64.nxv2f64(<vscale x 2 x double> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %28 = call <vscale x 4 x i64> @llvm.llrint.nxv4i64.nxv4f64(<vscale x 4 x double> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %29 = call <vscale x 8 x i64> @llvm.llrint.nxv8i64.nxv8f64(<vscale x 8 x double> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %27 = call <vscale x 2 x i64> @llvm.llrint.nxv2i64.nxv2f64(<vscale x 2 x double> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %28 = call <vscale x 4 x i64> @llvm.llrint.nxv4i64.nxv4f64(<vscale x 4 x double> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %29 = call <vscale x 8 x i64> @llvm.llrint.nxv8i64.nxv8f64(<vscale x 8 x double> undef)
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
   call i64 @llvm.llrint.i64.bf16(bfloat undef)
@@ -609,60 +609,60 @@ define void @llrint_fp16() {
 define void @lround() {
 ; CHECK-LABEL: 'lround'
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = call i32 @llvm.lround.i32.bf16(bfloat poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %2 = call <2 x i32> @llvm.lround.v2i32.v2bf16(<2 x bfloat> poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %3 = call <4 x i32> @llvm.lround.v4i32.v4bf16(<4 x bfloat> poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %4 = call <8 x i32> @llvm.lround.v8i32.v8bf16(<8 x bfloat> poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %5 = call <16 x i32> @llvm.lround.v16i32.v16bf16(<16 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %2 = call <2 x i32> @llvm.lround.v2i32.v2bf16(<2 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %3 = call <4 x i32> @llvm.lround.v4i32.v4bf16(<4 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %4 = call <8 x i32> @llvm.lround.v8i32.v8bf16(<8 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %5 = call <16 x i32> @llvm.lround.v16i32.v16bf16(<16 x bfloat> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %6 = call <vscale x 1 x i32> @llvm.lround.nxv1i32.nxv1bf16(<vscale x 1 x bfloat> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %7 = call <vscale x 2 x i32> @llvm.lround.nxv2i32.nxv2bf16(<vscale x 2 x bfloat> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %8 = call <vscale x 4 x i32> @llvm.lround.nxv4i32.nxv4bf16(<vscale x 4 x bfloat> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %9 = call <vscale x 8 x i32> @llvm.lround.nxv8i32.nxv8bf16(<vscale x 8 x bfloat> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %10 = call <vscale x 16 x i32> @llvm.lround.nxv16i32.nxv16bf16(<vscale x 16 x bfloat> poison)
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %11 = call i32 @llvm.lround.i32.f32(float poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %12 = call <2 x i32> @llvm.lround.v2i32.v2f32(<2 x float> poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %13 = call <4 x i32> @llvm.lround.v4i32.v4f32(<4 x float> poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %14 = call <8 x i32> @llvm.lround.v8i32.v8f32(<8 x float> poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %15 = call <16 x i32> @llvm.lround.v16i32.v16f32(<16 x float> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %12 = call <2 x i32> @llvm.lround.v2i32.v2f32(<2 x float> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %13 = call <4 x i32> @llvm.lround.v4i32.v4f32(<4 x float> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %14 = call <8 x i32> @llvm.lround.v8i32.v8f32(<8 x float> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %15 = call <16 x i32> @llvm.lround.v16i32.v16f32(<16 x float> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %16 = call <vscale x 1 x i32> @llvm.lround.nxv1i32.nxv1f32(<vscale x 1 x float> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %17 = call <vscale x 2 x i32> @llvm.lround.nxv2i32.nxv2f32(<vscale x 2 x float> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %18 = call <vscale x 4 x i32> @llvm.lround.nxv4i32.nxv4f32(<vscale x 4 x float> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %19 = call <vscale x 8 x i32> @llvm.lround.nxv8i32.nxv8f32(<vscale x 8 x float> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %20 = call <vscale x 16 x i32> @llvm.lround.nxv16i32.nxv16f32(<vscale x 16 x float> poison)
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %21 = call i32 @llvm.lround.i32.f64(double poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %22 = call <2 x i32> @llvm.lround.v2i32.v2f64(<2 x double> poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %23 = call <4 x i32> @llvm.lround.v4i32.v4f64(<4 x double> poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %24 = call <8 x i32> @llvm.lround.v8i32.v8f64(<8 x double> poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %25 = call <16 x i32> @llvm.lround.v16i32.v16f64(<16 x double> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %22 = call <2 x i32> @llvm.lround.v2i32.v2f64(<2 x double> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %23 = call <4 x i32> @llvm.lround.v4i32.v4f64(<4 x double> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %24 = call <8 x i32> @llvm.lround.v8i32.v8f64(<8 x double> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %25 = call <16 x i32> @llvm.lround.v16i32.v16f64(<16 x double> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %26 = call <vscale x 1 x i32> @llvm.lround.nxv1i32.nxv1f64(<vscale x 1 x double> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %27 = call <vscale x 2 x i32> @llvm.lround.nxv2i32.nxv2f64(<vscale x 2 x double> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %28 = call <vscale x 4 x i32> @llvm.lround.nxv4i32.nxv4f64(<vscale x 4 x double> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %29 = call <vscale x 8 x i32> @llvm.lround.nxv8i32.nxv8f64(<vscale x 8 x double> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %30 = call <vscale x 16 x i32> @llvm.lround.nxv16i32.nxv16f64(<vscale x 16 x double> poison)
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %31 = call i64 @llvm.lround.i64.bf16(bfloat poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %32 = call <2 x i64> @llvm.lround.v2i64.v2bf16(<2 x bfloat> poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %33 = call <4 x i64> @llvm.lround.v4i64.v4bf16(<4 x bfloat> poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %34 = call <8 x i64> @llvm.lround.v8i64.v8bf16(<8 x bfloat> poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %35 = call <16 x i64> @llvm.lround.v16i64.v16bf16(<16 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %32 = call <2 x i64> @llvm.lround.v2i64.v2bf16(<2 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %33 = call <4 x i64> @llvm.lround.v4i64.v4bf16(<4 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %34 = call <8 x i64> @llvm.lround.v8i64.v8bf16(<8 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %35 = call <16 x i64> @llvm.lround.v16i64.v16bf16(<16 x bfloat> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %36 = call <vscale x 1 x i64> @llvm.lround.nxv1i64.nxv1bf16(<vscale x 1 x bfloat> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %37 = call <vscale x 2 x i64> @llvm.lround.nxv2i64.nxv2bf16(<vscale x 2 x bfloat> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %38 = call <vscale x 4 x i64> @llvm.lround.nxv4i64.nxv4bf16(<vscale x 4 x bfloat> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %39 = call <vscale x 8 x i64> @llvm.lround.nxv8i64.nxv8bf16(<vscale x 8 x bfloat> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %40 = call <vscale x 16 x i64> @llvm.lround.nxv16i64.nxv16bf16(<vscale x 16 x bfloat> poison)
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %41 = call i64 @llvm.lround.i64.f32(float poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %42 = call <2 x i64> @llvm.lround.v2i64.v2f32(<2 x float> poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %43 = call <4 x i64> @llvm.lround.v4i64.v4f32(<4 x float> poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %44 = call <8 x i64> @llvm.lround.v8i64.v8f32(<8 x float> poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %45 = call <16 x i64> @llvm.lround.v16i64.v16f32(<16 x float> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %42 = call <2 x i64> @llvm.lround.v2i64.v2f32(<2 x float> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %43 = call <4 x i64> @llvm.lround.v4i64.v4f32(<4 x float> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %44 = call <8 x i64> @llvm.lround.v8i64.v8f32(<8 x float> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %45 = call <16 x i64> @llvm.lround.v16i64.v16f32(<16 x float> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %46 = call <vscale x 1 x i64> @llvm.lround.nxv1i64.nxv1f32(<vscale x 1 x float> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %47 = call <vscale x 2 x i64> @llvm.lround.nxv2i64.nxv2f32(<vscale x 2 x float> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %48 = call <vscale x 4 x i64> @llvm.lround.nxv4i64.nxv4f32(<vscale x 4 x float> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %49 = call <vscale x 8 x i64> @llvm.lround.nxv8i64.nxv8f32(<vscale x 8 x float> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %50 = call <vscale x 16 x i64> @llvm.lround.nxv16i64.nxv16f32(<vscale x 16 x float> poison)
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %51 = call i64 @llvm.lround.i64.f64(double poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %52 = call <2 x i64> @llvm.lround.v2i64.v2f64(<2 x double> poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %53 = call <4 x i64> @llvm.lround.v4i64.v4f64(<4 x double> poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %54 = call <8 x i64> @llvm.lround.v8i64.v8f64(<8 x double> poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %55 = call <16 x i64> @llvm.lround.v16i64.v16f64(<16 x double> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %52 = call <2 x i64> @llvm.lround.v2i64.v2f64(<2 x double> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %53 = call <4 x i64> @llvm.lround.v4i64.v4f64(<4 x double> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %54 = call <8 x i64> @llvm.lround.v8i64.v8f64(<8 x double> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %55 = call <16 x i64> @llvm.lround.v16i64.v16f64(<16 x double> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %56 = call <vscale x 1 x i64> @llvm.lround.nxv1i64.nxv1f64(<vscale x 1 x double> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %57 = call <vscale x 2 x i64> @llvm.lround.nxv2i64.nxv2f64(<vscale x 2 x double> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %58 = call <vscale x 4 x i64> @llvm.lround.nxv4i64.nxv4f64(<vscale x 4 x double> poison)
@@ -736,30 +736,30 @@ define void @lround() {
 define void @llround() {
 ; CHECK-LABEL: 'llround'
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %1 = call i64 @llvm.llround.i64.bf16(bfloat poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %2 = call <2 x i64> @llvm.llround.v2i64.v2bf16(<2 x bfloat> poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %3 = call <4 x i64> @llvm.llround.v4i64.v4bf16(<4 x bfloat> poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %4 = call <8 x i64> @llvm.llround.v8i64.v8bf16(<8 x bfloat> poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %5 = call <16 x i64> @llvm.llround.v16i64.v16bf16(<16 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %2 = call <2 x i64> @llvm.llround.v2i64.v2bf16(<2 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %3 = call <4 x i64> @llvm.llround.v4i64.v4bf16(<4 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %4 = call <8 x i64> @llvm.llround.v8i64.v8bf16(<8 x bfloat> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %5 = call <16 x i64> @llvm.llround.v16i64.v16bf16(<16 x bfloat> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %6 = call <vscale x 1 x i64> @llvm.llround.nxv1i64.nxv1bf16(<vscale x 1 x bfloat> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %7 = call <vscale x 2 x i64> @llvm.llround.nxv2i64.nxv2bf16(<vscale x 2 x bfloat> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %8 = call <vscale x 4 x i64> @llvm.llround.nxv4i64.nxv4bf16(<vscale x 4 x bfloat> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %9 = call <vscale x 8 x i64> @llvm.llround.nxv8i64.nxv8bf16(<vscale x 8 x bfloat> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %10 = call <vscale x 16 x i64> @llvm.llround.nxv16i64.nxv16bf16(<vscale x 16 x bfloat> poison)
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %11 = call i64 @llvm.llround.i64.f32(float poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %12 = call <2 x i64> @llvm.llround.v2i64.v2f32(<2 x float> poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %13 = call <4 x i64> @llvm.llround.v4i64.v4f32(<4 x float> poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %14 = call <8 x i64> @llvm.llround.v8i64.v8f32(<8 x float> poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %15 = call <16 x i64> @llvm.llround.v16i64.v16f32(<16 x float> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %12 = call <2 x i64> @llvm.llround.v2i64.v2f32(<2 x float> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %13 = call <4 x i64> @llvm.llround.v4i64.v4f32(<4 x float> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %14 = call <8 x i64> @llvm.llround.v8i64.v8f32(<8 x float> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %15 = call <16 x i64> @llvm.llround.v16i64.v16f32(<16 x float> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %16 = call <vscale x 1 x i64> @llvm.llround.nxv1i64.nxv1f32(<vscale x 1 x float> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %17 = call <vscale x 2 x i64> @llvm.llround.nxv2i64.nxv2f32(<vscale x 2 x float> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %18 = call <vscale x 4 x i64> @llvm.llround.nxv4i64.nxv4f32(<vscale x 4 x float> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %19 = call <vscale x 8 x i64> @llvm.llround.nxv8i64.nxv8f32(<vscale x 8 x float> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %20 = call <vscale x 16 x i64> @llvm.llround.nxv16i64.nxv16f32(<vscale x 16 x float> poison)
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %21 = call i64 @llvm.llround.i64.f64(double poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %22 = call <2 x i64> @llvm.llround.v2i64.v2f64(<2 x double> poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %23 = call <4 x i64> @llvm.llround.v4i64.v4f64(<4 x double> poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %24 = call <8 x i64> @llvm.llround.v8i64.v8f64(<8 x double> poison)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %25 = call <16 x i64> @llvm.llround.v16i64.v16f64(<16 x double> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %22 = call <2 x i64> @llvm.llround.v2i64.v2f64(<2 x double> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %23 = call <4 x i64> @llvm.llround.v4i64.v4f64(<4 x double> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %24 = call <8 x i64> @llvm.llround.v8i64.v8f64(<8 x double> poison)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %25 = call <16 x i64> @llvm.llround.v16i64.v16f64(<16 x double> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %26 = call <vscale x 1 x i64> @llvm.llround.nxv1i64.nxv1f64(<vscale x 1 x double> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %27 = call <vscale x 2 x i64> @llvm.llround.nxv2i64.nxv2f64(<vscale x 2 x double> poison)
 ; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %28 = call <vscale x 4 x i64> @llvm.llround.nxv4i64.nxv4f64(<vscale x 4 x double> poison)
diff --git a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-llround.ll b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-llround.ll
new file mode 100644
index 0000000000000..d3e818c22c3ea
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-llround.ll
@@ -0,0 +1,1030 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+v,+f,+d -target-abi=ilp32d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s --check-prefix=RV32
+; RUN: llc -mtriple=riscv64 -mattr=+v,+f,+d -target-abi=lp64d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s --check-prefix=RV64
+
+define <1 x i64> @llround_v1i64_v1f32(<1 x float> %x) {
+; RV32-LABEL: llround_v1i64_v1f32:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
+; RV32-NEXT:    .cfi_offset ra, -4
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    sw a0, 0(sp)
+; RV32-NEXT:    sw a1, 4(sp)
+; RV32-NEXT:    mv a0, sp
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV32-NEXT:    vlse64.v v8, (a0), zero
+; RV32-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
+; RV32-NEXT:    .cfi_restore ra
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    .cfi_def_cfa_offset 0
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: llround_v1i64_v1f32:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV64-NEXT:    vfmv.f.s fa5, v8
+; RV64-NEXT:    fcvt.l.s a0, fa5, rmm
+; RV64-NEXT:    vsetvli zero, zero, e64, m2, ta, ma
+; RV64-NEXT:    vmv.s.x v8, a0
+; RV64-NEXT:    ret
+  %a = call <1 x i64> @llvm.llround.v1i64.v1f32(<1 x float> %x)
+  ret <1 x i64> %a
+}
+declare <1 x i64> @llvm.llround.v1i64.v1f32(<1 x float>)
+
+define <2 x i64> @llround_v2i64_v2f32(<2 x float> %x) {
+; RV32-LABEL: llround_v2i64_v2f32:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -32
+; RV32-NEXT:    .cfi_def_cfa_offset 32
+; RV32-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
+; RV32-NEXT:    .cfi_offset ra, -4
+; RV32-NEXT:    csrr a0, vlenb
+; RV32-NEXT:    slli a0, a0, 1
+; RV32-NEXT:    sub sp, sp, a0
+; RV32-NEXT:    .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x20, 0x22, 0x11, 0x02, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 32 + 2 * vlenb
+; RV32-NEXT:    addi a0, sp, 16
+; RV32-NEXT:    vs1r.v v8, (a0) # vscale x 8-byte Folded Spill
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
+; RV32-NEXT:    vmv.v.x v8, a0
+; RV32-NEXT:    vslide1down.vx v8, v8, a1
+; RV32-NEXT:    csrr a0, vlenb
+; RV32-NEXT:    add a0, sp, a0
+; RV32-NEXT:    addi a0, a0, 16
+; RV32-NEXT:    vs1r.v v8, (a0) # vscale x 8-byte Folded Spill
+; RV32-NEXT:    addi a0, sp, 16
+; RV32-NEXT:    vl1r.v v8, (a0) # vscale x 8-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 1, e32, mf2, ta, ma
+; RV32-NEXT:    vslidedown.vi v8, v8, 1
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    csrr a2, vlenb
+; RV32-NEXT:    add a2, sp, a2
+; RV32-NEXT:    addi a2, a2, 16
+; RV32-NEXT:    vl1r.v v8, (a2) # vscale x 8-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
+; RV32-NEXT:    vslide1down.vx v8, v8, a0
+; RV32-NEXT:    vslide1down.vx v8, v8, a1
+; RV32-NEXT:    csrr a0, vlenb
+; RV32-NEXT:    slli a0, a0, 1
+; RV32-NEXT:    add sp, sp, a0
+; RV32-NEXT:    .cfi_def_cfa sp, 32
+; RV32-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
+; RV32-NEXT:    .cfi_restore ra
+; RV32-NEXT:    addi sp, sp, 32
+; RV32-NEXT:    .cfi_def_cfa_offset 0
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: llround_v2i64_v2f32:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetivli zero, 1, e32, mf2, ta, ma
+; RV64-NEXT:    vslidedown.vi v9, v8, 1
+; RV64-NEXT:    vfmv.f.s fa5, v8
+; RV64-NEXT:    fcvt.l.s a0, fa5, rmm
+; RV64-NEXT:    vfmv.f.s fa5, v9
+; RV64-NEXT:    fcvt.l.s a1, fa5, rmm
+; RV64-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
+; RV64-NEXT:    vmv.v.x v8, a0
+; RV64-NEXT:    vslide1down.vx v8, v8, a1
+; RV64-NEXT:    ret
+  %a = call <2 x i64> @llvm.llround.v2i64.v2f32(<2 x float> %x)
+  ret <2 x i64> %a
+}
+declare <2 x i64> @llvm.llround.v2i64.v2f32(<2 x float>)
+
+define <3 x i64> @llround_v3i64_v3f32(<3 x float> %x) {
+; RV32-LABEL: llround_v3i64_v3f32:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -32
+; RV32-NEXT:    .cfi_def_cfa_offset 32
+; RV32-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
+; RV32-NEXT:    .cfi_offset ra, -4
+; RV32-NEXT:    csrr a0, vlenb
+; RV32-NEXT:    slli a1, a0, 1
+; RV32-NEXT:    add a0, a1, a0
+; RV32-NEXT:    sub sp, sp, a0
+; RV32-NEXT:    .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x20, 0x22, 0x11, 0x03, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 32 + 3 * vlenb
+; RV32-NEXT:    csrr a0, vlenb
+; RV32-NEXT:    slli a0, a0, 1
+; RV32-NEXT:    add a0, sp, a0
+; RV32-NEXT:    addi a0, a0, 16
+; RV32-NEXT:    vs1r.v v8, (a0) # vscale x 8-byte Folded Spill
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    vsetivli zero, 8, e32, m2, ta, ma
+; RV32-NEXT:    vmv.v.x v8, a0
+; RV32-NEXT:    vslide1down.vx v8, v8, a1
+; RV32-NEXT:    addi a0, sp, 16
+; RV32-NEXT:    vs2r.v v8, (a0) # vscale x 16-byte Folded Spill
+; RV32-NEXT:    csrr a0, vlenb
+; RV32-NEXT:    slli a0, a0, 1
+; RV32-NEXT:    add a0, sp, a0
+; RV32-NEXT:    addi a0, a0, 16
+; RV32-NEXT:    vl1r.v v8, (a0) # vscale x 8-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV32-NEXT:    vslidedown.vi v8, v8, 1
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    addi a2, sp, 16
+; RV32-NEXT:    vl2r.v v8, (a2) # vscale x 16-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 8, e32, m2, ta, ma
+; RV32-NEXT:    vslide1down.vx v8, v8, a0
+; RV32-NEXT:    vslide1down.vx v8, v8, a1
+; RV32-NEXT:    addi a0, sp, 16
+; RV32-NEXT:    vs2r.v v8, (a0) # vscale x 16-byte Folded Spill
+; RV32-NEXT:    csrr a0, vlenb
+; RV32-NEXT:    slli a0, a0, 1
+; RV32-NEXT:    add a0, sp, a0
+; RV32-NEXT:    addi a0, a0, 16
+; RV32-NEXT:    vl1r.v v8, (a0) # vscale x 8-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV32-NEXT:    vslidedown.vi v8, v8, 2
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    addi a2, sp, 16
+; RV32-NEXT:    vl2r.v v8, (a2) # vscale x 16-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 8, e32, m2, ta, ma
+; RV32-NEXT:    vslide1down.vx v8, v8, a0
+; RV32-NEXT:    vslide1down.vx v8, v8, a1
+; RV32-NEXT:    addi a0, sp, 16
+; RV32-NEXT:    vs2r.v v8, (a0) # vscale x 16-byte Folded Spill
+; RV32-NEXT:    csrr a0, vlenb
+; RV32-NEXT:    slli a0, a0, 1
+; RV32-NEXT:    add a0, sp, a0
+; RV32-NEXT:    addi a0, a0, 16
+; RV32-NEXT:    vl1r.v v8, (a0) # vscale x 8-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV32-NEXT:    vslidedown.vi v8, v8, 3
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    addi a2, sp, 16
+; RV32-NEXT:    vl2r.v v8, (a2) # vscale x 16-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 8, e32, m2, ta, ma
+; RV32-NEXT:    vslide1down.vx v8, v8, a0
+; RV32-NEXT:    vslide1down.vx v8, v8, a1
+; RV32-NEXT:    csrr a0, vlenb
+; RV32-NEXT:    slli a1, a0, 1
+; RV32-NEXT:    add a0, a1, a0
+; RV32-NEXT:    add sp, sp, a0
+; RV32-NEXT:    .cfi_def_cfa sp, 32
+; RV32-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
+; RV32-NEXT:    .cfi_restore ra
+; RV32-NEXT:    addi sp, sp, 32
+; RV32-NEXT:    .cfi_def_cfa_offset 0
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: llround_v3i64_v3f32:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV64-NEXT:    vslidedown.vi v9, v8, 1
+; RV64-NEXT:    vfmv.f.s fa5, v8
+; RV64-NEXT:    vslidedown.vi v10, v8, 2
+; RV64-NEXT:    vslidedown.vi v11, v8, 3
+; RV64-NEXT:    fcvt.l.s a0, fa5, rmm
+; RV64-NEXT:    vfmv.f.s fa5, v9
+; RV64-NEXT:    fcvt.l.s a1, fa5, rmm
+; RV64-NEXT:    vfmv.f.s fa5, v10
+; RV64-NEXT:    vsetivli zero, 4, e64, m2, ta, ma
+; RV64-NEXT:    vmv.v.x v8, a0
+; RV64-NEXT:    fcvt.l.s a0, fa5, rmm
+; RV64-NEXT:    vsetvli zero, zero, e32, m1, ta, ma
+; RV64-NEXT:    vfmv.f.s fa5, v11
+; RV64-NEXT:    vsetvli zero, zero, e64, m2, ta, ma
+; RV64-NEXT:    vslide1down.vx v8, v8, a1
+; RV64-NEXT:    vslide1down.vx v8, v8, a0
+; RV64-NEXT:    fcvt.l.s a0, fa5, rmm
+; RV64-NEXT:    vslide1down.vx v8, v8, a0
+; RV64-NEXT:    ret
+  %a = call <3 x i64> @llvm.llround.v3i64.v3f32(<3 x float> %x)
+  ret <3 x i64> %a
+}
+declare <3 x i64> @llvm.llround.v3i64.v3f32(<3 x float>)
+
+define <4 x i64> @llround_v4i64_v4f32(<4 x float> %x) {
+; RV32-LABEL: llround_v4i64_v4f32:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -32
+; RV32-NEXT:    .cfi_def_cfa_offset 32
+; RV32-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
+; RV32-NEXT:    .cfi_offset ra, -4
+; RV32-NEXT:    csrr a0, vlenb
+; RV32-NEXT:    slli a1, a0, 1
+; RV32-NEXT:    add a0, a1, a0
+; RV32-NEXT:    sub sp, sp, a0
+; RV32-NEXT:    .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x20, 0x22, 0x11, 0x03, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 32 + 3 * vlenb
+; RV32-NEXT:    csrr a0, vlenb
+; RV32-NEXT:    slli a0, a0, 1
+; RV32-NEXT:    add a0, sp, a0
+; RV32-NEXT:    addi a0, a0, 16
+; RV32-NEXT:    vs1r.v v8, (a0) # vscale x 8-byte Folded Spill
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    vsetivli zero, 8, e32, m2, ta, ma
+; RV32-NEXT:    vmv.v.x v8, a0
+; RV32-NEXT:    vslide1down.vx v8, v8, a1
+; RV32-NEXT:    addi a0, sp, 16
+; RV32-NEXT:    vs2r.v v8, (a0) # vscale x 16-byte Folded Spill
+; RV32-NEXT:    csrr a0, vlenb
+; RV32-NEXT:    slli a0, a0, 1
+; RV32-NEXT:    add a0, sp, a0
+; RV32-NEXT:    addi a0, a0, 16
+; RV32-NEXT:    vl1r.v v8, (a0) # vscale x 8-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV32-NEXT:    vslidedown.vi v8, v8, 1
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    addi a2, sp, 16
+; RV32-NEXT:    vl2r.v v8, (a2) # vscale x 16-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 8, e32, m2, ta, ma
+; RV32-NEXT:    vslide1down.vx v8, v8, a0
+; RV32-NEXT:    vslide1down.vx v8, v8, a1
+; RV32-NEXT:    addi a0, sp, 16
+; RV32-NEXT:    vs2r.v v8, (a0) # vscale x 16-byte Folded Spill
+; RV32-NEXT:    csrr a0, vlenb
+; RV32-NEXT:    slli a0, a0, 1
+; RV32-NEXT:    add a0, sp, a0
+; RV32-NEXT:    addi a0, a0, 16
+; RV32-NEXT:    vl1r.v v8, (a0) # vscale x 8-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV32-NEXT:    vslidedown.vi v8, v8, 2
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    addi a2, sp, 16
+; RV32-NEXT:    vl2r.v v8, (a2) # vscale x 16-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 8, e32, m2, ta, ma
+; RV32-NEXT:    vslide1down.vx v8, v8, a0
+; RV32-NEXT:    vslide1down.vx v8, v8, a1
+; RV32-NEXT:    addi a0, sp, 16
+; RV32-NEXT:    vs2r.v v8, (a0) # vscale x 16-byte Folded Spill
+; RV32-NEXT:    csrr a0, vlenb
+; RV32-NEXT:    slli a0, a0, 1
+; RV32-NEXT:    add a0, sp, a0
+; RV32-NEXT:    addi a0, a0, 16
+; RV32-NEXT:    vl1r.v v8, (a0) # vscale x 8-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV32-NEXT:    vslidedown.vi v8, v8, 3
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    addi a2, sp, 16
+; RV32-NEXT:    vl2r.v v8, (a2) # vscale x 16-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 8, e32, m2, ta, ma
+; RV32-NEXT:    vslide1down.vx v8, v8, a0
+; RV32-NEXT:    vslide1down.vx v8, v8, a1
+; RV32-NEXT:    csrr a0, vlenb
+; RV32-NEXT:    slli a1, a0, 1
+; RV32-NEXT:    add a0, a1, a0
+; RV32-NEXT:    add sp, sp, a0
+; RV32-NEXT:    .cfi_def_cfa sp, 32
+; RV32-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
+; RV32-NEXT:    .cfi_restore ra
+; RV32-NEXT:    addi sp, sp, 32
+; RV32-NEXT:    .cfi_def_cfa_offset 0
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: llround_v4i64_v4f32:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV64-NEXT:    vslidedown.vi v9, v8, 1
+; RV64-NEXT:    vfmv.f.s fa5, v8
+; RV64-NEXT:    vslidedown.vi v10, v8, 2
+; RV64-NEXT:    vslidedown.vi v11, v8, 3
+; RV64-NEXT:    fcvt.l.s a0, fa5, rmm
+; RV64-NEXT:    vfmv.f.s fa5, v9
+; RV64-NEXT:    fcvt.l.s a1, fa5, rmm
+; RV64-NEXT:    vfmv.f.s fa5, v10
+; RV64-NEXT:    vsetivli zero, 4, e64, m2, ta, ma
+; RV64-NEXT:    vmv.v.x v8, a0
+; RV64-NEXT:    fcvt.l.s a0, fa5, rmm
+; RV64-NEXT:    vsetvli zero, zero, e32, m1, ta, ma
+; RV64-NEXT:    vfmv.f.s fa5, v11
+; RV64-NEXT:    vsetvli zero, zero, e64, m2, ta, ma
+; RV64-NEXT:    vslide1down.vx v8, v8, a1
+; RV64-NEXT:    vslide1down.vx v8, v8, a0
+; RV64-NEXT:    fcvt.l.s a0, fa5, rmm
+; RV64-NEXT:    vslide1down.vx v8, v8, a0
+; RV64-NEXT:    ret
+  %a = call <4 x i64> @llvm.llround.v4i64.v4f32(<4 x float> %x)
+  ret <4 x i64> %a
+}
+declare <4 x i64> @llvm.llround.v4i64.v4f32(<4 x float>)
+
+define <8 x i64> @llround_v8i64_v8f32(<8 x float> %x) {
+; RV32-LABEL: llround_v8i64_v8f32:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -208
+; RV32-NEXT:    .cfi_def_cfa_offset 208
+; RV32-NEXT:    sw ra, 204(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw s0, 200(sp) # 4-byte Folded Spill
+; RV32-NEXT:    .cfi_offset ra, -4
+; RV32-NEXT:    .cfi_offset s0, -8
+; RV32-NEXT:    addi s0, sp, 208
+; RV32-NEXT:    .cfi_def_cfa s0, 0
+; RV32-NEXT:    csrr a0, vlenb
+; RV32-NEXT:    slli a0, a0, 1
+; RV32-NEXT:    sub sp, sp, a0
+; RV32-NEXT:    andi sp, sp, -64
+; RV32-NEXT:    addi a0, sp, 192
+; RV32-NEXT:    vs2r.v v8, (a0) # vscale x 16-byte Folded Spill
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    sw a0, 64(sp)
+; RV32-NEXT:    sw a1, 68(sp)
+; RV32-NEXT:    addi a0, sp, 192
+; RV32-NEXT:    vl2r.v v8, (a0) # vscale x 16-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
+; RV32-NEXT:    vslidedown.vi v8, v8, 7
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    sw a0, 120(sp)
+; RV32-NEXT:    sw a1, 124(sp)
+; RV32-NEXT:    addi a0, sp, 192
+; RV32-NEXT:    vl2r.v v8, (a0) # vscale x 16-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
+; RV32-NEXT:    vslidedown.vi v8, v8, 6
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    sw a0, 112(sp)
+; RV32-NEXT:    sw a1, 116(sp)
+; RV32-NEXT:    addi a0, sp, 192
+; RV32-NEXT:    vl2r.v v8, (a0) # vscale x 16-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
+; RV32-NEXT:    vslidedown.vi v8, v8, 5
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    sw a0, 104(sp)
+; RV32-NEXT:    sw a1, 108(sp)
+; RV32-NEXT:    addi a0, sp, 192
+; RV32-NEXT:    vl2r.v v8, (a0) # vscale x 16-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
+; RV32-NEXT:    vslidedown.vi v8, v8, 4
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    sw a0, 96(sp)
+; RV32-NEXT:    sw a1, 100(sp)
+; RV32-NEXT:    addi a0, sp, 192
+; RV32-NEXT:    vl2r.v v8, (a0) # vscale x 16-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV32-NEXT:    vslidedown.vi v8, v8, 3
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    sw a0, 88(sp)
+; RV32-NEXT:    sw a1, 92(sp)
+; RV32-NEXT:    addi a0, sp, 192
+; RV32-NEXT:    vl2r.v v8, (a0) # vscale x 16-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV32-NEXT:    vslidedown.vi v8, v8, 2
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    sw a0, 80(sp)
+; RV32-NEXT:    sw a1, 84(sp)
+; RV32-NEXT:    addi a0, sp, 192
+; RV32-NEXT:    vl2r.v v8, (a0) # vscale x 16-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV32-NEXT:    vslidedown.vi v8, v8, 1
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    sw a0, 72(sp)
+; RV32-NEXT:    sw a1, 76(sp)
+; RV32-NEXT:    addi a0, sp, 64
+; RV32-NEXT:    vsetivli zero, 16, e32, m4, ta, ma
+; RV32-NEXT:    vle32.v v8, (a0)
+; RV32-NEXT:    addi sp, s0, -208
+; RV32-NEXT:    .cfi_def_cfa sp, 208
+; RV32-NEXT:    lw ra, 204(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw s0, 200(sp) # 4-byte Folded Reload
+; RV32-NEXT:    .cfi_restore ra
+; RV32-NEXT:    .cfi_restore s0
+; RV32-NEXT:    addi sp, sp, 208
+; RV32-NEXT:    .cfi_def_cfa_offset 0
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: llround_v8i64_v8f32:
+; RV64:       # %bb.0:
+; RV64-NEXT:    addi sp, sp, -128
+; RV64-NEXT:    .cfi_def_cfa_offset 128
+; RV64-NEXT:    sd ra, 120(sp) # 8-byte Folded Spill
+; RV64-NEXT:    sd s0, 112(sp) # 8-byte Folded Spill
+; RV64-NEXT:    .cfi_offset ra, -8
+; RV64-NEXT:    .cfi_offset s0, -16
+; RV64-NEXT:    addi s0, sp, 128
+; RV64-NEXT:    .cfi_def_cfa s0, 0
+; RV64-NEXT:    andi sp, sp, -64
+; RV64-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
+; RV64-NEXT:    vfmv.f.s fa5, v8
+; RV64-NEXT:    vslidedown.vi v10, v8, 7
+; RV64-NEXT:    fcvt.l.s a0, fa5, rmm
+; RV64-NEXT:    vfmv.f.s fa5, v10
+; RV64-NEXT:    vslidedown.vi v10, v8, 6
+; RV64-NEXT:    fcvt.l.s a1, fa5, rmm
+; RV64-NEXT:    vfmv.f.s fa5, v10
+; RV64-NEXT:    vslidedown.vi v10, v8, 5
+; RV64-NEXT:    fcvt.l.s a2, fa5, rmm
+; RV64-NEXT:    vfmv.f.s fa5, v10
+; RV64-NEXT:    vslidedown.vi v10, v8, 4
+; RV64-NEXT:    fcvt.l.s a3, fa5, rmm
+; RV64-NEXT:    vfmv.f.s fa5, v10
+; RV64-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV64-NEXT:    vslidedown.vi v9, v8, 3
+; RV64-NEXT:    vslidedown.vi v10, v8, 2
+; RV64-NEXT:    vslidedown.vi v8, v8, 1
+; RV64-NEXT:    fcvt.l.s a4, fa5, rmm
+; RV64-NEXT:    vfmv.f.s fa5, v9
+; RV64-NEXT:    fcvt.l.s a5, fa5, rmm
+; RV64-NEXT:    vfmv.f.s fa5, v10
+; RV64-NEXT:    fcvt.l.s a6, fa5, rmm
+; RV64-NEXT:    vfmv.f.s fa5, v8
+; RV64-NEXT:    sd a4, 32(sp)
+; RV64-NEXT:    sd a3, 40(sp)
+; RV64-NEXT:    sd a2, 48(sp)
+; RV64-NEXT:    sd a1, 56(sp)
+; RV64-NEXT:    fcvt.l.s a1, fa5, rmm
+; RV64-NEXT:    sd a0, 0(sp)
+; RV64-NEXT:    sd a1, 8(sp)
+; RV64-NEXT:    sd a6, 16(sp)
+; RV64-NEXT:    sd a5, 24(sp)
+; RV64-NEXT:    mv a0, sp
+; RV64-NEXT:    vsetivli zero, 8, e64, m4, ta, ma
+; RV64-NEXT:    vle64.v v8, (a0)
+; RV64-NEXT:    addi sp, s0, -128
+; RV64-NEXT:    .cfi_def_cfa sp, 128
+; RV64-NEXT:    ld ra, 120(sp) # 8-byte Folded Reload
+; RV64-NEXT:    ld s0, 112(sp) # 8-byte Folded Reload
+; RV64-NEXT:    .cfi_restore ra
+; RV64-NEXT:    .cfi_restore s0
+; RV64-NEXT:    addi sp, sp, 128
+; RV64-NEXT:    .cfi_def_cfa_offset 0
+; RV64-NEXT:    ret
+  %a = call <8 x i64> @llvm.llround.v8i64.v8f32(<8 x float> %x)
+  ret <8 x i64> %a
+}
+declare <8 x i64> @llvm.llround.v8i64.v8f32(<8 x float>)
+
+define <16 x i64> @llround_v16i64_v16f32(<16 x float> %x) {
+; RV32-LABEL: llround_v16i64_v16f32:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -400
+; RV32-NEXT:    .cfi_def_cfa_offset 400
+; RV32-NEXT:    sw ra, 396(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw s0, 392(sp) # 4-byte Folded Spill
+; RV32-NEXT:    .cfi_offset ra, -4
+; RV32-NEXT:    .cfi_offset s0, -8
+; RV32-NEXT:    addi s0, sp, 400
+; RV32-NEXT:    .cfi_def_cfa s0, 0
+; RV32-NEXT:    csrr a0, vlenb
+; RV32-NEXT:    slli a0, a0, 2
+; RV32-NEXT:    sub sp, sp, a0
+; RV32-NEXT:    andi sp, sp, -128
+; RV32-NEXT:    addi a0, sp, 384
+; RV32-NEXT:    vs4r.v v8, (a0) # vscale x 32-byte Folded Spill
+; RV32-NEXT:    addi a0, sp, 64
+; RV32-NEXT:    vsetivli zero, 16, e32, m4, ta, ma
+; RV32-NEXT:    vse32.v v8, (a0)
+; RV32-NEXT:    flw fa0, 124(sp)
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    sw a0, 248(sp)
+; RV32-NEXT:    sw a1, 252(sp)
+; RV32-NEXT:    flw fa0, 120(sp)
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    sw a0, 240(sp)
+; RV32-NEXT:    sw a1, 244(sp)
+; RV32-NEXT:    flw fa0, 116(sp)
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    sw a0, 232(sp)
+; RV32-NEXT:    sw a1, 236(sp)
+; RV32-NEXT:    flw fa0, 112(sp)
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    sw a0, 224(sp)
+; RV32-NEXT:    sw a1, 228(sp)
+; RV32-NEXT:    flw fa0, 108(sp)
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    sw a0, 216(sp)
+; RV32-NEXT:    sw a1, 220(sp)
+; RV32-NEXT:    flw fa0, 104(sp)
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    sw a0, 208(sp)
+; RV32-NEXT:    sw a1, 212(sp)
+; RV32-NEXT:    flw fa0, 100(sp)
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    sw a0, 200(sp)
+; RV32-NEXT:    sw a1, 204(sp)
+; RV32-NEXT:    flw fa0, 96(sp)
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    sw a0, 192(sp)
+; RV32-NEXT:    sw a1, 196(sp)
+; RV32-NEXT:    addi a0, sp, 384
+; RV32-NEXT:    vl4r.v v8, (a0) # vscale x 32-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    sw a0, 128(sp)
+; RV32-NEXT:    sw a1, 132(sp)
+; RV32-NEXT:    addi a0, sp, 384
+; RV32-NEXT:    vl4r.v v8, (a0) # vscale x 32-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV32-NEXT:    vslidedown.vi v8, v8, 3
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    sw a0, 152(sp)
+; RV32-NEXT:    sw a1, 156(sp)
+; RV32-NEXT:    addi a0, sp, 384
+; RV32-NEXT:    vl4r.v v8, (a0) # vscale x 32-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV32-NEXT:    vslidedown.vi v8, v8, 2
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    sw a0, 144(sp)
+; RV32-NEXT:    sw a1, 148(sp)
+; RV32-NEXT:    addi a0, sp, 384
+; RV32-NEXT:    vl4r.v v8, (a0) # vscale x 32-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV32-NEXT:    vslidedown.vi v8, v8, 1
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    sw a0, 136(sp)
+; RV32-NEXT:    sw a1, 140(sp)
+; RV32-NEXT:    addi a0, sp, 384
+; RV32-NEXT:    vl4r.v v8, (a0) # vscale x 32-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
+; RV32-NEXT:    vslidedown.vi v8, v8, 7
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    sw a0, 184(sp)
+; RV32-NEXT:    sw a1, 188(sp)
+; RV32-NEXT:    addi a0, sp, 384
+; RV32-NEXT:    vl4r.v v8, (a0) # vscale x 32-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
+; RV32-NEXT:    vslidedown.vi v8, v8, 6
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    sw a0, 176(sp)
+; RV32-NEXT:    sw a1, 180(sp)
+; RV32-NEXT:    addi a0, sp, 384
+; RV32-NEXT:    vl4r.v v8, (a0) # vscale x 32-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
+; RV32-NEXT:    vslidedown.vi v8, v8, 5
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    sw a0, 168(sp)
+; RV32-NEXT:    sw a1, 172(sp)
+; RV32-NEXT:    addi a0, sp, 384
+; RV32-NEXT:    vl4r.v v8, (a0) # vscale x 32-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
+; RV32-NEXT:    vslidedown.vi v8, v8, 4
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llroundf
+; RV32-NEXT:    sw a0, 160(sp)
+; RV32-NEXT:    sw a1, 164(sp)
+; RV32-NEXT:    li a0, 32
+; RV32-NEXT:    addi a1, sp, 128
+; RV32-NEXT:    vsetvli zero, a0, e32, m8, ta, ma
+; RV32-NEXT:    vle32.v v8, (a1)
+; RV32-NEXT:    addi sp, s0, -400
+; RV32-NEXT:    .cfi_def_cfa sp, 400
+; RV32-NEXT:    lw ra, 396(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw s0, 392(sp) # 4-byte Folded Reload
+; RV32-NEXT:    .cfi_restore ra
+; RV32-NEXT:    .cfi_restore s0
+; RV32-NEXT:    addi sp, sp, 400
+; RV32-NEXT:    .cfi_def_cfa_offset 0
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: llround_v16i64_v16f32:
+; RV64:       # %bb.0:
+; RV64-NEXT:    addi sp, sp, -384
+; RV64-NEXT:    .cfi_def_cfa_offset 384
+; RV64-NEXT:    sd ra, 376(sp) # 8-byte Folded Spill
+; RV64-NEXT:    sd s0, 368(sp) # 8-byte Folded Spill
+; RV64-NEXT:    .cfi_offset ra, -8
+; RV64-NEXT:    .cfi_offset s0, -16
+; RV64-NEXT:    addi s0, sp, 384
+; RV64-NEXT:    .cfi_def_cfa s0, 0
+; RV64-NEXT:    andi sp, sp, -128
+; RV64-NEXT:    addi a0, sp, 64
+; RV64-NEXT:    vsetivli zero, 16, e32, m4, ta, ma
+; RV64-NEXT:    vse32.v v8, (a0)
+; RV64-NEXT:    flw fa5, 124(sp)
+; RV64-NEXT:    vfmv.f.s fa4, v8
+; RV64-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV64-NEXT:    vslidedown.vi v10, v8, 3
+; RV64-NEXT:    vslidedown.vi v12, v8, 2
+; RV64-NEXT:    fcvt.l.s a0, fa5, rmm
+; RV64-NEXT:    sd a0, 248(sp)
+; RV64-NEXT:    flw fa5, 120(sp)
+; RV64-NEXT:    vslidedown.vi v13, v8, 1
+; RV64-NEXT:    fcvt.l.s a0, fa4, rmm
+; RV64-NEXT:    vfmv.f.s fa4, v10
+; RV64-NEXT:    fcvt.l.s a1, fa5, rmm
+; RV64-NEXT:    sd a1, 240(sp)
+; RV64-NEXT:    flw fa5, 116(sp)
+; RV64-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
+; RV64-NEXT:    vslidedown.vi v10, v8, 7
+; RV64-NEXT:    fcvt.l.s a1, fa4, rmm
+; RV64-NEXT:    vfmv.f.s fa4, v12
+; RV64-NEXT:    fcvt.l.s a2, fa5, rmm
+; RV64-NEXT:    sd a2, 232(sp)
+; RV64-NEXT:    flw fa5, 112(sp)
+; RV64-NEXT:    fcvt.l.s a2, fa4, rmm
+; RV64-NEXT:    vfmv.f.s fa4, v13
+; RV64-NEXT:    vslidedown.vi v12, v8, 6
+; RV64-NEXT:    fcvt.l.s a3, fa5, rmm
+; RV64-NEXT:    sd a3, 224(sp)
+; RV64-NEXT:    flw fa5, 108(sp)
+; RV64-NEXT:    fcvt.l.s a3, fa4, rmm
+; RV64-NEXT:    vfmv.f.s fa4, v10
+; RV64-NEXT:    vslidedown.vi v10, v8, 5
+; RV64-NEXT:    fcvt.l.s a4, fa5, rmm
+; RV64-NEXT:    sd a4, 216(sp)
+; RV64-NEXT:    flw fa5, 104(sp)
+; RV64-NEXT:    fcvt.l.s a4, fa4, rmm
+; RV64-NEXT:    vfmv.f.s fa4, v12
+; RV64-NEXT:    fcvt.l.s a5, fa4, rmm
+; RV64-NEXT:    fcvt.l.s a6, fa5, rmm
+; RV64-NEXT:    sd a6, 208(sp)
+; RV64-NEXT:    flw fa5, 100(sp)
+; RV64-NEXT:    vfmv.f.s fa4, v10
+; RV64-NEXT:    fcvt.l.s a6, fa4, rmm
+; RV64-NEXT:    vslidedown.vi v8, v8, 4
+; RV64-NEXT:    fcvt.l.s a7, fa5, rmm
+; RV64-NEXT:    vfmv.f.s fa5, v8
+; RV64-NEXT:    sd a7, 200(sp)
+; RV64-NEXT:    fcvt.l.s a7, fa5, rmm
+; RV64-NEXT:    flw fa5, 96(sp)
+; RV64-NEXT:    sd a0, 128(sp)
+; RV64-NEXT:    sd a3, 136(sp)
+; RV64-NEXT:    sd a2, 144(sp)
+; RV64-NEXT:    sd a1, 152(sp)
+; RV64-NEXT:    sd a7, 160(sp)
+; RV64-NEXT:    sd a6, 168(sp)
+; RV64-NEXT:    sd a5, 176(sp)
+; RV64-NEXT:    sd a4, 184(sp)
+; RV64-NEXT:    fcvt.l.s a0, fa5, rmm
+; RV64-NEXT:    sd a0, 192(sp)
+; RV64-NEXT:    addi a0, sp, 128
+; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, ma
+; RV64-NEXT:    vle64.v v8, (a0)
+; RV64-NEXT:    addi sp, s0, -384
+; RV64-NEXT:    .cfi_def_cfa sp, 384
+; RV64-NEXT:    ld ra, 376(sp) # 8-byte Folded Reload
+; RV64-NEXT:    ld s0, 368(sp) # 8-byte Folded Reload
+; RV64-NEXT:    .cfi_restore ra
+; RV64-NEXT:    .cfi_restore s0
+; RV64-NEXT:    addi sp, sp, 384
+; RV64-NEXT:    .cfi_def_cfa_offset 0
+; RV64-NEXT:    ret
+  %a = call <16 x i64> @llvm.llround.v16i64.v16f32(<16 x float> %x)
+  ret <16 x i64> %a
+}
+declare <16 x i64> @llvm.llround.v16i64.v16f32(<16 x float>)
+
+define <1 x i64> @llround_v1i64_v1f64(<1 x double> %x) {
+; RV32-LABEL: llround_v1i64_v1f64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
+; RV32-NEXT:    .cfi_offset ra, -4
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llround
+; RV32-NEXT:    sw a0, 0(sp)
+; RV32-NEXT:    sw a1, 4(sp)
+; RV32-NEXT:    mv a0, sp
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV32-NEXT:    vlse64.v v8, (a0), zero
+; RV32-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
+; RV32-NEXT:    .cfi_restore ra
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    .cfi_def_cfa_offset 0
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: llround_v1i64_v1f64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV64-NEXT:    vfmv.f.s fa5, v8
+; RV64-NEXT:    fcvt.l.d a0, fa5, rmm
+; RV64-NEXT:    vmv.s.x v8, a0
+; RV64-NEXT:    ret
+  %a = call <1 x i64> @llvm.llround.v1i64.v1f64(<1 x double> %x)
+  ret <1 x i64> %a
+}
+declare <1 x i64> @llvm.llround.v1i64.v1f64(<1 x double>)
+
+define <2 x i64> @llround_v2i64_v2f64(<2 x double> %x) {
+; RV32-LABEL: llround_v2i64_v2f64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -32
+; RV32-NEXT:    .cfi_def_cfa_offset 32
+; RV32-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
+; RV32-NEXT:    .cfi_offset ra, -4
+; RV32-NEXT:    csrr a0, vlenb
+; RV32-NEXT:    slli a0, a0, 1
+; RV32-NEXT:    sub sp, sp, a0
+; RV32-NEXT:    .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x20, 0x22, 0x11, 0x02, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 32 + 2 * vlenb
+; RV32-NEXT:    addi a0, sp, 16
+; RV32-NEXT:    vs1r.v v8, (a0) # vscale x 8-byte Folded Spill
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llround
+; RV32-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
+; RV32-NEXT:    vmv.v.x v8, a0
+; RV32-NEXT:    vslide1down.vx v8, v8, a1
+; RV32-NEXT:    csrr a0, vlenb
+; RV32-NEXT:    add a0, sp, a0
+; RV32-NEXT:    addi a0, a0, 16
+; RV32-NEXT:    vs1r.v v8, (a0) # vscale x 8-byte Folded Spill
+; RV32-NEXT:    addi a0, sp, 16
+; RV32-NEXT:    vl1r.v v8, (a0) # vscale x 8-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV32-NEXT:    vslidedown.vi v8, v8, 1
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llround
+; RV32-NEXT:    csrr a2, vlenb
+; RV32-NEXT:    add a2, sp, a2
+; RV32-NEXT:    addi a2, a2, 16
+; RV32-NEXT:    vl1r.v v8, (a2) # vscale x 8-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
+; RV32-NEXT:    vslide1down.vx v8, v8, a0
+; RV32-NEXT:    vslide1down.vx v8, v8, a1
+; RV32-NEXT:    csrr a0, vlenb
+; RV32-NEXT:    slli a0, a0, 1
+; RV32-NEXT:    add sp, sp, a0
+; RV32-NEXT:    .cfi_def_cfa sp, 32
+; RV32-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
+; RV32-NEXT:    .cfi_restore ra
+; RV32-NEXT:    addi sp, sp, 32
+; RV32-NEXT:    .cfi_def_cfa_offset 0
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: llround_v2i64_v2f64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
+; RV64-NEXT:    vslidedown.vi v9, v8, 1
+; RV64-NEXT:    vfmv.f.s fa5, v8
+; RV64-NEXT:    fcvt.l.d a0, fa5, rmm
+; RV64-NEXT:    vfmv.f.s fa5, v9
+; RV64-NEXT:    fcvt.l.d a1, fa5, rmm
+; RV64-NEXT:    vmv.v.x v8, a0
+; RV64-NEXT:    vslide1down.vx v8, v8, a1
+; RV64-NEXT:    ret
+  %a = call <2 x i64> @llvm.llround.v2i64.v2f64(<2 x double> %x)
+  ret <2 x i64> %a
+}
+declare <2 x i64> @llvm.llround.v2i64.v2f64(<2 x double>)
+
+define <4 x i64> @llround_v4i64_v4f64(<4 x double> %x) {
+; RV32-LABEL: llround_v4i64_v4f64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -32
+; RV32-NEXT:    .cfi_def_cfa_offset 32
+; RV32-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
+; RV32-NEXT:    .cfi_offset ra, -4
+; RV32-NEXT:    csrr a0, vlenb
+; RV32-NEXT:    slli a0, a0, 2
+; RV32-NEXT:    sub sp, sp, a0
+; RV32-NEXT:    .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x20, 0x22, 0x11, 0x04, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 32 + 4 * vlenb
+; RV32-NEXT:    csrr a0, vlenb
+; RV32-NEXT:    slli a0, a0, 1
+; RV32-NEXT:    add a0, sp, a0
+; RV32-NEXT:    addi a0, a0, 16
+; RV32-NEXT:    vs2r.v v8, (a0) # vscale x 16-byte Folded Spill
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llround
+; RV32-NEXT:    vsetivli zero, 8, e32, m2, ta, ma
+; RV32-NEXT:    vmv.v.x v8, a0
+; RV32-NEXT:    vslide1down.vx v8, v8, a1
+; RV32-NEXT:    addi a0, sp, 16
+; RV32-NEXT:    vs2r.v v8, (a0) # vscale x 16-byte Folded Spill
+; RV32-NEXT:    csrr a0, vlenb
+; RV32-NEXT:    slli a0, a0, 1
+; RV32-NEXT:    add a0, sp, a0
+; RV32-NEXT:    addi a0, a0, 16
+; RV32-NEXT:    vl2r.v v8, (a0) # vscale x 16-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV32-NEXT:    vslidedown.vi v8, v8, 1
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llround
+; RV32-NEXT:    addi a2, sp, 16
+; RV32-NEXT:    vl2r.v v8, (a2) # vscale x 16-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 8, e32, m2, ta, ma
+; RV32-NEXT:    vslide1down.vx v8, v8, a0
+; RV32-NEXT:    vslide1down.vx v8, v8, a1
+; RV32-NEXT:    addi a0, sp, 16
+; RV32-NEXT:    vs2r.v v8, (a0) # vscale x 16-byte Folded Spill
+; RV32-NEXT:    csrr a0, vlenb
+; RV32-NEXT:    slli a0, a0, 1
+; RV32-NEXT:    add a0, sp, a0
+; RV32-NEXT:    addi a0, a0, 16
+; RV32-NEXT:    vl2r.v v8, (a0) # vscale x 16-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 1, e64, m2, ta, ma
+; RV32-NEXT:    vslidedown.vi v8, v8, 2
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llround
+; RV32-NEXT:    addi a2, sp, 16
+; RV32-NEXT:    vl2r.v v8, (a2) # vscale x 16-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 8, e32, m2, ta, ma
+; RV32-NEXT:    vslide1down.vx v8, v8, a0
+; RV32-NEXT:    vslide1down.vx v8, v8, a1
+; RV32-NEXT:    addi a0, sp, 16
+; RV32-NEXT:    vs2r.v v8, (a0) # vscale x 16-byte Folded Spill
+; RV32-NEXT:    csrr a0, vlenb
+; RV32-NEXT:    slli a0, a0, 1
+; RV32-NEXT:    add a0, sp, a0
+; RV32-NEXT:    addi a0, a0, 16
+; RV32-NEXT:    vl2r.v v8, (a0) # vscale x 16-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 1, e64, m2, ta, ma
+; RV32-NEXT:    vslidedown.vi v8, v8, 3
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llround
+; RV32-NEXT:    addi a2, sp, 16
+; RV32-NEXT:    vl2r.v v8, (a2) # vscale x 16-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 8, e32, m2, ta, ma
+; RV32-NEXT:    vslide1down.vx v8, v8, a0
+; RV32-NEXT:    vslide1down.vx v8, v8, a1
+; RV32-NEXT:    csrr a0, vlenb
+; RV32-NEXT:    slli a0, a0, 2
+; RV32-NEXT:    add sp, sp, a0
+; RV32-NEXT:    .cfi_def_cfa sp, 32
+; RV32-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
+; RV32-NEXT:    .cfi_restore ra
+; RV32-NEXT:    addi sp, sp, 32
+; RV32-NEXT:    .cfi_def_cfa_offset 0
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: llround_v4i64_v4f64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV64-NEXT:    vslidedown.vi v12, v8, 1
+; RV64-NEXT:    vfmv.f.s fa5, v8
+; RV64-NEXT:    vsetivli zero, 1, e64, m2, ta, ma
+; RV64-NEXT:    vslidedown.vi v10, v8, 2
+; RV64-NEXT:    vslidedown.vi v8, v8, 3
+; RV64-NEXT:    fcvt.l.d a0, fa5, rmm
+; RV64-NEXT:    vfmv.f.s fa5, v12
+; RV64-NEXT:    fcvt.l.d a1, fa5, rmm
+; RV64-NEXT:    vfmv.f.s fa5, v10
+; RV64-NEXT:    vsetivli zero, 4, e64, m2, ta, ma
+; RV64-NEXT:    vmv.v.x v10, a0
+; RV64-NEXT:    fcvt.l.d a0, fa5, rmm
+; RV64-NEXT:    vfmv.f.s fa5, v8
+; RV64-NEXT:    vslide1down.vx v8, v10, a1
+; RV64-NEXT:    vslide1down.vx v8, v8, a0
+; RV64-NEXT:    fcvt.l.d a0, fa5, rmm
+; RV64-NEXT:    vslide1down.vx v8, v8, a0
+; RV64-NEXT:    ret
+  %a = call <4 x i64> @llvm.llround.v4i64.v4f64(<4 x double> %x)
+  ret <4 x i64> %a
+}
+declare <4 x i64> @llvm.llround.v4i64.v4f64(<4 x double>)
+
+define <8 x i64> @llround_v8i64_v8f64(<8 x double> %x) {
+; RV32-LABEL: llround_v8i64_v8f64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -272
+; RV32-NEXT:    .cfi_def_cfa_offset 272
+; RV32-NEXT:    sw ra, 268(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw s0, 264(sp) # 4-byte Folded Spill
+; RV32-NEXT:    .cfi_offset ra, -4
+; RV32-NEXT:    .cfi_offset s0, -8
+; RV32-NEXT:    addi s0, sp, 272
+; RV32-NEXT:    .cfi_def_cfa s0, 0
+; RV32-NEXT:    csrr a0, vlenb
+; RV32-NEXT:    slli a0, a0, 2
+; RV32-NEXT:    sub sp, sp, a0
+; RV32-NEXT:    andi sp, sp, -64
+; RV32-NEXT:    addi a0, sp, 256
+; RV32-NEXT:    vs4r.v v8, (a0) # vscale x 32-byte Folded Spill
+; RV32-NEXT:    addi a0, sp, 64
+; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, ma
+; RV32-NEXT:    vse64.v v8, (a0)
+; RV32-NEXT:    fld fa0, 120(sp)
+; RV32-NEXT:    call llround
+; RV32-NEXT:    sw a0, 184(sp)
+; RV32-NEXT:    sw a1, 188(sp)
+; RV32-NEXT:    fld fa0, 112(sp)
+; RV32-NEXT:    call llround
+; RV32-NEXT:    sw a0, 176(sp)
+; RV32-NEXT:    sw a1, 180(sp)
+; RV32-NEXT:    fld fa0, 104(sp)
+; RV32-NEXT:    call llround
+; RV32-NEXT:    sw a0, 168(sp)
+; RV32-NEXT:    sw a1, 172(sp)
+; RV32-NEXT:    fld fa0, 96(sp)
+; RV32-NEXT:    call llround
+; RV32-NEXT:    sw a0, 160(sp)
+; RV32-NEXT:    sw a1, 164(sp)
+; RV32-NEXT:    addi a0, sp, 256
+; RV32-NEXT:    vl4r.v v8, (a0) # vscale x 32-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llround
+; RV32-NEXT:    sw a0, 128(sp)
+; RV32-NEXT:    sw a1, 132(sp)
+; RV32-NEXT:    addi a0, sp, 256
+; RV32-NEXT:    vl4r.v v8, (a0) # vscale x 32-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV32-NEXT:    vslidedown.vi v8, v8, 1
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llround
+; RV32-NEXT:    sw a0, 136(sp)
+; RV32-NEXT:    sw a1, 140(sp)
+; RV32-NEXT:    addi a0, sp, 256
+; RV32-NEXT:    vl4r.v v8, (a0) # vscale x 32-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 1, e64, m2, ta, ma
+; RV32-NEXT:    vslidedown.vi v8, v8, 3
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llround
+; RV32-NEXT:    sw a0, 152(sp)
+; RV32-NEXT:    sw a1, 156(sp)
+; RV32-NEXT:    addi a0, sp, 256
+; RV32-NEXT:    vl4r.v v8, (a0) # vscale x 32-byte Folded Reload
+; RV32-NEXT:    vsetivli zero, 1, e64, m2, ta, ma
+; RV32-NEXT:    vslidedown.vi v8, v8, 2
+; RV32-NEXT:    vfmv.f.s fa0, v8
+; RV32-NEXT:    call llround
+; RV32-NEXT:    sw a0, 144(sp)
+; RV32-NEXT:    sw a1, 148(sp)
+; RV32-NEXT:    addi a0, sp, 128
+; RV32-NEXT:    vsetivli zero, 16, e32, m4, ta, ma
+; RV32-NEXT:    vle32.v v8, (a0)
+; RV32-NEXT:    addi sp, s0, -272
+; RV32-NEXT:    .cfi_def_cfa sp, 272
+; RV32-NEXT:    lw ra, 268(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw s0, 264(sp) # 4-byte Folded Reload
+; RV32-NEXT:    .cfi_restore ra
+; RV32-NEXT:    .cfi_restore s0
+; RV32-NEXT:    addi sp, sp, 272
+; RV32-NEXT:    .cfi_def_cfa_offset 0
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: llround_v8i64_v8f64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    addi sp, sp, -192
+; RV64-NEXT:    .cfi_def_cfa_offset 192
+; RV64-NEXT:    sd ra, 184(sp) # 8-byte Folded Spill
+; RV64-NEXT:    sd s0, 176(sp) # 8-byte Folded Spill
+; RV64-NEXT:    .cfi_offset ra, -8
+; RV64-NEXT:    .cfi_offset s0, -16
+; RV64-NEXT:    addi s0, sp, 192
+; RV64-NEXT:    .cfi_def_cfa s0, 0
+; RV64-NEXT:    andi sp, sp, -64
+; RV64-NEXT:    mv a0, sp
+; RV64-NEXT:    vsetivli zero, 8, e64, m4, ta, ma
+; RV64-NEXT:    vse64.v v8, (a0)
+; RV64-NEXT:    fld fa5, 56(sp)
+; RV64-NEXT:    vfmv.f.s fa4, v8
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV64-NEXT:    vslidedown.vi v10, v8, 1
+; RV64-NEXT:    fcvt.l.d a0, fa4, rmm
+; RV64-NEXT:    fcvt.l.d a1, fa5, rmm
+; RV64-NEXT:    sd a1, 120(sp)
+; RV64-NEXT:    fld fa5, 48(sp)
+; RV64-NEXT:    vfmv.f.s fa4, v10
+; RV64-NEXT:    vsetivli zero, 1, e64, m2, ta, ma
+; RV64-NEXT:    vslidedown.vi v10, v8, 3
+; RV64-NEXT:    fcvt.l.d a1, fa4, rmm
+; RV64-NEXT:    fcvt.l.d a2, fa5, rmm
+; RV64-NEXT:    sd a2, 112(sp)
+; RV64-NEXT:    fld fa5, 40(sp)
+; RV64-NEXT:    vfmv.f.s fa4, v10
+; RV64-NEXT:    fcvt.l.d a2, fa4, rmm
+; RV64-NEXT:    vslidedown.vi v8, v8, 2
+; RV64-NEXT:    fcvt.l.d a3, fa5, rmm
+; RV64-NEXT:    vfmv.f.s fa5, v8
+; RV64-NEXT:    sd a3, 104(sp)
+; RV64-NEXT:    fcvt.l.d a3, fa5, rmm
+; RV64-NEXT:    fld fa5, 32(sp)
+; RV64-NEXT:    sd a0, 64(sp)
+; RV64-NEXT:    sd a1, 72(sp)
+; RV64-NEXT:    sd a3, 80(sp)
+; RV64-NEXT:    sd a2, 88(sp)
+; RV64-NEXT:    fcvt.l.d a0, fa5, rmm
+; RV64-NEXT:    sd a0, 96(sp)
+; RV64-NEXT:    addi a0, sp, 64
+; RV64-NEXT:    vsetivli zero, 8, e64, m4, ta, ma
+; RV64-NEXT:    vle64.v v8, (a0)
+; RV64-NEXT:    addi sp, s0, -192
+; RV64-NEXT:    .cfi_def_cfa sp, 192
+; RV64-NEXT:    ld ra, 184(sp) # 8-byte Folded Reload
+; RV64-NEXT:    ld s0, 176(sp) # 8-byte Folded Reload
+; RV64-NEXT:    .cfi_restore ra
+; RV64-NEXT:    .cfi_restore s0
+; RV64-NEXT:    addi sp, sp, 192
+; RV64-NEXT:    .cfi_def_cfa_offset 0
+; RV64-NEXT:    ret
+  %a = call <8 x i64> @llvm.llround.v8i64.v8f64(<8 x double> %x)
+  ret <8 x i64> %a
+}
+declare <8 x i64> @llvm.llround.v8i64.v8f64(<8 x double>)
diff --git a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-lround.ll b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-lround.ll
new file mode 100644
index 0000000000000..ca2b221828990
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-lround.ll
@@ -0,0 +1,926 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+f,+d \
+; RUN:     -target-abi=ilp32d -verify-machineinstrs | FileCheck %s --check-prefix=RV32
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv64 -mattr=+v,+f,+d \
+; RUN:     -target-abi=lp64d -verify-machineinstrs | FileCheck %s --check-prefix=RV64-i32
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+f,+d \
+; RUN:     -target-abi=lp64d -verify-machineinstrs | FileCheck %s --check-prefix=RV64-i64
+
+define <1 x iXLen> @lround_v1f32(<1 x float> %x) {
+; RV32-LABEL: lround_v1f32:
+; RV32:       # %bb.0:
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV32-NEXT:    vfmv.f.s fa5, v8
+; RV32-NEXT:    fcvt.w.s a0, fa5, rmm
+; RV32-NEXT:    vmv.s.x v8, a0
+; RV32-NEXT:    ret
+;
+; RV64-i32-LABEL: lround_v1f32:
+; RV64-i32:       # %bb.0:
+; RV64-i32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV64-i32-NEXT:    vfmv.f.s fa5, v8
+; RV64-i32-NEXT:    fcvt.w.s a0, fa5, rmm
+; RV64-i32-NEXT:    vmv.s.x v8, a0
+; RV64-i32-NEXT:    ret
+;
+; RV64-i64-LABEL: lround_v1f32:
+; RV64-i64:       # %bb.0:
+; RV64-i64-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV64-i64-NEXT:    vfmv.f.s fa5, v8
+; RV64-i64-NEXT:    fcvt.l.s a0, fa5, rmm
+; RV64-i64-NEXT:    vsetvli zero, zero, e64, m2, ta, ma
+; RV64-i64-NEXT:    vmv.s.x v8, a0
+; RV64-i64-NEXT:    ret
+  %a = call <1 x iXLen> @llvm.lround.v1iXLen.v1f32(<1 x float> %x)
+  ret <1 x iXLen> %a
+}
+declare <1 x iXLen> @llvm.lround.v1iXLen.v1f32(<1 x float>)
+
+define <2 x iXLen> @lround_v2f32(<2 x float> %x) {
+; RV32-LABEL: lround_v2f32:
+; RV32:       # %bb.0:
+; RV32-NEXT:    vsetivli zero, 2, e32, mf2, ta, ma
+; RV32-NEXT:    vslidedown.vi v9, v8, 1
+; RV32-NEXT:    vfmv.f.s fa5, v8
+; RV32-NEXT:    fcvt.w.s a0, fa5, rmm
+; RV32-NEXT:    vfmv.f.s fa5, v9
+; RV32-NEXT:    fcvt.w.s a1, fa5, rmm
+; RV32-NEXT:    vmv.v.x v8, a0
+; RV32-NEXT:    vslide1down.vx v8, v8, a1
+; RV32-NEXT:    ret
+;
+; RV64-i32-LABEL: lround_v2f32:
+; RV64-i32:       # %bb.0:
+; RV64-i32-NEXT:    vsetivli zero, 2, e32, mf2, ta, ma
+; RV64-i32-NEXT:    vslidedown.vi v9, v8, 1
+; RV64-i32-NEXT:    vfmv.f.s fa5, v8
+; RV64-i32-NEXT:    fcvt.w.s a0, fa5, rmm
+; RV64-i32-NEXT:    vfmv.f.s fa5, v9
+; RV64-i32-NEXT:    fcvt.w.s a1, fa5, rmm
+; RV64-i32-NEXT:    vmv.v.x v8, a0
+; RV64-i32-NEXT:    vslide1down.vx v8, v8, a1
+; RV64-i32-NEXT:    ret
+;
+; RV64-i64-LABEL: lround_v2f32:
+; RV64-i64:       # %bb.0:
+; RV64-i64-NEXT:    vsetivli zero, 1, e32, mf2, ta, ma
+; RV64-i64-NEXT:    vslidedown.vi v9, v8, 1
+; RV64-i64-NEXT:    vfmv.f.s fa5, v8
+; RV64-i64-NEXT:    fcvt.l.s a0, fa5, rmm
+; RV64-i64-NEXT:    vfmv.f.s fa5, v9
+; RV64-i64-NEXT:    fcvt.l.s a1, fa5, rmm
+; RV64-i64-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
+; RV64-i64-NEXT:    vmv.v.x v8, a0
+; RV64-i64-NEXT:    vslide1down.vx v8, v8, a1
+; RV64-i64-NEXT:    ret
+  %a = call <2 x iXLen> @llvm.lround.v2iXLen.v2f32(<2 x float> %x)
+  ret <2 x iXLen> %a
+}
+declare <2 x iXLen> @llvm.lround.v2iXLen.v2f32(<2 x float>)
+
+define <3 x iXLen> @lround_v3f32(<3 x float> %x) {
+; RV32-LABEL: lround_v3f32:
+; RV32:       # %bb.0:
+; RV32-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
+; RV32-NEXT:    vslidedown.vi v9, v8, 1
+; RV32-NEXT:    vfmv.f.s fa5, v8
+; RV32-NEXT:    vslidedown.vi v10, v8, 2
+; RV32-NEXT:    vslidedown.vi v8, v8, 3
+; RV32-NEXT:    fcvt.w.s a0, fa5, rmm
+; RV32-NEXT:    vfmv.f.s fa5, v9
+; RV32-NEXT:    fcvt.w.s a1, fa5, rmm
+; RV32-NEXT:    vfmv.f.s fa5, v10
+; RV32-NEXT:    vmv.v.x v9, a0
+; RV32-NEXT:    fcvt.w.s a0, fa5, rmm
+; RV32-NEXT:    vfmv.f.s fa5, v8
+; RV32-NEXT:    vslide1down.vx v8, v9, a1
+; RV32-NEXT:    vslide1down.vx v8, v8, a0
+; RV32-NEXT:    fcvt.w.s a0, fa5, rmm
+; RV32-NEXT:    vslide1down.vx v8, v8, a0
+; RV32-NEXT:    ret
+;
+; RV64-i32-LABEL: lround_v3f32:
+; RV64-i32:       # %bb.0:
+; RV64-i32-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
+; RV64-i32-NEXT:    vslidedown.vi v9, v8, 1
+; RV64-i32-NEXT:    vfmv.f.s fa5, v8
+; RV64-i32-NEXT:    vslidedown.vi v10, v8, 2
+; RV64-i32-NEXT:    vslidedown.vi v8, v8, 3
+; RV64-i32-NEXT:    fcvt.w.s a0, fa5, rmm
+; RV64-i32-NEXT:    vfmv.f.s fa5, v9
+; RV64-i32-NEXT:    fcvt.w.s a1, fa5, rmm
+; RV64-i32-NEXT:    vfmv.f.s fa5, v10
+; RV64-i32-NEXT:    vmv.v.x v9, a0
+; RV64-i32-NEXT:    fcvt.w.s a0, fa5, rmm
+; RV64-i32-NEXT:    vfmv.f.s fa5, v8
+; RV64-i32-NEXT:    vslide1down.vx v8, v9, a1
+; RV64-i32-NEXT:    vslide1down.vx v8, v8, a0
+; RV64-i32-NEXT:    fcvt.w.s a0, fa5, rmm
+; RV64-i32-NEXT:    vslide1down.vx v8, v8, a0
+; RV64-i32-NEXT:    ret
+;
+; RV64-i64-LABEL: lround_v3f32:
+; RV64-i64:       # %bb.0:
+; RV64-i64-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV64-i64-NEXT:    vslidedown.vi v9, v8, 1
+; RV64-i64-NEXT:    vfmv.f.s fa5, v8
+; RV64-i64-NEXT:    vslidedown.vi v10, v8, 2
+; RV64-i64-NEXT:    vslidedown.vi v11, v8, 3
+; RV64-i64-NEXT:    fcvt.l.s a0, fa5, rmm
+; RV64-i64-NEXT:    vfmv.f.s fa5, v9
+; RV64-i64-NEXT:    fcvt.l.s a1, fa5, rmm
+; RV64-i64-NEXT:    vfmv.f.s fa5, v10
+; RV64-i64-NEXT:    vsetivli zero, 4, e64, m2, ta, ma
+; RV64-i64-NEXT:    vmv.v.x v8, a0
+; RV64-i64-NEXT:    fcvt.l.s a0, fa5, rmm
+; RV64-i64-NEXT:    vsetvli zero, zero, e32, m1, ta, ma
+; RV64-i64-NEXT:    vfmv.f.s fa5, v11
+; RV64-i64-NEXT:    vsetvli zero, zero, e64, m2, ta, ma
+; RV64-i64-NEXT:    vslide1down.vx v8, v8, a1
+; RV64-i64-NEXT:    vslide1down.vx v8, v8, a0
+; RV64-i64-NEXT:    fcvt.l.s a0, fa5, rmm
+; RV64-i64-NEXT:    vslide1down.vx v8, v8, a0
+; RV64-i64-NEXT:    ret
+  %a = call <3 x iXLen> @llvm.lround.v3iXLen.v3f32(<3 x float> %x)
+  ret <3 x iXLen> %a
+}
+declare <3 x iXLen> @llvm.lround.v3iXLen.v3f32(<3 x float>)
+
+define <4 x iXLen> @lround_v4f32(<4 x float> %x) {
+; RV32-LABEL: lround_v4f32:
+; RV32:       # %bb.0:
+; RV32-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
+; RV32-NEXT:    vslidedown.vi v9, v8, 1
+; RV32-NEXT:    vfmv.f.s fa5, v8
+; RV32-NEXT:    vslidedown.vi v10, v8, 2
+; RV32-NEXT:    vslidedown.vi v8, v8, 3
+; RV32-NEXT:    fcvt.w.s a0, fa5, rmm
+; RV32-NEXT:    vfmv.f.s fa5, v9
+; RV32-NEXT:    fcvt.w.s a1, fa5, rmm
+; RV32-NEXT:    vfmv.f.s fa5, v10
+; RV32-NEXT:    vmv.v.x v9, a0
+; RV32-NEXT:    fcvt.w.s a0, fa5, rmm
+; RV32-NEXT:    vfmv.f.s fa5, v8
+; RV32-NEXT:    vslide1down.vx v8, v9, a1
+; RV32-NEXT:    vslide1down.vx v8, v8, a0
+; RV32-NEXT:    fcvt.w.s a0, fa5, rmm
+; RV32-NEXT:    vslide1down.vx v8, v8, a0
+; RV32-NEXT:    ret
+;
+; RV64-i32-LABEL: lround_v4f32:
+; RV64-i32:       # %bb.0:
+; RV64-i32-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
+; RV64-i32-NEXT:    vslidedown.vi v9, v8, 1
+; RV64-i32-NEXT:    vfmv.f.s fa5, v8
+; RV64-i32-NEXT:    vslidedown.vi v10, v8, 2
+; RV64-i32-NEXT:    vslidedown.vi v8, v8, 3
+; RV64-i32-NEXT:    fcvt.w.s a0, fa5, rmm
+; RV64-i32-NEXT:    vfmv.f.s fa5, v9
+; RV64-i32-NEXT:    fcvt.w.s a1, fa5, rmm
+; RV64-i32-NEXT:    vfmv.f.s fa5, v10
+; RV64-i32-NEXT:    vmv.v.x v9, a0
+; RV64-i32-NEXT:    fcvt.w.s a0, fa5, rmm
+; RV64-i32-NEXT:    vfmv.f.s fa5, v8
+; RV64-i32-NEXT:    vslide1down.vx v8, v9, a1
+; RV64-i32-NEXT:    vslide1down.vx v8, v8, a0
+; RV64-i32-NEXT:    fcvt.w.s a0, fa5, rmm
+; RV64-i32-NEXT:    vslide1down.vx v8, v8, a0
+; RV64-i32-NEXT:    ret
+;
+; RV64-i64-LABEL: lround_v4f32:
+; RV64-i64:       # %bb.0:
+; RV64-i64-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV64-i64-NEXT:    vslidedown.vi v9, v8, 1
+; RV64-i64-NEXT:    vfmv.f.s fa5, v8
+; RV64-i64-NEXT:    vslidedown.vi v10, v8, 2
+; RV64-i64-NEXT:    vslidedown.vi v11, v8, 3
+; RV64-i64-NEXT:    fcvt.l.s a0, fa5, rmm
+; RV64-i64-NEXT:    vfmv.f.s fa5, v9
+; RV64-i64-NEXT:    fcvt.l.s a1, fa5, rmm
+; RV64-i64-NEXT:    vfmv.f.s fa5, v10
+; RV64-i64-NEXT:    vsetivli zero, 4, e64, m2, ta, ma
+; RV64-i64-NEXT:    vmv.v.x v8, a0
+; RV64-i64-NEXT:    fcvt.l.s a0, fa5, rmm
+; RV64-i64-NEXT:    vsetvli zero, zero, e32, m1, ta, ma
+; RV64-i64-NEXT:    vfmv.f.s fa5, v11
+; RV64-i64-NEXT:    vsetvli zero, zero, e64, m2, ta, ma
+; RV64-i64-NEXT:    vslide1down.vx v8, v8, a1
+; RV64-i64-NEXT:    vslide1down.vx v8, v8, a0
+; RV64-i64-NEXT:    fcvt.l.s a0, fa5, rmm
+; RV64-i64-NEXT:    vslide1down.vx v8, v8, a0
+; RV64-i64-NEXT:    ret
+  %a = call <4 x iXLen> @llvm.lround.v4iXLen.v4f32(<4 x float> %x)
+  ret <4 x iXLen> %a
+}
+declare <4 x iXLen> @llvm.lround.v4iXLen.v4f32(<4 x float>)
+
+define <8 x iXLen> @lround_v8f32(<8 x float> %x) {
+; RV32-LABEL: lround_v8f32:
+; RV32:       # %bb.0:
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV32-NEXT:    vslidedown.vi v10, v8, 1
+; RV32-NEXT:    vfmv.f.s fa5, v8
+; RV32-NEXT:    vslidedown.vi v11, v8, 2
+; RV32-NEXT:    vslidedown.vi v12, v8, 3
+; RV32-NEXT:    fcvt.w.s a0, fa5, rmm
+; RV32-NEXT:    vfmv.f.s fa5, v10
+; RV32-NEXT:    fcvt.w.s a1, fa5, rmm
+; RV32-NEXT:    vfmv.f.s fa5, v11
+; RV32-NEXT:    vsetivli zero, 8, e32, m2, ta, ma
+; RV32-NEXT:    vmv.v.x v10, a0
+; RV32-NEXT:    fcvt.w.s a0, fa5, rmm
+; RV32-NEXT:    vfmv.f.s fa5, v12
+; RV32-NEXT:    vslidedown.vi v12, v8, 4
+; RV32-NEXT:    vslide1down.vx v10, v10, a1
+; RV32-NEXT:    fcvt.w.s a1, fa5, rmm
+; RV32-NEXT:    vfmv.f.s fa5, v12
+; RV32-NEXT:    vslidedown.vi v12, v8, 5
+; RV32-NEXT:    vslide1down.vx v10, v10, a0
+; RV32-NEXT:    fcvt.w.s a0, fa5, rmm
+; RV32-NEXT:    vfmv.f.s fa5, v12
+; RV32-NEXT:    vslidedown.vi v12, v8, 6
+; RV32-NEXT:    vslidedown.vi v8, v8, 7
+; RV32-NEXT:    vslide1down.vx v10, v10, a1
+; RV32-NEXT:    fcvt.w.s a1, fa5, rmm
+; RV32-NEXT:    vfmv.f.s fa5, v12
+; RV32-NEXT:    vslide1down.vx v10, v10, a0
+; RV32-NEXT:    fcvt.w.s a0, fa5, rmm
+; RV32-NEXT:    vfmv.f.s fa5, v8
+; RV32-NEXT:    vslide1down.vx v8, v10, a1
+; RV32-NEXT:    vslide1down.vx v8, v8, a0
+; RV32-NEXT:    fcvt.w.s a0, fa5, rmm
+; RV32-NEXT:    vslide1down.vx v8, v8, a0
+; RV32-NEXT:    ret
+;
+; RV64-i32-LABEL: lround_v8f32:
+; RV64-i32:       # %bb.0:
+; RV64-i32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV64-i32-NEXT:    vslidedown.vi v10, v8, 1
+; RV64-i32-NEXT:    vfmv.f.s fa5, v8
+; RV64-i32-NEXT:    vslidedown.vi v11, v8, 2
+; RV64-i32-NEXT:    vslidedown.vi v12, v8, 3
+; RV64-i32-NEXT:    fcvt.w.s a0, fa5, rmm
+; RV64-i32-NEXT:    vfmv.f.s fa5, v10
+; RV64-i32-NEXT:    fcvt.w.s a1, fa5, rmm
+; RV64-i32-NEXT:    vfmv.f.s fa5, v11
+; RV64-i32-NEXT:    vsetivli zero, 8, e32, m2, ta, ma
+; RV64-i32-NEXT:    vmv.v.x v10, a0
+; RV64-i32-NEXT:    fcvt.w.s a0, fa5, rmm
+; RV64-i32-NEXT:    vfmv.f.s fa5, v12
+; RV64-i32-NEXT:    vslidedown.vi v12, v8, 4
+; RV64-i32-NEXT:    vslide1down.vx v10, v10, a1
+; RV64-i32-NEXT:    fcvt.w.s a1, fa5, rmm
+; RV64-i32-NEXT:    vfmv.f.s fa5, v12
+; RV64-i32-NEXT:    vslidedown.vi v12, v8, 5
+; RV64-i32-NEXT:    vslide1down.vx v10, v10, a0
+; RV64-i32-NEXT:    fcvt.w.s a0, fa5, rmm
+; RV64-i32-NEXT:    vfmv.f.s fa5, v12
+; RV64-i32-NEXT:    vslidedown.vi v12, v8, 6
+; RV64-i32-NEXT:    vslidedown.vi v8, v8, 7
+; RV64-i32-NEXT:    vslide1down.vx v10, v10, a1
+; RV64-i32-NEXT:    fcvt.w.s a1, fa5, rmm
+; RV64-i32-NEXT:    vfmv.f.s fa5, v12
+; RV64-i32-NEXT:    vslide1down.vx v10, v10, a0
+; RV64-i32-NEXT:    fcvt.w.s a0, fa5, rmm
+; RV64-i32-NEXT:    vfmv.f.s fa5, v8
+; RV64-i32-NEXT:    vslide1down.vx v8, v10, a1
+; RV64-i32-NEXT:    vslide1down.vx v8, v8, a0
+; RV64-i32-NEXT:    fcvt.w.s a0, fa5, rmm
+; RV64-i32-NEXT:    vslide1down.vx v8, v8, a0
+; RV64-i32-NEXT:    ret
+;
+; RV64-i64-LABEL: lround_v8f32:
+; RV64-i64:       # %bb.0:
+; RV64-i64-NEXT:    addi sp, sp, -128
+; RV64-i64-NEXT:    .cfi_def_cfa_offset 128
+; RV64-i64-NEXT:    sd ra, 120(sp) # 8-byte Folded Spill
+; RV64-i64-NEXT:    sd s0, 112(sp) # 8-byte Folded Spill
+; RV64-i64-NEXT:    .cfi_offset ra, -8
+; RV64-i64-NEXT:    .cfi_offset s0, -16
+; RV64-i64-NEXT:    addi s0, sp, 128
+; RV64-i64-NEXT:    .cfi_def_cfa s0, 0
+; RV64-i64-NEXT:    andi sp, sp, -64
+; RV64-i64-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
+; RV64-i64-NEXT:    vfmv.f.s fa5, v8
+; RV64-i64-NEXT:    vslidedown.vi v10, v8, 7
+; RV64-i64-NEXT:    fcvt.l.s a0, fa5, rmm
+; RV64-i64-NEXT:    vfmv.f.s fa5, v10
+; RV64-i64-NEXT:    vslidedown.vi v10, v8, 6
+; RV64-i64-NEXT:    fcvt.l.s a1, fa5, rmm
+; RV64-i64-NEXT:    vfmv.f.s fa5, v10
+; RV64-i64-NEXT:    vslidedown.vi v10, v8, 5
+; RV64-i64-NEXT:    fcvt.l.s a2, fa5, rmm
+; RV64-i64-NEXT:    vfmv.f.s fa5, v10
+; RV64-i64-NEXT:    vslidedown.vi v10, v8, 4
+; RV64-i64-NEXT:    fcvt.l.s a3, fa5, rmm
+; RV64-i64-NEXT:    vfmv.f.s fa5, v10
+; RV64-i64-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV64-i64-NEXT:    vslidedown.vi v9, v8, 3
+; RV64-i64-NEXT:    vslidedown.vi v10, v8, 2
+; RV64-i64-NEXT:    vslidedown.vi v8, v8, 1
+; RV64-i64-NEXT:    fcvt.l.s a4, fa5, rmm
+; RV64-i64-NEXT:    vfmv.f.s fa5, v9
+; RV64-i64-NEXT:    fcvt.l.s a5, fa5, rmm
+; RV64-i64-NEXT:    vfmv.f.s fa5, v10
+; RV64-i64-NEXT:    fcvt.l.s a6, fa5, rmm
+; RV64-i64-NEXT:    vfmv.f.s fa5, v8
+; RV64-i64-NEXT:    sd a4, 32(sp)
+; RV64-i64-NEXT:    sd a3, 40(sp)
+; RV64-i64-NEXT:    sd a2, 48(sp)
+; RV64-i64-NEXT:    sd a1, 56(sp)
+; RV64-i64-NEXT:    fcvt.l.s a1, fa5, rmm
+; RV64-i64-NEXT:    sd a0, 0(sp)
+; RV64-i64-NEXT:    sd a1, 8(sp)
+; RV64-i64-NEXT:    sd a6, 16(sp)
+; RV64-i64-NEXT:    sd a5, 24(sp)
+; RV64-i64-NEXT:    mv a0, sp
+; RV64-i64-NEXT:    vsetivli zero, 8, e64, m4, ta, ma
+; RV64-i64-NEXT:    vle64.v v8, (a0)
+; RV64-i64-NEXT:    addi sp, s0, -128
+; RV64-i64-NEXT:    .cfi_def_cfa sp, 128
+; RV64-i64-NEXT:    ld ra, 120(sp) # 8-byte Folded Reload
+; RV64-i64-NEXT:    ld s0, 112(sp) # 8-byte Folded Reload
+; RV64-i64-NEXT:    .cfi_restore ra
+; RV64-i64-NEXT:    .cfi_restore s0
+; RV64-i64-NEXT:    addi sp, sp, 128
+; RV64-i64-NEXT:    .cfi_def_cfa_offset 0
+; RV64-i64-NEXT:    ret
+  %a = call <8 x iXLen> @llvm.lround.v8iXLen.v8f32(<8 x float> %x)
+  ret <8 x iXLen> %a
+}
+declare <8 x iXLen> @llvm.lround.v8iXLen.v8f32(<8 x float>)
+
+define <16 x iXLen> @lround_v16f32(<16 x float> %x) {
+; RV32-LABEL: lround_v16f32:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -192
+; RV32-NEXT:    .cfi_def_cfa_offset 192
+; RV32-NEXT:    sw ra, 188(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw s0, 184(sp) # 4-byte Folded Spill
+; RV32-NEXT:    .cfi_offset ra, -4
+; RV32-NEXT:    .cfi_offset s0, -8
+; RV32-NEXT:    addi s0, sp, 192
+; RV32-NEXT:    .cfi_def_cfa s0, 0
+; RV32-NEXT:    andi sp, sp, -64
+; RV32-NEXT:    mv a0, sp
+; RV32-NEXT:    vsetivli zero, 16, e32, m4, ta, ma
+; RV32-NEXT:    vse32.v v8, (a0)
+; RV32-NEXT:    flw fa5, 60(sp)
+; RV32-NEXT:    vfmv.f.s fa4, v8
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV32-NEXT:    vslidedown.vi v10, v8, 3
+; RV32-NEXT:    vslidedown.vi v11, v8, 2
+; RV32-NEXT:    fcvt.w.s a0, fa5, rmm
+; RV32-NEXT:    sw a0, 124(sp)
+; RV32-NEXT:    flw fa5, 56(sp)
+; RV32-NEXT:    fcvt.w.s a0, fa4, rmm
+; RV32-NEXT:    vfmv.f.s fa4, v10
+; RV32-NEXT:    vslidedown.vi v10, v8, 1
+; RV32-NEXT:    fcvt.w.s a1, fa5, rmm
+; RV32-NEXT:    sw a1, 120(sp)
+; RV32-NEXT:    flw fa5, 52(sp)
+; RV32-NEXT:    fcvt.w.s a1, fa4, rmm
+; RV32-NEXT:    vfmv.f.s fa4, v11
+; RV32-NEXT:    fcvt.w.s a2, fa4, rmm
+; RV32-NEXT:    fcvt.w.s a3, fa5, rmm
+; RV32-NEXT:    sw a3, 116(sp)
+; RV32-NEXT:    flw fa5, 48(sp)
+; RV32-NEXT:    vfmv.f.s fa4, v10
+; RV32-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
+; RV32-NEXT:    vslidedown.vi v10, v8, 7
+; RV32-NEXT:    fcvt.w.s a3, fa4, rmm
+; RV32-NEXT:    fcvt.w.s a4, fa5, rmm
+; RV32-NEXT:    sw a4, 112(sp)
+; RV32-NEXT:    flw fa5, 44(sp)
+; RV32-NEXT:    vfmv.f.s fa4, v10
+; RV32-NEXT:    vslidedown.vi v10, v8, 6
+; RV32-NEXT:    fcvt.w.s a4, fa4, rmm
+; RV32-NEXT:    fcvt.w.s a5, fa5, rmm
+; RV32-NEXT:    sw a5, 108(sp)
+; RV32-NEXT:    flw fa5, 40(sp)
+; RV32-NEXT:    vfmv.f.s fa4, v10
+; RV32-NEXT:    vslidedown.vi v10, v8, 5
+; RV32-NEXT:    fcvt.w.s a5, fa4, rmm
+; RV32-NEXT:    fcvt.w.s a6, fa5, rmm
+; RV32-NEXT:    sw a6, 104(sp)
+; RV32-NEXT:    flw fa5, 36(sp)
+; RV32-NEXT:    vfmv.f.s fa4, v10
+; RV32-NEXT:    fcvt.w.s a6, fa4, rmm
+; RV32-NEXT:    vslidedown.vi v8, v8, 4
+; RV32-NEXT:    fcvt.w.s a7, fa5, rmm
+; RV32-NEXT:    vfmv.f.s fa5, v8
+; RV32-NEXT:    sw a7, 100(sp)
+; RV32-NEXT:    fcvt.w.s a7, fa5, rmm
+; RV32-NEXT:    flw fa5, 32(sp)
+; RV32-NEXT:    sw a0, 64(sp)
+; RV32-NEXT:    sw a3, 68(sp)
+; RV32-NEXT:    sw a2, 72(sp)
+; RV32-NEXT:    sw a1, 76(sp)
+; RV32-NEXT:    sw a7, 80(sp)
+; RV32-NEXT:    sw a6, 84(sp)
+; RV32-NEXT:    sw a5, 88(sp)
+; RV32-NEXT:    sw a4, 92(sp)
+; RV32-NEXT:    fcvt.w.s a0, fa5, rmm
+; RV32-NEXT:    sw a0, 96(sp)
+; RV32-NEXT:    addi a0, sp, 64
+; RV32-NEXT:    vsetivli zero, 16, e32, m4, ta, ma
+; RV32-NEXT:    vle32.v v8, (a0)
+; RV32-NEXT:    addi sp, s0, -192
+; RV32-NEXT:    .cfi_def_cfa sp, 192
+; RV32-NEXT:    lw ra, 188(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw s0, 184(sp) # 4-byte Folded Reload
+; RV32-NEXT:    .cfi_restore ra
+; RV32-NEXT:    .cfi_restore s0
+; RV32-NEXT:    addi sp, sp, 192
+; RV32-NEXT:    .cfi_def_cfa_offset 0
+; RV32-NEXT:    ret
+;
+; RV64-i32-LABEL: lround_v16f32:
+; RV64-i32:       # %bb.0:
+; RV64-i32-NEXT:    addi sp, sp, -192
+; RV64-i32-NEXT:    .cfi_def_cfa_offset 192
+; RV64-i32-NEXT:    sd ra, 184(sp) # 8-byte Folded Spill
+; RV64-i32-NEXT:    sd s0, 176(sp) # 8-byte Folded Spill
+; RV64-i32-NEXT:    .cfi_offset ra, -8
+; RV64-i32-NEXT:    .cfi_offset s0, -16
+; RV64-i32-NEXT:    addi s0, sp, 192
+; RV64-i32-NEXT:    .cfi_def_cfa s0, 0
+; RV64-i32-NEXT:    andi sp, sp, -64
+; RV64-i32-NEXT:    mv a0, sp
+; RV64-i32-NEXT:    vsetivli zero, 16, e32, m4, ta, ma
+; RV64-i32-NEXT:    vse32.v v8, (a0)
+; RV64-i32-NEXT:    flw fa5, 60(sp)
+; RV64-i32-NEXT:    vfmv.f.s fa4, v8
+; RV64-i32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV64-i32-NEXT:    vslidedown.vi v10, v8, 3
+; RV64-i32-NEXT:    vslidedown.vi v11, v8, 2
+; RV64-i32-NEXT:    fcvt.w.s a0, fa5, rmm
+; RV64-i32-NEXT:    sw a0, 124(sp)
+; RV64-i32-NEXT:    flw fa5, 56(sp)
+; RV64-i32-NEXT:    fcvt.w.s a0, fa4, rmm
+; RV64-i32-NEXT:    vfmv.f.s fa4, v10
+; RV64-i32-NEXT:    vslidedown.vi v10, v8, 1
+; RV64-i32-NEXT:    fcvt.w.s a1, fa5, rmm
+; RV64-i32-NEXT:    sw a1, 120(sp)
+; RV64-i32-NEXT:    flw fa5, 52(sp)
+; RV64-i32-NEXT:    fcvt.w.s a1, fa4, rmm
+; RV64-i32-NEXT:    vfmv.f.s fa4, v11
+; RV64-i32-NEXT:    fcvt.w.s a2, fa4, rmm
+; RV64-i32-NEXT:    fcvt.w.s a3, fa5, rmm
+; RV64-i32-NEXT:    sw a3, 116(sp)
+; RV64-i32-NEXT:    flw fa5, 48(sp)
+; RV64-i32-NEXT:    vfmv.f.s fa4, v10
+; RV64-i32-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
+; RV64-i32-NEXT:    vslidedown.vi v10, v8, 7
+; RV64-i32-NEXT:    fcvt.w.s a3, fa4, rmm
+; RV64-i32-NEXT:    fcvt.w.s a4, fa5, rmm
+; RV64-i32-NEXT:    sw a4, 112(sp)
+; RV64-i32-NEXT:    flw fa5, 44(sp)
+; RV64-i32-NEXT:    vfmv.f.s fa4, v10
+; RV64-i32-NEXT:    vslidedown.vi v10, v8, 6
+; RV64-i32-NEXT:    fcvt.w.s a4, fa4, rmm
+; RV64-i32-NEXT:    fcvt.w.s a5, fa5, rmm
+; RV64-i32-NEXT:    sw a5, 108(sp)
+; RV64-i32-NEXT:    flw fa5, 40(sp)
+; RV64-i32-NEXT:    vfmv.f.s fa4, v10
+; RV64-i32-NEXT:    vslidedown.vi v10, v8, 5
+; RV64-i32-NEXT:    fcvt.w.s a5, fa4, rmm
+; RV64-i32-NEXT:    fcvt.w.s a6, fa5, rmm
+; RV64-i32-NEXT:    sw a6, 104(sp)
+; RV64-i32-NEXT:    flw fa5, 36(sp)
+; RV64-i32-NEXT:    vfmv.f.s fa4, v10
+; RV64-i32-NEXT:    fcvt.w.s a6, fa4, rmm
+; RV64-i32-NEXT:    vslidedown.vi v8, v8, 4
+; RV64-i32-NEXT:    fcvt.w.s a7, fa5, rmm
+; RV64-i32-NEXT:    vfmv.f.s fa5, v8
+; RV64-i32-NEXT:    sw a7, 100(sp)
+; RV64-i32-NEXT:    fcvt.w.s a7, fa5, rmm
+; RV64-i32-NEXT:    flw fa5, 32(sp)
+; RV64-i32-NEXT:    sw a0, 64(sp)
+; RV64-i32-NEXT:    sw a3, 68(sp)
+; RV64-i32-NEXT:    sw a2, 72(sp)
+; RV64-i32-NEXT:    sw a1, 76(sp)
+; RV64-i32-NEXT:    sw a7, 80(sp)
+; RV64-i32-NEXT:    sw a6, 84(sp)
+; RV64-i32-NEXT:    sw a5, 88(sp)
+; RV64-i32-NEXT:    sw a4, 92(sp)
+; RV64-i32-NEXT:    fcvt.w.s a0, fa5, rmm
+; RV64-i32-NEXT:    sw a0, 96(sp)
+; RV64-i32-NEXT:    addi a0, sp, 64
+; RV64-i32-NEXT:    vsetivli zero, 16, e32, m4, ta, ma
+; RV64-i32-NEXT:    vle32.v v8, (a0)
+; RV64-i32-NEXT:    addi sp, s0, -192
+; RV64-i32-NEXT:    .cfi_def_cfa sp, 192
+; RV64-i32-NEXT:    ld ra, 184(sp) # 8-byte Folded Reload
+; RV64-i32-NEXT:    ld s0, 176(sp) # 8-byte Folded Reload
+; RV64-i32-NEXT:    .cfi_restore ra
+; RV64-i32-NEXT:    .cfi_restore s0
+; RV64-i32-NEXT:    addi sp, sp, 192
+; RV64-i32-NEXT:    .cfi_def_cfa_offset 0
+; RV64-i32-NEXT:    ret
+;
+; RV64-i64-LABEL: lround_v16f32:
+; RV64-i64:       # %bb.0:
+; RV64-i64-NEXT:    addi sp, sp, -384
+; RV64-i64-NEXT:    .cfi_def_cfa_offset 384
+; RV64-i64-NEXT:    sd ra, 376(sp) # 8-byte Folded Spill
+; RV64-i64-NEXT:    sd s0, 368(sp) # 8-byte Folded Spill
+; RV64-i64-NEXT:    .cfi_offset ra, -8
+; RV64-i64-NEXT:    .cfi_offset s0, -16
+; RV64-i64-NEXT:    addi s0, sp, 384
+; RV64-i64-NEXT:    .cfi_def_cfa s0, 0
+; RV64-i64-NEXT:    andi sp, sp, -128
+; RV64-i64-NEXT:    addi a0, sp, 64
+; RV64-i64-NEXT:    vsetivli zero, 16, e32, m4, ta, ma
+; RV64-i64-NEXT:    vse32.v v8, (a0)
+; RV64-i64-NEXT:    flw fa5, 124(sp)
+; RV64-i64-NEXT:    vfmv.f.s fa4, v8
+; RV64-i64-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
+; RV64-i64-NEXT:    vslidedown.vi v10, v8, 3
+; RV64-i64-NEXT:    vslidedown.vi v12, v8, 2
+; RV64-i64-NEXT:    fcvt.l.s a0, fa5, rmm
+; RV64-i64-NEXT:    sd a0, 248(sp)
+; RV64-i64-NEXT:    flw fa5, 120(sp)
+; RV64-i64-NEXT:    vslidedown.vi v13, v8, 1
+; RV64-i64-NEXT:    fcvt.l.s a0, fa4, rmm
+; RV64-i64-NEXT:    vfmv.f.s fa4, v10
+; RV64-i64-NEXT:    fcvt.l.s a1, fa5, rmm
+; RV64-i64-NEXT:    sd a1, 240(sp)
+; RV64-i64-NEXT:    flw fa5, 116(sp)
+; RV64-i64-NEXT:    vsetivli zero, 1, e32, m2, ta, ma
+; RV64-i64-NEXT:    vslidedown.vi v10, v8, 7
+; RV64-i64-NEXT:    fcvt.l.s a1, fa4, rmm
+; RV64-i64-NEXT:    vfmv.f.s fa4, v12
+; RV64-i64-NEXT:    fcvt.l.s a2, fa5, rmm
+; RV64-i64-NEXT:    sd a2, 232(sp)
+; RV64-i64-NEXT:    flw fa5, 112(sp)
+; RV64-i64-NEXT:    fcvt.l.s a2, fa4, rmm
+; RV64-i64-NEXT:    vfmv.f.s fa4, v13
+; RV64-i64-NEXT:    vslidedown.vi v12, v8, 6
+; RV64-i64-NEXT:    fcvt.l.s a3, fa5, rmm
+; RV64-i64-NEXT:    sd a3, 224(sp)
+; RV64-i64-NEXT:    flw fa5, 108(sp)
+; RV64-i64-NEXT:    fcvt.l.s a3, fa4, rmm
+; RV64-i64-NEXT:    vfmv.f.s fa4, v10
+; RV64-i64-NEXT:    vslidedown.vi v10, v8, 5
+; RV64-i64-NEXT:    fcvt.l.s a4, fa5, rmm
+; RV64-i64-NEXT:    sd a4, 216(sp)
+; RV64-i64-NEXT:    flw fa5, 104(sp)
+; RV64-i64-NEXT:    fcvt.l.s a4, fa4, rmm
+; RV64-i64-NEXT:    vfmv.f.s fa4, v12
+; RV64-i64-NEXT:    fcvt.l.s a5, fa4, rmm
+; RV64-i64-NEXT:    fcvt.l.s a6, fa5, rmm
+; RV64-i64-NEXT:    sd a6, 208(sp)
+; RV64-i64-NEXT:    flw fa5, 100(sp)
+; RV64-i64-NEXT:    vfmv.f.s fa4, v10
+; RV64-i64-NEXT:    fcvt.l.s a6, fa4, rmm
+; RV64-i64-NEXT:    vslidedown.vi v8, v8, 4
+; RV64-i64-NEXT:    fcvt.l.s a7, fa5, rmm
+; RV64-i64-NEXT:    vfmv.f.s fa5, v8
+; RV64-i64-NEXT:    sd a7, 200(sp)
+; RV64-i64-NEXT:    fcvt.l.s a7, fa5, rmm
+; RV64-i64-NEXT:    flw fa5, 96(sp)
+; RV64-i64-NEXT:    sd a0, 128(sp)
+; RV64-i64-NEXT:    sd a3, 136(sp)
+; RV64-i64-NEXT:    sd a2, 144(sp)
+; RV64-i64-NEXT:    sd a1, 152(sp)
+; RV64-i64-NEXT:    sd a7, 160(sp)
+; RV64-i64-NEXT:    sd a6, 168(sp)
+; RV64-i64-NEXT:    sd a5, 176(sp)
+; RV64-i64-NEXT:    sd a4, 184(sp)
+; RV64-i64-NEXT:    fcvt.l.s a0, fa5, rmm
+; RV64-i64-NEXT:    sd a0, 192(sp)
+; RV64-i64-NEXT:    addi a0, sp, 128
+; RV64-i64-NEXT:    vsetivli zero, 16, e64, m8, ta, ma
+; RV64-i64-NEXT:    vle64.v v8, (a0)
+; RV64-i64-NEXT:    addi sp, s0, -384
+; RV64-i64-NEXT:    .cfi_def_cfa sp, 384
+; RV64-i64-NEXT:    ld ra, 376(sp) # 8-byte Folded Reload
+; RV64-i64-NEXT:    ld s0, 368(sp) # 8-byte Folded Reload
+; RV64-i64-NEXT:    .cfi_restore ra
+; RV64-i64-NEXT:    .cfi_restore s0
+; RV64-i64-NEXT:    addi sp, sp, 384
+; RV64-i64-NEXT:    .cfi_def_cfa_offset 0
+; RV64-i64-NEXT:    ret
+  %a = call <16 x iXLen> @llvm.lround.v16iXLen.v16f32(<16 x float> %x)
+  ret <16 x iXLen> %a
+}
+declare <16 x iXLen> @llvm.lround.v16iXLen.v16f32(<16 x float>)
+
+define <1 x iXLen> @lround_v1f64(<1 x double> %x) {
+; RV32-LABEL: lround_v1f64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV32-NEXT:    vfmv.f.s fa5, v8
+; RV32-NEXT:    fcvt.w.d a0, fa5, rmm
+; RV32-NEXT:    vmv.s.x v8, a0
+; RV32-NEXT:    ret
+;
+; RV64-i32-LABEL: lround_v1f64:
+; RV64-i32:       # %bb.0:
+; RV64-i32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV64-i32-NEXT:    vfmv.f.s fa5, v8
+; RV64-i32-NEXT:    fcvt.w.d a0, fa5, rmm
+; RV64-i32-NEXT:    vmv.s.x v8, a0
+; RV64-i32-NEXT:    ret
+;
+; RV64-i64-LABEL: lround_v1f64:
+; RV64-i64:       # %bb.0:
+; RV64-i64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV64-i64-NEXT:    vfmv.f.s fa5, v8
+; RV64-i64-NEXT:    fcvt.l.d a0, fa5, rmm
+; RV64-i64-NEXT:    vmv.s.x v8, a0
+; RV64-i64-NEXT:    ret
+  %a = call <1 x iXLen> @llvm.lround.v1iXLen.v1f64(<1 x double> %x)
+  ret <1 x iXLen> %a
+}
+declare <1 x iXLen> @llvm.lround.v1iXLen.v1f64(<1 x double>)
+
+define <2 x iXLen> @lround_v2f64(<2 x double> %x) {
+; RV32-LABEL: lround_v2f64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV32-NEXT:    vslidedown.vi v9, v8, 1
+; RV32-NEXT:    vfmv.f.s fa5, v8
+; RV32-NEXT:    fcvt.w.d a0, fa5, rmm
+; RV32-NEXT:    vfmv.f.s fa5, v9
+; RV32-NEXT:    fcvt.w.d a1, fa5, rmm
+; RV32-NEXT:    vsetivli zero, 2, e32, mf2, ta, ma
+; RV32-NEXT:    vmv.v.x v8, a0
+; RV32-NEXT:    vslide1down.vx v8, v8, a1
+; RV32-NEXT:    ret
+;
+; RV64-i32-LABEL: lround_v2f64:
+; RV64-i32:       # %bb.0:
+; RV64-i32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV64-i32-NEXT:    vslidedown.vi v9, v8, 1
+; RV64-i32-NEXT:    vfmv.f.s fa5, v8
+; RV64-i32-NEXT:    fcvt.w.d a0, fa5, rmm
+; RV64-i32-NEXT:    vfmv.f.s fa5, v9
+; RV64-i32-NEXT:    fcvt.w.d a1, fa5, rmm
+; RV64-i32-NEXT:    vsetivli zero, 2, e32, mf2, ta, ma
+; RV64-i32-NEXT:    vmv.v.x v8, a0
+; RV64-i32-NEXT:    vslide1down.vx v8, v8, a1
+; RV64-i32-NEXT:    ret
+;
+; RV64-i64-LABEL: lround_v2f64:
+; RV64-i64:       # %bb.0:
+; RV64-i64-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
+; RV64-i64-NEXT:    vslidedown.vi v9, v8, 1
+; RV64-i64-NEXT:    vfmv.f.s fa5, v8
+; RV64-i64-NEXT:    fcvt.l.d a0, fa5, rmm
+; RV64-i64-NEXT:    vfmv.f.s fa5, v9
+; RV64-i64-NEXT:    fcvt.l.d a1, fa5, rmm
+; RV64-i64-NEXT:    vmv.v.x v8, a0
+; RV64-i64-NEXT:    vslide1down.vx v8, v8, a1
+; RV64-i64-NEXT:    ret
+  %a = call <2 x iXLen> @llvm.lround.v2iXLen.v2f64(<2 x double> %x)
+  ret <2 x iXLen> %a
+}
+declare <2 x iXLen> @llvm.lround.v2iXLen.v2f64(<2 x double>)
+
+define <4 x iXLen> @lround_v4f64(<4 x double> %x) {
+; RV32-LABEL: lround_v4f64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV32-NEXT:    vslidedown.vi v12, v8, 1
+; RV32-NEXT:    vfmv.f.s fa5, v8
+; RV32-NEXT:    vsetivli zero, 1, e64, m2, ta, ma
+; RV32-NEXT:    vslidedown.vi v10, v8, 2
+; RV32-NEXT:    vslidedown.vi v8, v8, 3
+; RV32-NEXT:    fcvt.w.d a0, fa5, rmm
+; RV32-NEXT:    vfmv.f.s fa5, v12
+; RV32-NEXT:    fcvt.w.d a1, fa5, rmm
+; RV32-NEXT:    vfmv.f.s fa5, v10
+; RV32-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
+; RV32-NEXT:    vmv.v.x v9, a0
+; RV32-NEXT:    fcvt.w.d a0, fa5, rmm
+; RV32-NEXT:    vsetvli zero, zero, e64, m2, ta, ma
+; RV32-NEXT:    vfmv.f.s fa5, v8
+; RV32-NEXT:    vsetvli zero, zero, e32, m1, ta, ma
+; RV32-NEXT:    vslide1down.vx v8, v9, a1
+; RV32-NEXT:    vslide1down.vx v8, v8, a0
+; RV32-NEXT:    fcvt.w.d a0, fa5, rmm
+; RV32-NEXT:    vslide1down.vx v8, v8, a0
+; RV32-NEXT:    ret
+;
+; RV64-i32-LABEL: lround_v4f64:
+; RV64-i32:       # %bb.0:
+; RV64-i32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV64-i32-NEXT:    vslidedown.vi v12, v8, 1
+; RV64-i32-NEXT:    vfmv.f.s fa5, v8
+; RV64-i32-NEXT:    vsetivli zero, 1, e64, m2, ta, ma
+; RV64-i32-NEXT:    vslidedown.vi v10, v8, 2
+; RV64-i32-NEXT:    vslidedown.vi v8, v8, 3
+; RV64-i32-NEXT:    fcvt.w.d a0, fa5, rmm
+; RV64-i32-NEXT:    vfmv.f.s fa5, v12
+; RV64-i32-NEXT:    fcvt.w.d a1, fa5, rmm
+; RV64-i32-NEXT:    vfmv.f.s fa5, v10
+; RV64-i32-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
+; RV64-i32-NEXT:    vmv.v.x v9, a0
+; RV64-i32-NEXT:    fcvt.w.d a0, fa5, rmm
+; RV64-i32-NEXT:    vsetvli zero, zero, e64, m2, ta, ma
+; RV64-i32-NEXT:    vfmv.f.s fa5, v8
+; RV64-i32-NEXT:    vsetvli zero, zero, e32, m1, ta, ma
+; RV64-i32-NEXT:    vslide1down.vx v8, v9, a1
+; RV64-i32-NEXT:    vslide1down.vx v8, v8, a0
+; RV64-i32-NEXT:    fcvt.w.d a0, fa5, rmm
+; RV64-i32-NEXT:    vslide1down.vx v8, v8, a0
+; RV64-i32-NEXT:    ret
+;
+; RV64-i64-LABEL: lround_v4f64:
+; RV64-i64:       # %bb.0:
+; RV64-i64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV64-i64-NEXT:    vslidedown.vi v12, v8, 1
+; RV64-i64-NEXT:    vfmv.f.s fa5, v8
+; RV64-i64-NEXT:    vsetivli zero, 1, e64, m2, ta, ma
+; RV64-i64-NEXT:    vslidedown.vi v10, v8, 2
+; RV64-i64-NEXT:    vslidedown.vi v8, v8, 3
+; RV64-i64-NEXT:    fcvt.l.d a0, fa5, rmm
+; RV64-i64-NEXT:    vfmv.f.s fa5, v12
+; RV64-i64-NEXT:    fcvt.l.d a1, fa5, rmm
+; RV64-i64-NEXT:    vfmv.f.s fa5, v10
+; RV64-i64-NEXT:    vsetivli zero, 4, e64, m2, ta, ma
+; RV64-i64-NEXT:    vmv.v.x v10, a0
+; RV64-i64-NEXT:    fcvt.l.d a0, fa5, rmm
+; RV64-i64-NEXT:    vfmv.f.s fa5, v8
+; RV64-i64-NEXT:    vslide1down.vx v8, v10, a1
+; RV64-i64-NEXT:    vslide1down.vx v8, v8, a0
+; RV64-i64-NEXT:    fcvt.l.d a0, fa5, rmm
+; RV64-i64-NEXT:    vslide1down.vx v8, v8, a0
+; RV64-i64-NEXT:    ret
+  %a = call <4 x iXLen> @llvm.lround.v4iXLen.v4f64(<4 x double> %x)
+  ret <4 x iXLen> %a
+}
+declare <4 x iXLen> @llvm.lround.v4iXLen.v4f64(<4 x double>)
+
+define <8 x iXLen> @lround_v8f64(<8 x double> %x) {
+; RV32-LABEL: lround_v8f64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -128
+; RV32-NEXT:    .cfi_def_cfa_offset 128
+; RV32-NEXT:    sw ra, 124(sp) # 4-byte Folded Spill
+; RV32-NEXT:    sw s0, 120(sp) # 4-byte Folded Spill
+; RV32-NEXT:    .cfi_offset ra, -4
+; RV32-NEXT:    .cfi_offset s0, -8
+; RV32-NEXT:    addi s0, sp, 128
+; RV32-NEXT:    .cfi_def_cfa s0, 0
+; RV32-NEXT:    andi sp, sp, -64
+; RV32-NEXT:    mv a0, sp
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV32-NEXT:    vslidedown.vi v14, v8, 1
+; RV32-NEXT:    vfmv.f.s fa5, v8
+; RV32-NEXT:    vsetivli zero, 1, e64, m2, ta, ma
+; RV32-NEXT:    vslidedown.vi v12, v8, 2
+; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, ma
+; RV32-NEXT:    vse64.v v8, (a0)
+; RV32-NEXT:    vsetivli zero, 1, e64, m2, ta, ma
+; RV32-NEXT:    vslidedown.vi v8, v8, 3
+; RV32-NEXT:    vfmv.f.s fa4, v14
+; RV32-NEXT:    fcvt.w.d a0, fa5, rmm
+; RV32-NEXT:    vfmv.f.s fa5, v12
+; RV32-NEXT:    vfmv.f.s fa3, v8
+; RV32-NEXT:    fcvt.w.d a1, fa4, rmm
+; RV32-NEXT:    fcvt.w.d a2, fa5, rmm
+; RV32-NEXT:    fcvt.w.d a3, fa3, rmm
+; RV32-NEXT:    fld fa5, 32(sp)
+; RV32-NEXT:    fld fa4, 40(sp)
+; RV32-NEXT:    fld fa3, 48(sp)
+; RV32-NEXT:    fld fa2, 56(sp)
+; RV32-NEXT:    fcvt.w.d a4, fa5, rmm
+; RV32-NEXT:    fcvt.w.d a5, fa4, rmm
+; RV32-NEXT:    fcvt.w.d a6, fa3, rmm
+; RV32-NEXT:    vsetivli zero, 8, e32, m2, ta, ma
+; RV32-NEXT:    vmv.v.x v8, a0
+; RV32-NEXT:    vslide1down.vx v8, v8, a1
+; RV32-NEXT:    vslide1down.vx v8, v8, a2
+; RV32-NEXT:    vslide1down.vx v8, v8, a3
+; RV32-NEXT:    vslide1down.vx v8, v8, a4
+; RV32-NEXT:    vslide1down.vx v8, v8, a5
+; RV32-NEXT:    vslide1down.vx v8, v8, a6
+; RV32-NEXT:    fcvt.w.d a0, fa2, rmm
+; RV32-NEXT:    vslide1down.vx v8, v8, a0
+; RV32-NEXT:    addi sp, s0, -128
+; RV32-NEXT:    .cfi_def_cfa sp, 128
+; RV32-NEXT:    lw ra, 124(sp) # 4-byte Folded Reload
+; RV32-NEXT:    lw s0, 120(sp) # 4-byte Folded Reload
+; RV32-NEXT:    .cfi_restore ra
+; RV32-NEXT:    .cfi_restore s0
+; RV32-NEXT:    addi sp, sp, 128
+; RV32-NEXT:    .cfi_def_cfa_offset 0
+; RV32-NEXT:    ret
+;
+; RV64-i32-LABEL: lround_v8f64:
+; RV64-i32:       # %bb.0:
+; RV64-i32-NEXT:    addi sp, sp, -128
+; RV64-i32-NEXT:    .cfi_def_cfa_offset 128
+; RV64-i32-NEXT:    sd ra, 120(sp) # 8-byte Folded Spill
+; RV64-i32-NEXT:    sd s0, 112(sp) # 8-byte Folded Spill
+; RV64-i32-NEXT:    .cfi_offset ra, -8
+; RV64-i32-NEXT:    .cfi_offset s0, -16
+; RV64-i32-NEXT:    addi s0, sp, 128
+; RV64-i32-NEXT:    .cfi_def_cfa s0, 0
+; RV64-i32-NEXT:    andi sp, sp, -64
+; RV64-i32-NEXT:    mv a0, sp
+; RV64-i32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV64-i32-NEXT:    vslidedown.vi v14, v8, 1
+; RV64-i32-NEXT:    vfmv.f.s fa5, v8
+; RV64-i32-NEXT:    vsetivli zero, 1, e64, m2, ta, ma
+; RV64-i32-NEXT:    vslidedown.vi v12, v8, 2
+; RV64-i32-NEXT:    vsetivli zero, 8, e64, m4, ta, ma
+; RV64-i32-NEXT:    vse64.v v8, (a0)
+; RV64-i32-NEXT:    vsetivli zero, 1, e64, m2, ta, ma
+; RV64-i32-NEXT:    vslidedown.vi v8, v8, 3
+; RV64-i32-NEXT:    vfmv.f.s fa4, v14
+; RV64-i32-NEXT:    fcvt.w.d a0, fa5, rmm
+; RV64-i32-NEXT:    vfmv.f.s fa5, v12
+; RV64-i32-NEXT:    vfmv.f.s fa3, v8
+; RV64-i32-NEXT:    fcvt.w.d a1, fa4, rmm
+; RV64-i32-NEXT:    fcvt.w.d a2, fa5, rmm
+; RV64-i32-NEXT:    fcvt.w.d a3, fa3, rmm
+; RV64-i32-NEXT:    fld fa5, 32(sp)
+; RV64-i32-NEXT:    fld fa4, 40(sp)
+; RV64-i32-NEXT:    fld fa3, 48(sp)
+; RV64-i32-NEXT:    fld fa2, 56(sp)
+; RV64-i32-NEXT:    fcvt.w.d a4, fa5, rmm
+; RV64-i32-NEXT:    fcvt.w.d a5, fa4, rmm
+; RV64-i32-NEXT:    fcvt.w.d a6, fa3, rmm
+; RV64-i32-NEXT:    vsetivli zero, 8, e32, m2, ta, ma
+; RV64-i32-NEXT:    vmv.v.x v8, a0
+; RV64-i32-NEXT:    vslide1down.vx v8, v8, a1
+; RV64-i32-NEXT:    vslide1down.vx v8, v8, a2
+; RV64-i32-NEXT:    vslide1down.vx v8, v8, a3
+; RV64-i32-NEXT:    vslide1down.vx v8, v8, a4
+; RV64-i32-NEXT:    vslide1down.vx v8, v8, a5
+; RV64-i32-NEXT:    vslide1down.vx v8, v8, a6
+; RV64-i32-NEXT:    fcvt.w.d a0, fa2, rmm
+; RV64-i32-NEXT:    vslide1down.vx v8, v8, a0
+; RV64-i32-NEXT:    addi sp, s0, -128
+; RV64-i32-NEXT:    .cfi_def_cfa sp, 128
+; RV64-i32-NEXT:    ld ra, 120(sp) # 8-byte Folded Reload
+; RV64-i32-NEXT:    ld s0, 112(sp) # 8-byte Folded Reload
+; RV64-i32-NEXT:    .cfi_restore ra
+; RV64-i32-NEXT:    .cfi_restore s0
+; RV64-i32-NEXT:    addi sp, sp, 128
+; RV64-i32-NEXT:    .cfi_def_cfa_offset 0
+; RV64-i32-NEXT:    ret
+;
+; RV64-i64-LABEL: lround_v8f64:
+; RV64-i64:       # %bb.0:
+; RV64-i64-NEXT:    addi sp, sp, -192
+; RV64-i64-NEXT:    .cfi_def_cfa_offset 192
+; RV64-i64-NEXT:    sd ra, 184(sp) # 8-byte Folded Spill
+; RV64-i64-NEXT:    sd s0, 176(sp) # 8-byte Folded Spill
+; RV64-i64-NEXT:    .cfi_offset ra, -8
+; RV64-i64-NEXT:    .cfi_offset s0, -16
+; RV64-i64-NEXT:    addi s0, sp, 192
+; RV64-i64-NEXT:    .cfi_def_cfa s0, 0
+; RV64-i64-NEXT:    andi sp, sp, -64
+; RV64-i64-NEXT:    mv a0, sp
+; RV64-i64-NEXT:    vsetivli zero, 8, e64, m4, ta, ma
+; RV64-i64-NEXT:    vse64.v v8, (a0)
+; RV64-i64-NEXT:    fld fa5, 56(sp)
+; RV64-i64-NEXT:    vfmv.f.s fa4, v8
+; RV64-i64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV64-i64-NEXT:    vslidedown.vi v10, v8, 1
+; RV64-i64-NEXT:    fcvt.l.d a0, fa4, rmm
+; RV64-i64-NEXT:    fcvt.l.d a1, fa5, rmm
+; RV64-i64-NEXT:    sd a1, 120(sp)
+; RV64-i64-NEXT:    fld fa5, 48(sp)
+; RV64-i64-NEXT:    vfmv.f.s fa4, v10
+; RV64-i64-NEXT:    vsetivli zero, 1, e64, m2, ta, ma
+; RV64-i64-NEXT:    vslidedown.vi v10, v8, 3
+; RV64-i64-NEXT:    fcvt.l.d a1, fa4, rmm
+; RV64-i64-NEXT:    fcvt.l.d a2, fa5, rmm
+; RV64-i64-NEXT:    sd a2, 112(sp)
+; RV64-i64-NEXT:    fld fa5, 40(sp)
+; RV64-i64-NEXT:    vfmv.f.s fa4, v10
+; RV64-i64-NEXT:    fcvt.l.d a2, fa4, rmm
+; RV64-i64-NEXT:    vslidedown.vi v8, v8, 2
+; RV64-i64-NEXT:    fcvt.l.d a3, fa5, rmm
+; RV64-i64-NEXT:    vfmv.f.s fa5, v8
+; RV64-i64-NEXT:    sd a3, 104(sp)
+; RV64-i64-NEXT:    fcvt.l.d a3, fa5, rmm
+; RV64-i64-NEXT:    fld fa5, 32(sp)
+; RV64-i64-NEXT:    sd a0, 64(sp)
+; RV64-i64-NEXT:    sd a1, 72(sp)
+; RV64-i64-NEXT:    sd a3, 80(sp)
+; RV64-i64-NEXT:    sd a2, 88(sp)
+; RV64-i64-NEXT:    fcvt.l.d a0, fa5, rmm
+; RV64-i64-NEXT:    sd a0, 96(sp)
+; RV64-i64-NEXT:    addi a0, sp, 64
+; RV64-i64-NEXT:    vsetivli zero, 8, e64, m4, ta, ma
+; RV64-i64-NEXT:    vle64.v v8, (a0)
+; RV64-i64-NEXT:    addi sp, s0, -192
+; RV64-i64-NEXT:    .cfi_def_cfa sp, 192
+; RV64-i64-NEXT:    ld ra, 184(sp) # 8-byte Folded Reload
+; RV64-i64-NEXT:    ld s0, 176(sp) # 8-byte Folded Reload
+; RV64-i64-NEXT:    .cfi_restore ra
+; RV64-i64-NEXT:    .cfi_restore s0
+; RV64-i64-NEXT:    addi sp, sp, 192
+; RV64-i64-NEXT:    .cfi_def_cfa_offset 0
+; RV64-i64-NEXT:    ret
+  %a = call <8 x iXLen> @llvm.lround.v8iXLen.v8f64(<8 x double> %x)
+  ret <8 x iXLen> %a
+}
+declare <8 x iXLen> @llvm.lround.v8iXLen.v8f64(<8 x double>)