[llvm] [TTI] Handle experimental.vp.reverse in BasicTTIImpl (PR #147868)

[Luke Lau via llvm-commits]

https://github.com/lukel97 updated https://github.com/llvm/llvm-project/pull/147868

>From 5503a8a929da56ef0944304754155150e5f33261 Mon Sep 17 00:00:00 2001
From: Luke Lau <luke at igalia.com>
Date: Wed, 9 Jul 2025 16:49:17 +0800
Subject: [PATCH 1/2] Precommit tests

---
 .../Analysis/CostModel/RISCV/vp-intrinsics.ll | 228 ++++++++++++++++++
 1 file changed, 228 insertions(+)

diff --git a/llvm/test/Analysis/CostModel/RISCV/vp-intrinsics.ll b/llvm/test/Analysis/CostModel/RISCV/vp-intrinsics.ll
index 3a2e7d5580ac0..dda00d34ecf10 100644
--- a/llvm/test/Analysis/CostModel/RISCV/vp-intrinsics.ll
+++ b/llvm/test/Analysis/CostModel/RISCV/vp-intrinsics.ll
@@ -1752,5 +1752,233 @@ define void @is.fpclass() {
   call <vscale x 4 x i1> @llvm.vp.is.fpclass.nxv4f64(<vscale x 4 x double> poison, i32 0, <vscale x 4 x i1> poison, i32 poison)
   call <vscale x 8 x i1> @llvm.vp.is.fpclass.nxv8f64(<vscale x 8 x double> poison, i32 0, <vscale x 8 x i1> poison, i32 poison)
   call <vscale x 16 x i1> @llvm.vp.is.fpclass.nxv16f64(<vscale x 16 x double> poison, i32 0, <vscale x 16 x i1> poison, i32 poison)
+
+  ret void
+}
+
+define void @reverse() {
+; ARGBASED-LABEL: 'reverse'
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %1 = call <2 x i1> @llvm.experimental.vp.reverse.v2i1(<2 x i1> poison, <2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %2 = call <4 x i1> @llvm.experimental.vp.reverse.v4i1(<4 x i1> poison, <4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %3 = call <8 x i1> @llvm.experimental.vp.reverse.v8i1(<8 x i1> poison, <8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %4 = call <16 x i1> @llvm.experimental.vp.reverse.v16i1(<16 x i1> poison, <16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %5 = call <2 x i8> @llvm.experimental.vp.reverse.v2i8(<2 x i8> poison, <2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %6 = call <4 x i8> @llvm.experimental.vp.reverse.v4i8(<4 x i8> poison, <4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %7 = call <8 x i8> @llvm.experimental.vp.reverse.v8i8(<8 x i8> poison, <8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %8 = call <16 x i8> @llvm.experimental.vp.reverse.v16i8(<16 x i8> poison, <16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %9 = call <2 x i16> @llvm.experimental.vp.reverse.v2i16(<2 x i16> poison, <2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %10 = call <4 x i16> @llvm.experimental.vp.reverse.v4i16(<4 x i16> poison, <4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %11 = call <8 x i16> @llvm.experimental.vp.reverse.v8i16(<8 x i16> poison, <8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %12 = call <16 x i16> @llvm.experimental.vp.reverse.v16i16(<16 x i16> poison, <16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %13 = call <2 x i32> @llvm.experimental.vp.reverse.v2i32(<2 x i32> poison, <2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %14 = call <4 x i32> @llvm.experimental.vp.reverse.v4i32(<4 x i32> poison, <4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %15 = call <8 x i32> @llvm.experimental.vp.reverse.v8i32(<8 x i32> poison, <8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %16 = call <16 x i32> @llvm.experimental.vp.reverse.v16i32(<16 x i32> poison, <16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %17 = call <2 x i64> @llvm.experimental.vp.reverse.v2i64(<2 x i64> poison, <2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %18 = call <4 x i64> @llvm.experimental.vp.reverse.v4i64(<4 x i64> poison, <4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %19 = call <8 x i64> @llvm.experimental.vp.reverse.v8i64(<8 x i64> poison, <8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %20 = call <16 x i64> @llvm.experimental.vp.reverse.v16i64(<16 x i64> poison, <16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %21 = call <2 x bfloat> @llvm.experimental.vp.reverse.v2bf16(<2 x bfloat> poison, <2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %22 = call <4 x bfloat> @llvm.experimental.vp.reverse.v4bf16(<4 x bfloat> poison, <4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %23 = call <8 x bfloat> @llvm.experimental.vp.reverse.v8bf16(<8 x bfloat> poison, <8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %24 = call <16 x bfloat> @llvm.experimental.vp.reverse.v16bf16(<16 x bfloat> poison, <16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %25 = call <2 x half> @llvm.experimental.vp.reverse.v2f16(<2 x half> poison, <2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %26 = call <4 x half> @llvm.experimental.vp.reverse.v4f16(<4 x half> poison, <4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %27 = call <8 x half> @llvm.experimental.vp.reverse.v8f16(<8 x half> poison, <8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %28 = call <16 x half> @llvm.experimental.vp.reverse.v16f16(<16 x half> poison, <16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %29 = call <2 x float> @llvm.experimental.vp.reverse.v2f32(<2 x float> poison, <2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %30 = call <4 x float> @llvm.experimental.vp.reverse.v4f32(<4 x float> poison, <4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %31 = call <8 x float> @llvm.experimental.vp.reverse.v8f32(<8 x float> poison, <8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %32 = call <16 x float> @llvm.experimental.vp.reverse.v16f32(<16 x float> poison, <16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %33 = call <2 x double> @llvm.experimental.vp.reverse.v2f64(<2 x double> poison, <2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %34 = call <4 x double> @llvm.experimental.vp.reverse.v4f64(<4 x double> poison, <4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %35 = call <8 x double> @llvm.experimental.vp.reverse.v8f64(<8 x double> poison, <8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %36 = call <16 x double> @llvm.experimental.vp.reverse.v16f64(<16 x double> poison, <16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %37 = call <vscale x 2 x i1> @llvm.experimental.vp.reverse.nxv2i1(<vscale x 2 x i1> poison, <vscale x 2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %38 = call <vscale x 4 x i1> @llvm.experimental.vp.reverse.nxv4i1(<vscale x 4 x i1> poison, <vscale x 4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %39 = call <vscale x 8 x i1> @llvm.experimental.vp.reverse.nxv8i1(<vscale x 8 x i1> poison, <vscale x 8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %40 = call <vscale x 16 x i1> @llvm.experimental.vp.reverse.nxv16i1(<vscale x 16 x i1> poison, <vscale x 16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %41 = call <vscale x 2 x i8> @llvm.experimental.vp.reverse.nxv2i8(<vscale x 2 x i8> poison, <vscale x 2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %42 = call <vscale x 4 x i8> @llvm.experimental.vp.reverse.nxv4i8(<vscale x 4 x i8> poison, <vscale x 4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %43 = call <vscale x 8 x i8> @llvm.experimental.vp.reverse.nxv8i8(<vscale x 8 x i8> poison, <vscale x 8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %44 = call <vscale x 16 x i8> @llvm.experimental.vp.reverse.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %45 = call <vscale x 2 x i16> @llvm.experimental.vp.reverse.nxv2i16(<vscale x 2 x i16> poison, <vscale x 2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %46 = call <vscale x 4 x i16> @llvm.experimental.vp.reverse.nxv4i16(<vscale x 4 x i16> poison, <vscale x 4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %47 = call <vscale x 8 x i16> @llvm.experimental.vp.reverse.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %48 = call <vscale x 16 x i16> @llvm.experimental.vp.reverse.nxv16i16(<vscale x 16 x i16> poison, <vscale x 16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %49 = call <vscale x 2 x i32> @llvm.experimental.vp.reverse.nxv2i32(<vscale x 2 x i32> poison, <vscale x 2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %50 = call <vscale x 4 x i32> @llvm.experimental.vp.reverse.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %51 = call <vscale x 8 x i32> @llvm.experimental.vp.reverse.nxv8i32(<vscale x 8 x i32> poison, <vscale x 8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %52 = call <vscale x 16 x i32> @llvm.experimental.vp.reverse.nxv16i32(<vscale x 16 x i32> poison, <vscale x 16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %53 = call <vscale x 2 x i64> @llvm.experimental.vp.reverse.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %54 = call <vscale x 4 x i64> @llvm.experimental.vp.reverse.nxv4i64(<vscale x 4 x i64> poison, <vscale x 4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %55 = call <vscale x 8 x i64> @llvm.experimental.vp.reverse.nxv8i64(<vscale x 8 x i64> poison, <vscale x 8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %56 = call <vscale x 16 x i64> @llvm.experimental.vp.reverse.nxv16i64(<vscale x 16 x i64> poison, <vscale x 16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %57 = call <vscale x 2 x bfloat> @llvm.experimental.vp.reverse.nxv2bf16(<vscale x 2 x bfloat> poison, <vscale x 2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %58 = call <vscale x 4 x bfloat> @llvm.experimental.vp.reverse.nxv4bf16(<vscale x 4 x bfloat> poison, <vscale x 4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %59 = call <vscale x 8 x bfloat> @llvm.experimental.vp.reverse.nxv8bf16(<vscale x 8 x bfloat> poison, <vscale x 8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %60 = call <vscale x 16 x bfloat> @llvm.experimental.vp.reverse.nxv16bf16(<vscale x 16 x bfloat> poison, <vscale x 16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %61 = call <vscale x 2 x half> @llvm.experimental.vp.reverse.nxv2f16(<vscale x 2 x half> poison, <vscale x 2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %62 = call <vscale x 4 x half> @llvm.experimental.vp.reverse.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %63 = call <vscale x 8 x half> @llvm.experimental.vp.reverse.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %64 = call <vscale x 16 x half> @llvm.experimental.vp.reverse.nxv16f16(<vscale x 16 x half> poison, <vscale x 16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %65 = call <vscale x 2 x float> @llvm.experimental.vp.reverse.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %66 = call <vscale x 4 x float> @llvm.experimental.vp.reverse.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %67 = call <vscale x 8 x float> @llvm.experimental.vp.reverse.nxv8f32(<vscale x 8 x float> poison, <vscale x 8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %68 = call <vscale x 16 x float> @llvm.experimental.vp.reverse.nxv16f32(<vscale x 16 x float> poison, <vscale x 16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %69 = call <vscale x 2 x double> @llvm.experimental.vp.reverse.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %70 = call <vscale x 4 x double> @llvm.experimental.vp.reverse.nxv4f64(<vscale x 4 x double> poison, <vscale x 4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %71 = call <vscale x 8 x double> @llvm.experimental.vp.reverse.nxv8f64(<vscale x 8 x double> poison, <vscale x 8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %72 = call <vscale x 16 x double> @llvm.experimental.vp.reverse.nxv16f64(<vscale x 16 x double> poison, <vscale x 16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
+; TYPEBASED-LABEL: 'reverse'
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %1 = call <2 x i1> @llvm.experimental.vp.reverse.v2i1(<2 x i1> poison, <2 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %2 = call <4 x i1> @llvm.experimental.vp.reverse.v4i1(<4 x i1> poison, <4 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %3 = call <8 x i1> @llvm.experimental.vp.reverse.v8i1(<8 x i1> poison, <8 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %4 = call <16 x i1> @llvm.experimental.vp.reverse.v16i1(<16 x i1> poison, <16 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %5 = call <2 x i8> @llvm.experimental.vp.reverse.v2i8(<2 x i8> poison, <2 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %6 = call <4 x i8> @llvm.experimental.vp.reverse.v4i8(<4 x i8> poison, <4 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: %7 = call <8 x i8> @llvm.experimental.vp.reverse.v8i8(<8 x i8> poison, <8 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 126 for instruction: %8 = call <16 x i8> @llvm.experimental.vp.reverse.v16i8(<16 x i8> poison, <16 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %9 = call <2 x i16> @llvm.experimental.vp.reverse.v2i16(<2 x i16> poison, <2 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %10 = call <4 x i16> @llvm.experimental.vp.reverse.v4i16(<4 x i16> poison, <4 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: %11 = call <8 x i16> @llvm.experimental.vp.reverse.v8i16(<8 x i16> poison, <8 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 141 for instruction: %12 = call <16 x i16> @llvm.experimental.vp.reverse.v16i16(<16 x i16> poison, <16 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %13 = call <2 x i32> @llvm.experimental.vp.reverse.v2i32(<2 x i32> poison, <2 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %14 = call <4 x i32> @llvm.experimental.vp.reverse.v4i32(<4 x i32> poison, <4 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 69 for instruction: %15 = call <8 x i32> @llvm.experimental.vp.reverse.v8i32(<8 x i32> poison, <8 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 141 for instruction: %16 = call <16 x i32> @llvm.experimental.vp.reverse.v16i32(<16 x i32> poison, <16 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %17 = call <2 x i64> @llvm.experimental.vp.reverse.v2i64(<2 x i64> poison, <2 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %18 = call <4 x i64> @llvm.experimental.vp.reverse.v4i64(<4 x i64> poison, <4 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 69 for instruction: %19 = call <8 x i64> @llvm.experimental.vp.reverse.v8i64(<8 x i64> poison, <8 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 141 for instruction: %20 = call <16 x i64> @llvm.experimental.vp.reverse.v16i64(<16 x i64> poison, <16 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %21 = call <2 x bfloat> @llvm.experimental.vp.reverse.v2bf16(<2 x bfloat> poison, <2 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %22 = call <4 x bfloat> @llvm.experimental.vp.reverse.v4bf16(<4 x bfloat> poison, <4 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: %23 = call <8 x bfloat> @llvm.experimental.vp.reverse.v8bf16(<8 x bfloat> poison, <8 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 141 for instruction: %24 = call <16 x bfloat> @llvm.experimental.vp.reverse.v16bf16(<16 x bfloat> poison, <16 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %25 = call <2 x half> @llvm.experimental.vp.reverse.v2f16(<2 x half> poison, <2 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %26 = call <4 x half> @llvm.experimental.vp.reverse.v4f16(<4 x half> poison, <4 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: %27 = call <8 x half> @llvm.experimental.vp.reverse.v8f16(<8 x half> poison, <8 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 141 for instruction: %28 = call <16 x half> @llvm.experimental.vp.reverse.v16f16(<16 x half> poison, <16 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %29 = call <2 x float> @llvm.experimental.vp.reverse.v2f32(<2 x float> poison, <2 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %30 = call <4 x float> @llvm.experimental.vp.reverse.v4f32(<4 x float> poison, <4 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 69 for instruction: %31 = call <8 x float> @llvm.experimental.vp.reverse.v8f32(<8 x float> poison, <8 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 141 for instruction: %32 = call <16 x float> @llvm.experimental.vp.reverse.v16f32(<16 x float> poison, <16 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %33 = call <2 x double> @llvm.experimental.vp.reverse.v2f64(<2 x double> poison, <2 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %34 = call <4 x double> @llvm.experimental.vp.reverse.v4f64(<4 x double> poison, <4 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 69 for instruction: %35 = call <8 x double> @llvm.experimental.vp.reverse.v8f64(<8 x double> poison, <8 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 141 for instruction: %36 = call <16 x double> @llvm.experimental.vp.reverse.v16f64(<16 x double> poison, <16 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %37 = call <vscale x 2 x i1> @llvm.experimental.vp.reverse.nxv2i1(<vscale x 2 x i1> poison, <vscale x 2 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %38 = call <vscale x 4 x i1> @llvm.experimental.vp.reverse.nxv4i1(<vscale x 4 x i1> poison, <vscale x 4 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %39 = call <vscale x 8 x i1> @llvm.experimental.vp.reverse.nxv8i1(<vscale x 8 x i1> poison, <vscale x 8 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %40 = call <vscale x 16 x i1> @llvm.experimental.vp.reverse.nxv16i1(<vscale x 16 x i1> poison, <vscale x 16 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %41 = call <vscale x 2 x i8> @llvm.experimental.vp.reverse.nxv2i8(<vscale x 2 x i8> poison, <vscale x 2 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %42 = call <vscale x 4 x i8> @llvm.experimental.vp.reverse.nxv4i8(<vscale x 4 x i8> poison, <vscale x 4 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %43 = call <vscale x 8 x i8> @llvm.experimental.vp.reverse.nxv8i8(<vscale x 8 x i8> poison, <vscale x 8 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %44 = call <vscale x 16 x i8> @llvm.experimental.vp.reverse.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %45 = call <vscale x 2 x i16> @llvm.experimental.vp.reverse.nxv2i16(<vscale x 2 x i16> poison, <vscale x 2 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %46 = call <vscale x 4 x i16> @llvm.experimental.vp.reverse.nxv4i16(<vscale x 4 x i16> poison, <vscale x 4 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %47 = call <vscale x 8 x i16> @llvm.experimental.vp.reverse.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %48 = call <vscale x 16 x i16> @llvm.experimental.vp.reverse.nxv16i16(<vscale x 16 x i16> poison, <vscale x 16 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %49 = call <vscale x 2 x i32> @llvm.experimental.vp.reverse.nxv2i32(<vscale x 2 x i32> poison, <vscale x 2 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %50 = call <vscale x 4 x i32> @llvm.experimental.vp.reverse.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %51 = call <vscale x 8 x i32> @llvm.experimental.vp.reverse.nxv8i32(<vscale x 8 x i32> poison, <vscale x 8 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %52 = call <vscale x 16 x i32> @llvm.experimental.vp.reverse.nxv16i32(<vscale x 16 x i32> poison, <vscale x 16 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %53 = call <vscale x 2 x i64> @llvm.experimental.vp.reverse.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %54 = call <vscale x 4 x i64> @llvm.experimental.vp.reverse.nxv4i64(<vscale x 4 x i64> poison, <vscale x 4 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %55 = call <vscale x 8 x i64> @llvm.experimental.vp.reverse.nxv8i64(<vscale x 8 x i64> poison, <vscale x 8 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %56 = call <vscale x 16 x i64> @llvm.experimental.vp.reverse.nxv16i64(<vscale x 16 x i64> poison, <vscale x 16 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %57 = call <vscale x 2 x bfloat> @llvm.experimental.vp.reverse.nxv2bf16(<vscale x 2 x bfloat> poison, <vscale x 2 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %58 = call <vscale x 4 x bfloat> @llvm.experimental.vp.reverse.nxv4bf16(<vscale x 4 x bfloat> poison, <vscale x 4 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %59 = call <vscale x 8 x bfloat> @llvm.experimental.vp.reverse.nxv8bf16(<vscale x 8 x bfloat> poison, <vscale x 8 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %60 = call <vscale x 16 x bfloat> @llvm.experimental.vp.reverse.nxv16bf16(<vscale x 16 x bfloat> poison, <vscale x 16 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %61 = call <vscale x 2 x half> @llvm.experimental.vp.reverse.nxv2f16(<vscale x 2 x half> poison, <vscale x 2 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %62 = call <vscale x 4 x half> @llvm.experimental.vp.reverse.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %63 = call <vscale x 8 x half> @llvm.experimental.vp.reverse.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %64 = call <vscale x 16 x half> @llvm.experimental.vp.reverse.nxv16f16(<vscale x 16 x half> poison, <vscale x 16 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %65 = call <vscale x 2 x float> @llvm.experimental.vp.reverse.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %66 = call <vscale x 4 x float> @llvm.experimental.vp.reverse.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %67 = call <vscale x 8 x float> @llvm.experimental.vp.reverse.nxv8f32(<vscale x 8 x float> poison, <vscale x 8 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %68 = call <vscale x 16 x float> @llvm.experimental.vp.reverse.nxv16f32(<vscale x 16 x float> poison, <vscale x 16 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %69 = call <vscale x 2 x double> @llvm.experimental.vp.reverse.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %70 = call <vscale x 4 x double> @llvm.experimental.vp.reverse.nxv4f64(<vscale x 4 x double> poison, <vscale x 4 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %71 = call <vscale x 8 x double> @llvm.experimental.vp.reverse.nxv8f64(<vscale x 8 x double> poison, <vscale x 8 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %72 = call <vscale x 16 x double> @llvm.experimental.vp.reverse.nxv16f64(<vscale x 16 x double> poison, <vscale x 16 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
+  call <2 x i1> @llvm.experimental.vp.reverse.v2i1(<2 x i1> poison, <2 x i1> poison, i32 poison)
+  call <4 x i1> @llvm.experimental.vp.reverse.v4i1(<4 x i1> poison, <4 x i1> poison, i32 poison)
+  call <8 x i1> @llvm.experimental.vp.reverse.v8i1(<8 x i1> poison, <8 x i1> poison, i32 poison)
+  call <16 x i1> @llvm.experimental.vp.reverse.v16i1(<16 x i1> poison, <16 x i1> poison, i32 poison)
+  call <2 x i8> @llvm.experimental.vp.reverse.v2i8(<2 x i8> poison, <2 x i1> poison, i32 poison)
+  call <4 x i8> @llvm.experimental.vp.reverse.v4i8(<4 x i8> poison, <4 x i1> poison, i32 poison)
+  call <8 x i8> @llvm.experimental.vp.reverse.v8i8(<8 x i8> poison, <8 x i1> poison, i32 poison)
+  call <16 x i8> @llvm.experimental.vp.reverse.v16i8(<16 x i8> poison, <16 x i1> poison, i32 poison)
+  call <2 x i16> @llvm.experimental.vp.reverse.v2i16(<2 x i16> poison, <2 x i1> poison, i32 poison)
+  call <4 x i16> @llvm.experimental.vp.reverse.v4i16(<4 x i16> poison, <4 x i1> poison, i32 poison)
+  call <8 x i16> @llvm.experimental.vp.reverse.v8i16(<8 x i16> poison, <8 x i1> poison, i32 poison)
+  call <16 x i16> @llvm.experimental.vp.reverse.v16i16(<16 x i16> poison, <16 x i1> poison, i32 poison)
+  call <2 x i32> @llvm.experimental.vp.reverse.v2i32(<2 x i32> poison, <2 x i1> poison, i32 poison)
+  call <4 x i32> @llvm.experimental.vp.reverse.v4i32(<4 x i32> poison, <4 x i1> poison, i32 poison)
+  call <8 x i32> @llvm.experimental.vp.reverse.v8i32(<8 x i32> poison, <8 x i1> poison, i32 poison)
+  call <16 x i32> @llvm.experimental.vp.reverse.v16i32(<16 x i32> poison, <16 x i1> poison, i32 poison)
+  call <2 x i64> @llvm.experimental.vp.reverse.v2i64(<2 x i64> poison, <2 x i1> poison, i32 poison)
+  call <4 x i64> @llvm.experimental.vp.reverse.v4i64(<4 x i64> poison, <4 x i1> poison, i32 poison)
+  call <8 x i64> @llvm.experimental.vp.reverse.v8i64(<8 x i64> poison, <8 x i1> poison, i32 poison)
+  call <16 x i64> @llvm.experimental.vp.reverse.v16i64(<16 x i64> poison, <16 x i1> poison, i32 poison)
+  call <2 x bfloat> @llvm.experimental.vp.reverse.v2bf16(<2 x bfloat> poison, <2 x i1> poison, i32 poison)
+  call <4 x bfloat> @llvm.experimental.vp.reverse.v4bf16(<4 x bfloat> poison, <4 x i1> poison, i32 poison)
+  call <8 x bfloat> @llvm.experimental.vp.reverse.v8bf16(<8 x bfloat> poison, <8 x i1> poison, i32 poison)
+  call <16 x bfloat> @llvm.experimental.vp.reverse.v16bf16(<16 x bfloat> poison, <16 x i1> poison, i32 poison)
+  call <2 x half> @llvm.experimental.vp.reverse.v2f16(<2 x half> poison, <2 x i1> poison, i32 poison)
+  call <4 x half> @llvm.experimental.vp.reverse.v4f16(<4 x half> poison, <4 x i1> poison, i32 poison)
+  call <8 x half> @llvm.experimental.vp.reverse.v8f16(<8 x half> poison, <8 x i1> poison, i32 poison)
+  call <16 x half> @llvm.experimental.vp.reverse.v16f16(<16 x half> poison, <16 x i1> poison, i32 poison)
+  call <2 x float> @llvm.experimental.vp.reverse.v2f32(<2 x float> poison, <2 x i1> poison, i32 poison)
+  call <4 x float> @llvm.experimental.vp.reverse.v4f32(<4 x float> poison, <4 x i1> poison, i32 poison)
+  call <8 x float> @llvm.experimental.vp.reverse.v8f32(<8 x float> poison, <8 x i1> poison, i32 poison)
+  call <16 x float> @llvm.experimental.vp.reverse.v16f32(<16 x float> poison, <16 x i1> poison, i32 poison)
+  call <2 x double> @llvm.experimental.vp.reverse.v2f64(<2 x double> poison, <2 x i1> poison, i32 poison)
+  call <4 x double> @llvm.experimental.vp.reverse.v4f64(<4 x double> poison, <4 x i1> poison, i32 poison)
+  call <8 x double> @llvm.experimental.vp.reverse.v8f64(<8 x double> poison, <8 x i1> poison, i32 poison)
+  call <16 x double> @llvm.experimental.vp.reverse.v16f64(<16 x double> poison, <16 x i1> poison, i32 poison)
+  call <vscale x 2 x i1> @llvm.experimental.vp.reverse.nxv2i1(<vscale x 2 x i1> poison, <vscale x 2 x i1> poison, i32 poison)
+  call <vscale x 4 x i1> @llvm.experimental.vp.reverse.nxv4i1(<vscale x 4 x i1> poison, <vscale x 4 x i1> poison, i32 poison)
+  call <vscale x 8 x i1> @llvm.experimental.vp.reverse.nxv8i1(<vscale x 8 x i1> poison, <vscale x 8 x i1> poison, i32 poison)
+  call <vscale x 16 x i1> @llvm.experimental.vp.reverse.nxv16i1(<vscale x 16 x i1> poison, <vscale x 16 x i1> poison, i32 poison)
+  call <vscale x 2 x i8> @llvm.experimental.vp.reverse.nxv2i8(<vscale x 2 x i8> poison, <vscale x 2 x i1> poison, i32 poison)
+  call <vscale x 4 x i8> @llvm.experimental.vp.reverse.nxv4i8(<vscale x 4 x i8> poison, <vscale x 4 x i1> poison, i32 poison)
+  call <vscale x 8 x i8> @llvm.experimental.vp.reverse.nxv8i8(<vscale x 8 x i8> poison, <vscale x 8 x i1> poison, i32 poison)
+  call <vscale x 16 x i8> @llvm.experimental.vp.reverse.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> poison, i32 poison)
+  call <vscale x 2 x i16> @llvm.experimental.vp.reverse.nxv2i16(<vscale x 2 x i16> poison, <vscale x 2 x i1> poison, i32 poison)
+  call <vscale x 4 x i16> @llvm.experimental.vp.reverse.nxv4i16(<vscale x 4 x i16> poison, <vscale x 4 x i1> poison, i32 poison)
+  call <vscale x 8 x i16> @llvm.experimental.vp.reverse.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> poison, i32 poison)
+  call <vscale x 16 x i16> @llvm.experimental.vp.reverse.nxv16i16(<vscale x 16 x i16> poison, <vscale x 16 x i1> poison, i32 poison)
+  call <vscale x 2 x i32> @llvm.experimental.vp.reverse.nxv2i32(<vscale x 2 x i32> poison, <vscale x 2 x i1> poison, i32 poison)
+  call <vscale x 4 x i32> @llvm.experimental.vp.reverse.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> poison, i32 poison)
+  call <vscale x 8 x i32> @llvm.experimental.vp.reverse.nxv8i32(<vscale x 8 x i32> poison, <vscale x 8 x i1> poison, i32 poison)
+  call <vscale x 16 x i32> @llvm.experimental.vp.reverse.nxv16i32(<vscale x 16 x i32> poison, <vscale x 16 x i1> poison, i32 poison)
+  call <vscale x 2 x i64> @llvm.experimental.vp.reverse.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> poison, i32 poison)
+  call <vscale x 4 x i64> @llvm.experimental.vp.reverse.nxv4i64(<vscale x 4 x i64> poison, <vscale x 4 x i1> poison, i32 poison)
+  call <vscale x 8 x i64> @llvm.experimental.vp.reverse.nxv8i64(<vscale x 8 x i64> poison, <vscale x 8 x i1> poison, i32 poison)
+  call <vscale x 16 x i64> @llvm.experimental.vp.reverse.nxv16i64(<vscale x 16 x i64> poison, <vscale x 16 x i1> poison, i32 poison)
+  call <vscale x 2 x bfloat> @llvm.experimental.vp.reverse.nxv2bf16(<vscale x 2 x bfloat> poison, <vscale x 2 x i1> poison, i32 poison)
+  call <vscale x 4 x bfloat> @llvm.experimental.vp.reverse.nxv4bf16(<vscale x 4 x bfloat> poison, <vscale x 4 x i1> poison, i32 poison)
+  call <vscale x 8 x bfloat> @llvm.experimental.vp.reverse.nxv8bf16(<vscale x 8 x bfloat> poison, <vscale x 8 x i1> poison, i32 poison)
+  call <vscale x 16 x bfloat> @llvm.experimental.vp.reverse.nxv16bf16(<vscale x 16 x bfloat> poison, <vscale x 16 x i1> poison, i32 poison)
+  call <vscale x 2 x half> @llvm.experimental.vp.reverse.nxv2f16(<vscale x 2 x half> poison, <vscale x 2 x i1> poison, i32 poison)
+  call <vscale x 4 x half> @llvm.experimental.vp.reverse.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> poison, i32 poison)
+  call <vscale x 8 x half> @llvm.experimental.vp.reverse.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> poison, i32 poison)
+  call <vscale x 16 x half> @llvm.experimental.vp.reverse.nxv16f16(<vscale x 16 x half> poison, <vscale x 16 x i1> poison, i32 poison)
+  call <vscale x 2 x float> @llvm.experimental.vp.reverse.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> poison, i32 poison)
+  call <vscale x 4 x float> @llvm.experimental.vp.reverse.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> poison, i32 poison)
+  call <vscale x 8 x float> @llvm.experimental.vp.reverse.nxv8f32(<vscale x 8 x float> poison, <vscale x 8 x i1> poison, i32 poison)
+  call <vscale x 16 x float> @llvm.experimental.vp.reverse.nxv16f32(<vscale x 16 x float> poison, <vscale x 16 x i1> poison, i32 poison)
+  call <vscale x 2 x double> @llvm.experimental.vp.reverse.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> poison, i32 poison)
+  call <vscale x 4 x double> @llvm.experimental.vp.reverse.nxv4f64(<vscale x 4 x double> poison, <vscale x 4 x i1> poison, i32 poison)
+  call <vscale x 8 x double> @llvm.experimental.vp.reverse.nxv8f64(<vscale x 8 x double> poison, <vscale x 8 x i1> poison, i32 poison)
+  call <vscale x 16 x double> @llvm.experimental.vp.reverse.nxv16f64(<vscale x 16 x double> poison, <vscale x 16 x i1> poison, i32 poison)
+
   ret void
 }

>From ed1fbd81c32eb068a01e9191bdfdde621273f677 Mon Sep 17 00:00:00 2001
From: Luke Lau <luke at igalia.com>
Date: Wed, 9 Jul 2025 17:15:37 +0800
Subject: [PATCH 2/2] [TTI] Handle experimental.vp.reverse in BasicTTIImpl

A experimental_vp_reverse isn't exactly functionally the same as vector_reverse, so previously it wasn't getting picked up by the generic VP costing code that reuses the non-VP equivalents.

But for costing purposes it's good enough so we can reuse it.

The RISC-V costs are still incorrect and are showing up as scalarized. #147677 aims to fix part of this.
---
 llvm/include/llvm/CodeGen/BasicTTIImpl.h      |   5 +
 .../Analysis/CostModel/RISCV/vp-intrinsics.ll | 216 +++++++++---------
 2 files changed, 113 insertions(+), 108 deletions(-)

diff --git a/llvm/include/llvm/CodeGen/BasicTTIImpl.h b/llvm/include/llvm/CodeGen/BasicTTIImpl.h
index 39955558fdd0a..fefd36ec54ae2 100644
--- a/llvm/include/llvm/CodeGen/BasicTTIImpl.h
+++ b/llvm/include/llvm/CodeGen/BasicTTIImpl.h
@@ -1787,6 +1787,11 @@ class BasicTTIImplBase : public TargetTransformInfoImplCRTPBase<T> {
 
       std::optional<Intrinsic::ID> FID =
           VPIntrinsic::getFunctionalIntrinsicIDForVP(ICA.getID());
+
+      // Not functionally equivalent but close enough for cost modelling.
+      if (ICA.getID() == Intrinsic::experimental_vp_reverse)
+        FID = Intrinsic::vector_reverse;
+
       if (FID) {
         // Non-vp version will have same arg types except mask and vector
         // length.
diff --git a/llvm/test/Analysis/CostModel/RISCV/vp-intrinsics.ll b/llvm/test/Analysis/CostModel/RISCV/vp-intrinsics.ll
index dda00d34ecf10..39debd8e3dddc 100644
--- a/llvm/test/Analysis/CostModel/RISCV/vp-intrinsics.ll
+++ b/llvm/test/Analysis/CostModel/RISCV/vp-intrinsics.ll
@@ -1758,117 +1758,117 @@ define void @is.fpclass() {
 
 define void @reverse() {
 ; ARGBASED-LABEL: 'reverse'
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %1 = call <2 x i1> @llvm.experimental.vp.reverse.v2i1(<2 x i1> poison, <2 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %2 = call <4 x i1> @llvm.experimental.vp.reverse.v4i1(<4 x i1> poison, <4 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %3 = call <8 x i1> @llvm.experimental.vp.reverse.v8i1(<8 x i1> poison, <8 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %4 = call <16 x i1> @llvm.experimental.vp.reverse.v16i1(<16 x i1> poison, <16 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %5 = call <2 x i8> @llvm.experimental.vp.reverse.v2i8(<2 x i8> poison, <2 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %6 = call <4 x i8> @llvm.experimental.vp.reverse.v4i8(<4 x i8> poison, <4 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %7 = call <8 x i8> @llvm.experimental.vp.reverse.v8i8(<8 x i8> poison, <8 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %8 = call <16 x i8> @llvm.experimental.vp.reverse.v16i8(<16 x i8> poison, <16 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %9 = call <2 x i16> @llvm.experimental.vp.reverse.v2i16(<2 x i16> poison, <2 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %10 = call <4 x i16> @llvm.experimental.vp.reverse.v4i16(<4 x i16> poison, <4 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %11 = call <8 x i16> @llvm.experimental.vp.reverse.v8i16(<8 x i16> poison, <8 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %12 = call <16 x i16> @llvm.experimental.vp.reverse.v16i16(<16 x i16> poison, <16 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %13 = call <2 x i32> @llvm.experimental.vp.reverse.v2i32(<2 x i32> poison, <2 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %14 = call <4 x i32> @llvm.experimental.vp.reverse.v4i32(<4 x i32> poison, <4 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %15 = call <8 x i32> @llvm.experimental.vp.reverse.v8i32(<8 x i32> poison, <8 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %16 = call <16 x i32> @llvm.experimental.vp.reverse.v16i32(<16 x i32> poison, <16 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %17 = call <2 x i64> @llvm.experimental.vp.reverse.v2i64(<2 x i64> poison, <2 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %18 = call <4 x i64> @llvm.experimental.vp.reverse.v4i64(<4 x i64> poison, <4 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %19 = call <8 x i64> @llvm.experimental.vp.reverse.v8i64(<8 x i64> poison, <8 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %20 = call <16 x i64> @llvm.experimental.vp.reverse.v16i64(<16 x i64> poison, <16 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %21 = call <2 x bfloat> @llvm.experimental.vp.reverse.v2bf16(<2 x bfloat> poison, <2 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %22 = call <4 x bfloat> @llvm.experimental.vp.reverse.v4bf16(<4 x bfloat> poison, <4 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %23 = call <8 x bfloat> @llvm.experimental.vp.reverse.v8bf16(<8 x bfloat> poison, <8 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %24 = call <16 x bfloat> @llvm.experimental.vp.reverse.v16bf16(<16 x bfloat> poison, <16 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %25 = call <2 x half> @llvm.experimental.vp.reverse.v2f16(<2 x half> poison, <2 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %26 = call <4 x half> @llvm.experimental.vp.reverse.v4f16(<4 x half> poison, <4 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %27 = call <8 x half> @llvm.experimental.vp.reverse.v8f16(<8 x half> poison, <8 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %28 = call <16 x half> @llvm.experimental.vp.reverse.v16f16(<16 x half> poison, <16 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %29 = call <2 x float> @llvm.experimental.vp.reverse.v2f32(<2 x float> poison, <2 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %30 = call <4 x float> @llvm.experimental.vp.reverse.v4f32(<4 x float> poison, <4 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %31 = call <8 x float> @llvm.experimental.vp.reverse.v8f32(<8 x float> poison, <8 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %32 = call <16 x float> @llvm.experimental.vp.reverse.v16f32(<16 x float> poison, <16 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %33 = call <2 x double> @llvm.experimental.vp.reverse.v2f64(<2 x double> poison, <2 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %34 = call <4 x double> @llvm.experimental.vp.reverse.v4f64(<4 x double> poison, <4 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %35 = call <8 x double> @llvm.experimental.vp.reverse.v8f64(<8 x double> poison, <8 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %36 = call <16 x double> @llvm.experimental.vp.reverse.v16f64(<16 x double> poison, <16 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %37 = call <vscale x 2 x i1> @llvm.experimental.vp.reverse.nxv2i1(<vscale x 2 x i1> poison, <vscale x 2 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %38 = call <vscale x 4 x i1> @llvm.experimental.vp.reverse.nxv4i1(<vscale x 4 x i1> poison, <vscale x 4 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %39 = call <vscale x 8 x i1> @llvm.experimental.vp.reverse.nxv8i1(<vscale x 8 x i1> poison, <vscale x 8 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %40 = call <vscale x 16 x i1> @llvm.experimental.vp.reverse.nxv16i1(<vscale x 16 x i1> poison, <vscale x 16 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %41 = call <vscale x 2 x i8> @llvm.experimental.vp.reverse.nxv2i8(<vscale x 2 x i8> poison, <vscale x 2 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %42 = call <vscale x 4 x i8> @llvm.experimental.vp.reverse.nxv4i8(<vscale x 4 x i8> poison, <vscale x 4 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %43 = call <vscale x 8 x i8> @llvm.experimental.vp.reverse.nxv8i8(<vscale x 8 x i8> poison, <vscale x 8 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %44 = call <vscale x 16 x i8> @llvm.experimental.vp.reverse.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %45 = call <vscale x 2 x i16> @llvm.experimental.vp.reverse.nxv2i16(<vscale x 2 x i16> poison, <vscale x 2 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %46 = call <vscale x 4 x i16> @llvm.experimental.vp.reverse.nxv4i16(<vscale x 4 x i16> poison, <vscale x 4 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %47 = call <vscale x 8 x i16> @llvm.experimental.vp.reverse.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %48 = call <vscale x 16 x i16> @llvm.experimental.vp.reverse.nxv16i16(<vscale x 16 x i16> poison, <vscale x 16 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %49 = call <vscale x 2 x i32> @llvm.experimental.vp.reverse.nxv2i32(<vscale x 2 x i32> poison, <vscale x 2 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %50 = call <vscale x 4 x i32> @llvm.experimental.vp.reverse.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %51 = call <vscale x 8 x i32> @llvm.experimental.vp.reverse.nxv8i32(<vscale x 8 x i32> poison, <vscale x 8 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %52 = call <vscale x 16 x i32> @llvm.experimental.vp.reverse.nxv16i32(<vscale x 16 x i32> poison, <vscale x 16 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %53 = call <vscale x 2 x i64> @llvm.experimental.vp.reverse.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %54 = call <vscale x 4 x i64> @llvm.experimental.vp.reverse.nxv4i64(<vscale x 4 x i64> poison, <vscale x 4 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %55 = call <vscale x 8 x i64> @llvm.experimental.vp.reverse.nxv8i64(<vscale x 8 x i64> poison, <vscale x 8 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %56 = call <vscale x 16 x i64> @llvm.experimental.vp.reverse.nxv16i64(<vscale x 16 x i64> poison, <vscale x 16 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %57 = call <vscale x 2 x bfloat> @llvm.experimental.vp.reverse.nxv2bf16(<vscale x 2 x bfloat> poison, <vscale x 2 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %58 = call <vscale x 4 x bfloat> @llvm.experimental.vp.reverse.nxv4bf16(<vscale x 4 x bfloat> poison, <vscale x 4 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %59 = call <vscale x 8 x bfloat> @llvm.experimental.vp.reverse.nxv8bf16(<vscale x 8 x bfloat> poison, <vscale x 8 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %60 = call <vscale x 16 x bfloat> @llvm.experimental.vp.reverse.nxv16bf16(<vscale x 16 x bfloat> poison, <vscale x 16 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %61 = call <vscale x 2 x half> @llvm.experimental.vp.reverse.nxv2f16(<vscale x 2 x half> poison, <vscale x 2 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %62 = call <vscale x 4 x half> @llvm.experimental.vp.reverse.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %63 = call <vscale x 8 x half> @llvm.experimental.vp.reverse.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %64 = call <vscale x 16 x half> @llvm.experimental.vp.reverse.nxv16f16(<vscale x 16 x half> poison, <vscale x 16 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %65 = call <vscale x 2 x float> @llvm.experimental.vp.reverse.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %66 = call <vscale x 4 x float> @llvm.experimental.vp.reverse.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %67 = call <vscale x 8 x float> @llvm.experimental.vp.reverse.nxv8f32(<vscale x 8 x float> poison, <vscale x 8 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %68 = call <vscale x 16 x float> @llvm.experimental.vp.reverse.nxv16f32(<vscale x 16 x float> poison, <vscale x 16 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %69 = call <vscale x 2 x double> @llvm.experimental.vp.reverse.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %70 = call <vscale x 4 x double> @llvm.experimental.vp.reverse.nxv4f64(<vscale x 4 x double> poison, <vscale x 4 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %71 = call <vscale x 8 x double> @llvm.experimental.vp.reverse.nxv8f64(<vscale x 8 x double> poison, <vscale x 8 x i1> poison, i32 poison)
-; ARGBASED-NEXT:  Cost Model: Invalid cost for instruction: %72 = call <vscale x 16 x double> @llvm.experimental.vp.reverse.nxv16f64(<vscale x 16 x double> poison, <vscale x 16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %1 = call <2 x i1> @llvm.experimental.vp.reverse.v2i1(<2 x i1> poison, <2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %2 = call <4 x i1> @llvm.experimental.vp.reverse.v4i1(<4 x i1> poison, <4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %3 = call <8 x i1> @llvm.experimental.vp.reverse.v8i1(<8 x i1> poison, <8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %4 = call <16 x i1> @llvm.experimental.vp.reverse.v16i1(<16 x i1> poison, <16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %5 = call <2 x i8> @llvm.experimental.vp.reverse.v2i8(<2 x i8> poison, <2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %6 = call <4 x i8> @llvm.experimental.vp.reverse.v4i8(<4 x i8> poison, <4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %7 = call <8 x i8> @llvm.experimental.vp.reverse.v8i8(<8 x i8> poison, <8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %8 = call <16 x i8> @llvm.experimental.vp.reverse.v16i8(<16 x i8> poison, <16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %9 = call <2 x i16> @llvm.experimental.vp.reverse.v2i16(<2 x i16> poison, <2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %10 = call <4 x i16> @llvm.experimental.vp.reverse.v4i16(<4 x i16> poison, <4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %11 = call <8 x i16> @llvm.experimental.vp.reverse.v8i16(<8 x i16> poison, <8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %12 = call <16 x i16> @llvm.experimental.vp.reverse.v16i16(<16 x i16> poison, <16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %13 = call <2 x i32> @llvm.experimental.vp.reverse.v2i32(<2 x i32> poison, <2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %14 = call <4 x i32> @llvm.experimental.vp.reverse.v4i32(<4 x i32> poison, <4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %15 = call <8 x i32> @llvm.experimental.vp.reverse.v8i32(<8 x i32> poison, <8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %16 = call <16 x i32> @llvm.experimental.vp.reverse.v16i32(<16 x i32> poison, <16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %17 = call <2 x i64> @llvm.experimental.vp.reverse.v2i64(<2 x i64> poison, <2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %18 = call <4 x i64> @llvm.experimental.vp.reverse.v4i64(<4 x i64> poison, <4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %19 = call <8 x i64> @llvm.experimental.vp.reverse.v8i64(<8 x i64> poison, <8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %20 = call <16 x i64> @llvm.experimental.vp.reverse.v16i64(<16 x i64> poison, <16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %21 = call <2 x bfloat> @llvm.experimental.vp.reverse.v2bf16(<2 x bfloat> poison, <2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %22 = call <4 x bfloat> @llvm.experimental.vp.reverse.v4bf16(<4 x bfloat> poison, <4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %23 = call <8 x bfloat> @llvm.experimental.vp.reverse.v8bf16(<8 x bfloat> poison, <8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %24 = call <16 x bfloat> @llvm.experimental.vp.reverse.v16bf16(<16 x bfloat> poison, <16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %25 = call <2 x half> @llvm.experimental.vp.reverse.v2f16(<2 x half> poison, <2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %26 = call <4 x half> @llvm.experimental.vp.reverse.v4f16(<4 x half> poison, <4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %27 = call <8 x half> @llvm.experimental.vp.reverse.v8f16(<8 x half> poison, <8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %28 = call <16 x half> @llvm.experimental.vp.reverse.v16f16(<16 x half> poison, <16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %29 = call <2 x float> @llvm.experimental.vp.reverse.v2f32(<2 x float> poison, <2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %30 = call <4 x float> @llvm.experimental.vp.reverse.v4f32(<4 x float> poison, <4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %31 = call <8 x float> @llvm.experimental.vp.reverse.v8f32(<8 x float> poison, <8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %32 = call <16 x float> @llvm.experimental.vp.reverse.v16f32(<16 x float> poison, <16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %33 = call <2 x double> @llvm.experimental.vp.reverse.v2f64(<2 x double> poison, <2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %34 = call <4 x double> @llvm.experimental.vp.reverse.v4f64(<4 x double> poison, <4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %35 = call <8 x double> @llvm.experimental.vp.reverse.v8f64(<8 x double> poison, <8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %36 = call <16 x double> @llvm.experimental.vp.reverse.v16f64(<16 x double> poison, <16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %37 = call <vscale x 2 x i1> @llvm.experimental.vp.reverse.nxv2i1(<vscale x 2 x i1> poison, <vscale x 2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %38 = call <vscale x 4 x i1> @llvm.experimental.vp.reverse.nxv4i1(<vscale x 4 x i1> poison, <vscale x 4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %39 = call <vscale x 8 x i1> @llvm.experimental.vp.reverse.nxv8i1(<vscale x 8 x i1> poison, <vscale x 8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %40 = call <vscale x 16 x i1> @llvm.experimental.vp.reverse.nxv16i1(<vscale x 16 x i1> poison, <vscale x 16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %41 = call <vscale x 2 x i8> @llvm.experimental.vp.reverse.nxv2i8(<vscale x 2 x i8> poison, <vscale x 2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %42 = call <vscale x 4 x i8> @llvm.experimental.vp.reverse.nxv4i8(<vscale x 4 x i8> poison, <vscale x 4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %43 = call <vscale x 8 x i8> @llvm.experimental.vp.reverse.nxv8i8(<vscale x 8 x i8> poison, <vscale x 8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %44 = call <vscale x 16 x i8> @llvm.experimental.vp.reverse.nxv16i8(<vscale x 16 x i8> poison, <vscale x 16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %45 = call <vscale x 2 x i16> @llvm.experimental.vp.reverse.nxv2i16(<vscale x 2 x i16> poison, <vscale x 2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %46 = call <vscale x 4 x i16> @llvm.experimental.vp.reverse.nxv4i16(<vscale x 4 x i16> poison, <vscale x 4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %47 = call <vscale x 8 x i16> @llvm.experimental.vp.reverse.nxv8i16(<vscale x 8 x i16> poison, <vscale x 8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %48 = call <vscale x 16 x i16> @llvm.experimental.vp.reverse.nxv16i16(<vscale x 16 x i16> poison, <vscale x 16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %49 = call <vscale x 2 x i32> @llvm.experimental.vp.reverse.nxv2i32(<vscale x 2 x i32> poison, <vscale x 2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %50 = call <vscale x 4 x i32> @llvm.experimental.vp.reverse.nxv4i32(<vscale x 4 x i32> poison, <vscale x 4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %51 = call <vscale x 8 x i32> @llvm.experimental.vp.reverse.nxv8i32(<vscale x 8 x i32> poison, <vscale x 8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %52 = call <vscale x 16 x i32> @llvm.experimental.vp.reverse.nxv16i32(<vscale x 16 x i32> poison, <vscale x 16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %53 = call <vscale x 2 x i64> @llvm.experimental.vp.reverse.nxv2i64(<vscale x 2 x i64> poison, <vscale x 2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %54 = call <vscale x 4 x i64> @llvm.experimental.vp.reverse.nxv4i64(<vscale x 4 x i64> poison, <vscale x 4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %55 = call <vscale x 8 x i64> @llvm.experimental.vp.reverse.nxv8i64(<vscale x 8 x i64> poison, <vscale x 8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %56 = call <vscale x 16 x i64> @llvm.experimental.vp.reverse.nxv16i64(<vscale x 16 x i64> poison, <vscale x 16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %57 = call <vscale x 2 x bfloat> @llvm.experimental.vp.reverse.nxv2bf16(<vscale x 2 x bfloat> poison, <vscale x 2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %58 = call <vscale x 4 x bfloat> @llvm.experimental.vp.reverse.nxv4bf16(<vscale x 4 x bfloat> poison, <vscale x 4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %59 = call <vscale x 8 x bfloat> @llvm.experimental.vp.reverse.nxv8bf16(<vscale x 8 x bfloat> poison, <vscale x 8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %60 = call <vscale x 16 x bfloat> @llvm.experimental.vp.reverse.nxv16bf16(<vscale x 16 x bfloat> poison, <vscale x 16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %61 = call <vscale x 2 x half> @llvm.experimental.vp.reverse.nxv2f16(<vscale x 2 x half> poison, <vscale x 2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %62 = call <vscale x 4 x half> @llvm.experimental.vp.reverse.nxv4f16(<vscale x 4 x half> poison, <vscale x 4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %63 = call <vscale x 8 x half> @llvm.experimental.vp.reverse.nxv8f16(<vscale x 8 x half> poison, <vscale x 8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %64 = call <vscale x 16 x half> @llvm.experimental.vp.reverse.nxv16f16(<vscale x 16 x half> poison, <vscale x 16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %65 = call <vscale x 2 x float> @llvm.experimental.vp.reverse.nxv2f32(<vscale x 2 x float> poison, <vscale x 2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %66 = call <vscale x 4 x float> @llvm.experimental.vp.reverse.nxv4f32(<vscale x 4 x float> poison, <vscale x 4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %67 = call <vscale x 8 x float> @llvm.experimental.vp.reverse.nxv8f32(<vscale x 8 x float> poison, <vscale x 8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %68 = call <vscale x 16 x float> @llvm.experimental.vp.reverse.nxv16f32(<vscale x 16 x float> poison, <vscale x 16 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %69 = call <vscale x 2 x double> @llvm.experimental.vp.reverse.nxv2f64(<vscale x 2 x double> poison, <vscale x 2 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %70 = call <vscale x 4 x double> @llvm.experimental.vp.reverse.nxv4f64(<vscale x 4 x double> poison, <vscale x 4 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %71 = call <vscale x 8 x double> @llvm.experimental.vp.reverse.nxv8f64(<vscale x 8 x double> poison, <vscale x 8 x i1> poison, i32 poison)
+; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %72 = call <vscale x 16 x double> @llvm.experimental.vp.reverse.nxv16f64(<vscale x 16 x double> poison, <vscale x 16 x i1> poison, i32 poison)
 ; ARGBASED-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; TYPEBASED-LABEL: 'reverse'
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %1 = call <2 x i1> @llvm.experimental.vp.reverse.v2i1(<2 x i1> poison, <2 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %2 = call <4 x i1> @llvm.experimental.vp.reverse.v4i1(<4 x i1> poison, <4 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %3 = call <8 x i1> @llvm.experimental.vp.reverse.v8i1(<8 x i1> poison, <8 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %4 = call <16 x i1> @llvm.experimental.vp.reverse.v16i1(<16 x i1> poison, <16 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %5 = call <2 x i8> @llvm.experimental.vp.reverse.v2i8(<2 x i8> poison, <2 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %6 = call <4 x i8> @llvm.experimental.vp.reverse.v4i8(<4 x i8> poison, <4 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: %7 = call <8 x i8> @llvm.experimental.vp.reverse.v8i8(<8 x i8> poison, <8 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 126 for instruction: %8 = call <16 x i8> @llvm.experimental.vp.reverse.v16i8(<16 x i8> poison, <16 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %9 = call <2 x i16> @llvm.experimental.vp.reverse.v2i16(<2 x i16> poison, <2 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %10 = call <4 x i16> @llvm.experimental.vp.reverse.v4i16(<4 x i16> poison, <4 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: %11 = call <8 x i16> @llvm.experimental.vp.reverse.v8i16(<8 x i16> poison, <8 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 141 for instruction: %12 = call <16 x i16> @llvm.experimental.vp.reverse.v16i16(<16 x i16> poison, <16 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %13 = call <2 x i32> @llvm.experimental.vp.reverse.v2i32(<2 x i32> poison, <2 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %14 = call <4 x i32> @llvm.experimental.vp.reverse.v4i32(<4 x i32> poison, <4 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 69 for instruction: %15 = call <8 x i32> @llvm.experimental.vp.reverse.v8i32(<8 x i32> poison, <8 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 141 for instruction: %16 = call <16 x i32> @llvm.experimental.vp.reverse.v16i32(<16 x i32> poison, <16 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %17 = call <2 x i64> @llvm.experimental.vp.reverse.v2i64(<2 x i64> poison, <2 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %18 = call <4 x i64> @llvm.experimental.vp.reverse.v4i64(<4 x i64> poison, <4 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 69 for instruction: %19 = call <8 x i64> @llvm.experimental.vp.reverse.v8i64(<8 x i64> poison, <8 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 141 for instruction: %20 = call <16 x i64> @llvm.experimental.vp.reverse.v16i64(<16 x i64> poison, <16 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %21 = call <2 x bfloat> @llvm.experimental.vp.reverse.v2bf16(<2 x bfloat> poison, <2 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %22 = call <4 x bfloat> @llvm.experimental.vp.reverse.v4bf16(<4 x bfloat> poison, <4 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: %23 = call <8 x bfloat> @llvm.experimental.vp.reverse.v8bf16(<8 x bfloat> poison, <8 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 141 for instruction: %24 = call <16 x bfloat> @llvm.experimental.vp.reverse.v16bf16(<16 x bfloat> poison, <16 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %25 = call <2 x half> @llvm.experimental.vp.reverse.v2f16(<2 x half> poison, <2 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %26 = call <4 x half> @llvm.experimental.vp.reverse.v4f16(<4 x half> poison, <4 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: %27 = call <8 x half> @llvm.experimental.vp.reverse.v8f16(<8 x half> poison, <8 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 141 for instruction: %28 = call <16 x half> @llvm.experimental.vp.reverse.v16f16(<16 x half> poison, <16 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %29 = call <2 x float> @llvm.experimental.vp.reverse.v2f32(<2 x float> poison, <2 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %30 = call <4 x float> @llvm.experimental.vp.reverse.v4f32(<4 x float> poison, <4 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 69 for instruction: %31 = call <8 x float> @llvm.experimental.vp.reverse.v8f32(<8 x float> poison, <8 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 141 for instruction: %32 = call <16 x float> @llvm.experimental.vp.reverse.v16f32(<16 x float> poison, <16 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %33 = call <2 x double> @llvm.experimental.vp.reverse.v2f64(<2 x double> poison, <2 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %34 = call <4 x double> @llvm.experimental.vp.reverse.v4f64(<4 x double> poison, <4 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 69 for instruction: %35 = call <8 x double> @llvm.experimental.vp.reverse.v8f64(<8 x double> poison, <8 x i1> poison, i32 poison)
-; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 141 for instruction: %36 = call <16 x double> @llvm.experimental.vp.reverse.v16f64(<16 x double> poison, <16 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %1 = call <2 x i1> @llvm.experimental.vp.reverse.v2i1(<2 x i1> poison, <2 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %2 = call <4 x i1> @llvm.experimental.vp.reverse.v4i1(<4 x i1> poison, <4 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %3 = call <8 x i1> @llvm.experimental.vp.reverse.v8i1(<8 x i1> poison, <8 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: %4 = call <16 x i1> @llvm.experimental.vp.reverse.v16i1(<16 x i1> poison, <16 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %5 = call <2 x i8> @llvm.experimental.vp.reverse.v2i8(<2 x i8> poison, <2 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %6 = call <4 x i8> @llvm.experimental.vp.reverse.v4i8(<4 x i8> poison, <4 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %7 = call <8 x i8> @llvm.experimental.vp.reverse.v8i8(<8 x i8> poison, <8 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: %8 = call <16 x i8> @llvm.experimental.vp.reverse.v16i8(<16 x i8> poison, <16 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %9 = call <2 x i16> @llvm.experimental.vp.reverse.v2i16(<2 x i16> poison, <2 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %10 = call <4 x i16> @llvm.experimental.vp.reverse.v4i16(<4 x i16> poison, <4 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %11 = call <8 x i16> @llvm.experimental.vp.reverse.v8i16(<8 x i16> poison, <8 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 78 for instruction: %12 = call <16 x i16> @llvm.experimental.vp.reverse.v16i16(<16 x i16> poison, <16 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %13 = call <2 x i32> @llvm.experimental.vp.reverse.v2i32(<2 x i32> poison, <2 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %14 = call <4 x i32> @llvm.experimental.vp.reverse.v4i32(<4 x i32> poison, <4 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %15 = call <8 x i32> @llvm.experimental.vp.reverse.v8i32(<8 x i32> poison, <8 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 78 for instruction: %16 = call <16 x i32> @llvm.experimental.vp.reverse.v16i32(<16 x i32> poison, <16 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %17 = call <2 x i64> @llvm.experimental.vp.reverse.v2i64(<2 x i64> poison, <2 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %18 = call <4 x i64> @llvm.experimental.vp.reverse.v4i64(<4 x i64> poison, <4 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %19 = call <8 x i64> @llvm.experimental.vp.reverse.v8i64(<8 x i64> poison, <8 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 78 for instruction: %20 = call <16 x i64> @llvm.experimental.vp.reverse.v16i64(<16 x i64> poison, <16 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %21 = call <2 x bfloat> @llvm.experimental.vp.reverse.v2bf16(<2 x bfloat> poison, <2 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %22 = call <4 x bfloat> @llvm.experimental.vp.reverse.v4bf16(<4 x bfloat> poison, <4 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %23 = call <8 x bfloat> @llvm.experimental.vp.reverse.v8bf16(<8 x bfloat> poison, <8 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 78 for instruction: %24 = call <16 x bfloat> @llvm.experimental.vp.reverse.v16bf16(<16 x bfloat> poison, <16 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %25 = call <2 x half> @llvm.experimental.vp.reverse.v2f16(<2 x half> poison, <2 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %26 = call <4 x half> @llvm.experimental.vp.reverse.v4f16(<4 x half> poison, <4 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %27 = call <8 x half> @llvm.experimental.vp.reverse.v8f16(<8 x half> poison, <8 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 78 for instruction: %28 = call <16 x half> @llvm.experimental.vp.reverse.v16f16(<16 x half> poison, <16 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %29 = call <2 x float> @llvm.experimental.vp.reverse.v2f32(<2 x float> poison, <2 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %30 = call <4 x float> @llvm.experimental.vp.reverse.v4f32(<4 x float> poison, <4 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %31 = call <8 x float> @llvm.experimental.vp.reverse.v8f32(<8 x float> poison, <8 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 78 for instruction: %32 = call <16 x float> @llvm.experimental.vp.reverse.v16f32(<16 x float> poison, <16 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %33 = call <2 x double> @llvm.experimental.vp.reverse.v2f64(<2 x double> poison, <2 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %34 = call <4 x double> @llvm.experimental.vp.reverse.v4f64(<4 x double> poison, <4 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %35 = call <8 x double> @llvm.experimental.vp.reverse.v8f64(<8 x double> poison, <8 x i1> poison, i32 poison)
+; TYPEBASED-NEXT:  Cost Model: Found an estimated cost of 78 for instruction: %36 = call <16 x double> @llvm.experimental.vp.reverse.v16f64(<16 x double> poison, <16 x i1> poison, i32 poison)
 ; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %37 = call <vscale x 2 x i1> @llvm.experimental.vp.reverse.nxv2i1(<vscale x 2 x i1> poison, <vscale x 2 x i1> poison, i32 poison)
 ; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %38 = call <vscale x 4 x i1> @llvm.experimental.vp.reverse.nxv4i1(<vscale x 4 x i1> poison, <vscale x 4 x i1> poison, i32 poison)
 ; TYPEBASED-NEXT:  Cost Model: Invalid cost for instruction: %39 = call <vscale x 8 x i1> @llvm.experimental.vp.reverse.nxv8i1(<vscale x 8 x i1> poison, <vscale x 8 x i1> poison, i32 poison)