[llvm] [X86] Add ctpop/ctlz/cttz test coverage for very large integers (PR #164450)
Simon Pilgrim via llvm-commits
llvm-commits at lists.llvm.org
Tue Oct 21 09:02:25 PDT 2025
https://github.com/RKSimon created https://github.com/llvm/llvm-project/pull/164450
Inspired by #164275
>From 66d4b11205df5a3ace4151cbabed8da18243aed5 Mon Sep 17 00:00:00 2001
From: Simon Pilgrim <llvm-dev at redking.me.uk>
Date: Tue, 21 Oct 2025 16:53:50 +0100
Subject: [PATCH] [X86] Add ctpop/ctlz/cttz test coverage for very large
integers
Inspired by #164275
---
llvm/test/CodeGen/X86/bitcnt-big-integer.ll | 3021 +++++++++++++++++++
1 file changed, 3021 insertions(+)
create mode 100644 llvm/test/CodeGen/X86/bitcnt-big-integer.ll
diff --git a/llvm/test/CodeGen/X86/bitcnt-big-integer.ll b/llvm/test/CodeGen/X86/bitcnt-big-integer.ll
new file mode 100644
index 0000000000000..13149d78b16fb
--- /dev/null
+++ b/llvm/test/CodeGen/X86/bitcnt-big-integer.ll
@@ -0,0 +1,3021 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=x86-64-v2 | FileCheck %s --check-prefixes=CHECK,SSE
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=x86-64-v3 | FileCheck %s --check-prefixes=CHECK,AVX2
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=x86-64-v4 | FileCheck %s --check-prefixes=CHECK,AVX512
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=x86-64-v4 -mattr=+avx512vpopcntdq | FileCheck %s --check-prefixes=CHECK,AVX512
+
+;
+; CTPOP
+;
+
+define i32 @test_ctpop_i128(i128 %a0) nounwind {
+; CHECK-LABEL: test_ctpop_i128:
+; CHECK: # %bb.0:
+; CHECK-NEXT: popcntq %rsi, %rcx
+; CHECK-NEXT: popcntq %rdi, %rax
+; CHECK-NEXT: addl %ecx, %eax
+; CHECK-NEXT: # kill: def $eax killed $eax killed $rax
+; CHECK-NEXT: retq
+ %cnt = call i128 @llvm.ctpop.i128(i128 %a0)
+ %res = trunc i128 %cnt to i32
+ ret i32 %res
+}
+
+define i32 @load_ctpop_i128(ptr %p0) nounwind {
+; CHECK-LABEL: load_ctpop_i128:
+; CHECK: # %bb.0:
+; CHECK-NEXT: popcntq 8(%rdi), %rcx
+; CHECK-NEXT: popcntq (%rdi), %rax
+; CHECK-NEXT: addl %ecx, %eax
+; CHECK-NEXT: # kill: def $eax killed $eax killed $rax
+; CHECK-NEXT: retq
+ %a0 = load i128, ptr %p0
+ %cnt = call i128 @llvm.ctpop.i128(i128 %a0)
+ %res = trunc i128 %cnt to i32
+ ret i32 %res
+}
+
+define i32 @test_ctpop_i256(i256 %a0) nounwind {
+; CHECK-LABEL: test_ctpop_i256:
+; CHECK: # %bb.0:
+; CHECK-NEXT: popcntq %rcx, %rax
+; CHECK-NEXT: xorl %ecx, %ecx
+; CHECK-NEXT: popcntq %rdx, %rcx
+; CHECK-NEXT: addl %eax, %ecx
+; CHECK-NEXT: xorl %edx, %edx
+; CHECK-NEXT: popcntq %rsi, %rdx
+; CHECK-NEXT: xorl %eax, %eax
+; CHECK-NEXT: popcntq %rdi, %rax
+; CHECK-NEXT: addl %edx, %eax
+; CHECK-NEXT: addl %ecx, %eax
+; CHECK-NEXT: # kill: def $eax killed $eax killed $rax
+; CHECK-NEXT: retq
+ %cnt = call i256 @llvm.ctpop.i256(i256 %a0)
+ %res = trunc i256 %cnt to i32
+ ret i32 %res
+}
+
+define i32 @load_ctpop_i256(ptr %p0) nounwind {
+; SSE-LABEL: load_ctpop_i256:
+; SSE: # %bb.0:
+; SSE-NEXT: popcntq 24(%rdi), %rcx
+; SSE-NEXT: popcntq 16(%rdi), %rdx
+; SSE-NEXT: popcntq 8(%rdi), %rsi
+; SSE-NEXT: popcntq (%rdi), %rax
+; SSE-NEXT: addl %ecx, %edx
+; SSE-NEXT: addl %esi, %eax
+; SSE-NEXT: addl %edx, %eax
+; SSE-NEXT: # kill: def $eax killed $eax killed $rax
+; SSE-NEXT: retq
+;
+; AVX2-LABEL: load_ctpop_i256:
+; AVX2: # %bb.0:
+; AVX2-NEXT: popcntq 24(%rdi), %rax
+; AVX2-NEXT: popcntq 16(%rdi), %rcx
+; AVX2-NEXT: addl %eax, %ecx
+; AVX2-NEXT: popcntq 8(%rdi), %rdx
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: popcntq (%rdi), %rax
+; AVX2-NEXT: addl %edx, %eax
+; AVX2-NEXT: addl %ecx, %eax
+; AVX2-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: load_ctpop_i256:
+; AVX512: # %bb.0:
+; AVX512-NEXT: popcntq 24(%rdi), %rax
+; AVX512-NEXT: popcntq 16(%rdi), %rcx
+; AVX512-NEXT: addl %eax, %ecx
+; AVX512-NEXT: popcntq 8(%rdi), %rdx
+; AVX512-NEXT: xorl %eax, %eax
+; AVX512-NEXT: popcntq (%rdi), %rax
+; AVX512-NEXT: addl %edx, %eax
+; AVX512-NEXT: addl %ecx, %eax
+; AVX512-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX512-NEXT: retq
+ %a0 = load i256, ptr %p0
+ %cnt = call i256 @llvm.ctpop.i256(i256 %a0)
+ %res = trunc i256 %cnt to i32
+ ret i32 %res
+}
+
+define i32 @test_ctpop_i512(i512 %a0) nounwind {
+; CHECK-LABEL: test_ctpop_i512:
+; CHECK: # %bb.0:
+; CHECK-NEXT: popcntq {{[0-9]+}}(%rsp), %rax
+; CHECK-NEXT: popcntq {{[0-9]+}}(%rsp), %r10
+; CHECK-NEXT: addl %eax, %r10d
+; CHECK-NEXT: xorl %eax, %eax
+; CHECK-NEXT: popcntq %r9, %rax
+; CHECK-NEXT: popcntq %r8, %r8
+; CHECK-NEXT: addl %eax, %r8d
+; CHECK-NEXT: addl %r10d, %r8d
+; CHECK-NEXT: xorl %eax, %eax
+; CHECK-NEXT: popcntq %rcx, %rax
+; CHECK-NEXT: xorl %ecx, %ecx
+; CHECK-NEXT: popcntq %rdx, %rcx
+; CHECK-NEXT: addl %eax, %ecx
+; CHECK-NEXT: xorl %edx, %edx
+; CHECK-NEXT: popcntq %rsi, %rdx
+; CHECK-NEXT: xorl %eax, %eax
+; CHECK-NEXT: popcntq %rdi, %rax
+; CHECK-NEXT: addl %edx, %eax
+; CHECK-NEXT: addl %ecx, %eax
+; CHECK-NEXT: addl %r8d, %eax
+; CHECK-NEXT: # kill: def $eax killed $eax killed $rax
+; CHECK-NEXT: retq
+ %cnt = call i512 @llvm.ctpop.i512(i512 %a0)
+ %res = trunc i512 %cnt to i32
+ ret i32 %res
+}
+
+define i32 @load_ctpop_i512(ptr %p0) nounwind {
+; SSE-LABEL: load_ctpop_i512:
+; SSE: # %bb.0:
+; SSE-NEXT: popcntq 56(%rdi), %rax
+; SSE-NEXT: popcntq 48(%rdi), %rcx
+; SSE-NEXT: popcntq 40(%rdi), %rdx
+; SSE-NEXT: popcntq 32(%rdi), %rsi
+; SSE-NEXT: addl %eax, %ecx
+; SSE-NEXT: addl %edx, %esi
+; SSE-NEXT: xorl %eax, %eax
+; SSE-NEXT: popcntq 24(%rdi), %rax
+; SSE-NEXT: addl %ecx, %esi
+; SSE-NEXT: xorl %ecx, %ecx
+; SSE-NEXT: popcntq 16(%rdi), %rcx
+; SSE-NEXT: addl %eax, %ecx
+; SSE-NEXT: xorl %edx, %edx
+; SSE-NEXT: popcntq 8(%rdi), %rdx
+; SSE-NEXT: xorl %eax, %eax
+; SSE-NEXT: popcntq (%rdi), %rax
+; SSE-NEXT: addl %edx, %eax
+; SSE-NEXT: addl %ecx, %eax
+; SSE-NEXT: addl %esi, %eax
+; SSE-NEXT: # kill: def $eax killed $eax killed $rax
+; SSE-NEXT: retq
+;
+; AVX2-LABEL: load_ctpop_i512:
+; AVX2: # %bb.0:
+; AVX2-NEXT: popcntq 56(%rdi), %rax
+; AVX2-NEXT: popcntq 48(%rdi), %rcx
+; AVX2-NEXT: addl %eax, %ecx
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: popcntq 40(%rdi), %rax
+; AVX2-NEXT: popcntq 32(%rdi), %rdx
+; AVX2-NEXT: addl %eax, %edx
+; AVX2-NEXT: addl %ecx, %edx
+; AVX2-NEXT: xorl %ecx, %ecx
+; AVX2-NEXT: popcntq 24(%rdi), %rcx
+; AVX2-NEXT: popcntq 16(%rdi), %rsi
+; AVX2-NEXT: popcntq 8(%rdi), %r8
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: popcntq (%rdi), %rax
+; AVX2-NEXT: addl %ecx, %esi
+; AVX2-NEXT: addl %r8d, %eax
+; AVX2-NEXT: addl %esi, %eax
+; AVX2-NEXT: addl %edx, %eax
+; AVX2-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: load_ctpop_i512:
+; AVX512: # %bb.0:
+; AVX512-NEXT: popcntq 56(%rdi), %rax
+; AVX512-NEXT: popcntq 48(%rdi), %rcx
+; AVX512-NEXT: addl %eax, %ecx
+; AVX512-NEXT: xorl %eax, %eax
+; AVX512-NEXT: popcntq 40(%rdi), %rax
+; AVX512-NEXT: popcntq 32(%rdi), %rdx
+; AVX512-NEXT: addl %eax, %edx
+; AVX512-NEXT: addl %ecx, %edx
+; AVX512-NEXT: xorl %eax, %eax
+; AVX512-NEXT: popcntq 24(%rdi), %rax
+; AVX512-NEXT: xorl %ecx, %ecx
+; AVX512-NEXT: popcntq 16(%rdi), %rcx
+; AVX512-NEXT: popcntq 8(%rdi), %rsi
+; AVX512-NEXT: addl %eax, %ecx
+; AVX512-NEXT: xorl %eax, %eax
+; AVX512-NEXT: popcntq (%rdi), %rax
+; AVX512-NEXT: addl %esi, %eax
+; AVX512-NEXT: addl %ecx, %eax
+; AVX512-NEXT: addl %edx, %eax
+; AVX512-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX512-NEXT: retq
+ %a0 = load i512, ptr %p0
+ %cnt = call i512 @llvm.ctpop.i512(i512 %a0)
+ %res = trunc i512 %cnt to i32
+ ret i32 %res
+}
+
+define i32 @test_ctpop_i1024(i1024 %a0) nounwind {
+; SSE-LABEL: test_ctpop_i1024:
+; SSE: # %bb.0:
+; SSE-NEXT: pushq %rbx
+; SSE-NEXT: popcntq {{[0-9]+}}(%rsp), %rax
+; SSE-NEXT: popcntq {{[0-9]+}}(%rsp), %r10
+; SSE-NEXT: popcntq {{[0-9]+}}(%rsp), %r11
+; SSE-NEXT: addl %eax, %r10d
+; SSE-NEXT: xorl %eax, %eax
+; SSE-NEXT: popcntq {{[0-9]+}}(%rsp), %rax
+; SSE-NEXT: addl %r11d, %eax
+; SSE-NEXT: xorl %r11d, %r11d
+; SSE-NEXT: popcntq {{[0-9]+}}(%rsp), %r11
+; SSE-NEXT: xorl %ebx, %ebx
+; SSE-NEXT: popcntq {{[0-9]+}}(%rsp), %rbx
+; SSE-NEXT: addl %r10d, %eax
+; SSE-NEXT: addl %r11d, %ebx
+; SSE-NEXT: xorl %r11d, %r11d
+; SSE-NEXT: popcntq {{[0-9]+}}(%rsp), %r11
+; SSE-NEXT: xorl %r10d, %r10d
+; SSE-NEXT: popcntq {{[0-9]+}}(%rsp), %r10
+; SSE-NEXT: addl %r11d, %r10d
+; SSE-NEXT: addl %ebx, %r10d
+; SSE-NEXT: xorl %r11d, %r11d
+; SSE-NEXT: popcntq {{[0-9]+}}(%rsp), %r11
+; SSE-NEXT: xorl %ebx, %ebx
+; SSE-NEXT: popcntq {{[0-9]+}}(%rsp), %rbx
+; SSE-NEXT: addl %eax, %r10d
+; SSE-NEXT: addl %r11d, %ebx
+; SSE-NEXT: xorl %eax, %eax
+; SSE-NEXT: popcntq %r9, %rax
+; SSE-NEXT: popcntq %r8, %r8
+; SSE-NEXT: addl %eax, %r8d
+; SSE-NEXT: addl %ebx, %r8d
+; SSE-NEXT: xorl %eax, %eax
+; SSE-NEXT: popcntq %rcx, %rax
+; SSE-NEXT: xorl %ecx, %ecx
+; SSE-NEXT: popcntq %rdx, %rcx
+; SSE-NEXT: addl %eax, %ecx
+; SSE-NEXT: xorl %edx, %edx
+; SSE-NEXT: popcntq %rsi, %rdx
+; SSE-NEXT: xorl %eax, %eax
+; SSE-NEXT: popcntq %rdi, %rax
+; SSE-NEXT: addl %edx, %eax
+; SSE-NEXT: addl %ecx, %eax
+; SSE-NEXT: addl %r8d, %eax
+; SSE-NEXT: addl %r10d, %eax
+; SSE-NEXT: # kill: def $eax killed $eax killed $rax
+; SSE-NEXT: popq %rbx
+; SSE-NEXT: retq
+;
+; AVX2-LABEL: test_ctpop_i1024:
+; AVX2: # %bb.0:
+; AVX2-NEXT: pushq %r14
+; AVX2-NEXT: pushq %rbx
+; AVX2-NEXT: popcntq {{[0-9]+}}(%rsp), %rax
+; AVX2-NEXT: popcntq {{[0-9]+}}(%rsp), %r10
+; AVX2-NEXT: addl %eax, %r10d
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: popcntq {{[0-9]+}}(%rsp), %rax
+; AVX2-NEXT: popcntq {{[0-9]+}}(%rsp), %r11
+; AVX2-NEXT: addl %eax, %r11d
+; AVX2-NEXT: addl %r10d, %r11d
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: popcntq {{[0-9]+}}(%rsp), %rax
+; AVX2-NEXT: xorl %ebx, %ebx
+; AVX2-NEXT: popcntq {{[0-9]+}}(%rsp), %rbx
+; AVX2-NEXT: xorl %r14d, %r14d
+; AVX2-NEXT: popcntq {{[0-9]+}}(%rsp), %r14
+; AVX2-NEXT: addl %eax, %ebx
+; AVX2-NEXT: xorl %r10d, %r10d
+; AVX2-NEXT: popcntq {{[0-9]+}}(%rsp), %r10
+; AVX2-NEXT: addl %r14d, %r10d
+; AVX2-NEXT: addl %ebx, %r10d
+; AVX2-NEXT: addl %r11d, %r10d
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: popcntq {{[0-9]+}}(%rsp), %rax
+; AVX2-NEXT: xorl %r11d, %r11d
+; AVX2-NEXT: popcntq {{[0-9]+}}(%rsp), %r11
+; AVX2-NEXT: addl %eax, %r11d
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: popcntq %r9, %rax
+; AVX2-NEXT: popcntq %r8, %r8
+; AVX2-NEXT: addl %eax, %r8d
+; AVX2-NEXT: addl %r11d, %r8d
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: popcntq %rcx, %rax
+; AVX2-NEXT: xorl %ecx, %ecx
+; AVX2-NEXT: popcntq %rdx, %rcx
+; AVX2-NEXT: addl %eax, %ecx
+; AVX2-NEXT: xorl %edx, %edx
+; AVX2-NEXT: popcntq %rsi, %rdx
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: popcntq %rdi, %rax
+; AVX2-NEXT: addl %edx, %eax
+; AVX2-NEXT: addl %ecx, %eax
+; AVX2-NEXT: addl %r8d, %eax
+; AVX2-NEXT: addl %r10d, %eax
+; AVX2-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX2-NEXT: popq %rbx
+; AVX2-NEXT: popq %r14
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_ctpop_i1024:
+; AVX512: # %bb.0:
+; AVX512-NEXT: pushq %r14
+; AVX512-NEXT: pushq %rbx
+; AVX512-NEXT: popcntq {{[0-9]+}}(%rsp), %rax
+; AVX512-NEXT: popcntq {{[0-9]+}}(%rsp), %r10
+; AVX512-NEXT: addl %eax, %r10d
+; AVX512-NEXT: xorl %eax, %eax
+; AVX512-NEXT: popcntq {{[0-9]+}}(%rsp), %rax
+; AVX512-NEXT: popcntq {{[0-9]+}}(%rsp), %r11
+; AVX512-NEXT: addl %eax, %r11d
+; AVX512-NEXT: addl %r10d, %r11d
+; AVX512-NEXT: xorl %eax, %eax
+; AVX512-NEXT: popcntq {{[0-9]+}}(%rsp), %rax
+; AVX512-NEXT: xorl %ebx, %ebx
+; AVX512-NEXT: popcntq {{[0-9]+}}(%rsp), %rbx
+; AVX512-NEXT: xorl %r14d, %r14d
+; AVX512-NEXT: popcntq {{[0-9]+}}(%rsp), %r14
+; AVX512-NEXT: addl %eax, %ebx
+; AVX512-NEXT: xorl %r10d, %r10d
+; AVX512-NEXT: popcntq {{[0-9]+}}(%rsp), %r10
+; AVX512-NEXT: addl %r14d, %r10d
+; AVX512-NEXT: addl %ebx, %r10d
+; AVX512-NEXT: addl %r11d, %r10d
+; AVX512-NEXT: xorl %eax, %eax
+; AVX512-NEXT: popcntq {{[0-9]+}}(%rsp), %rax
+; AVX512-NEXT: xorl %r11d, %r11d
+; AVX512-NEXT: popcntq {{[0-9]+}}(%rsp), %r11
+; AVX512-NEXT: addl %eax, %r11d
+; AVX512-NEXT: xorl %eax, %eax
+; AVX512-NEXT: popcntq %r9, %rax
+; AVX512-NEXT: popcntq %r8, %r8
+; AVX512-NEXT: addl %eax, %r8d
+; AVX512-NEXT: addl %r11d, %r8d
+; AVX512-NEXT: xorl %eax, %eax
+; AVX512-NEXT: popcntq %rcx, %rax
+; AVX512-NEXT: xorl %ecx, %ecx
+; AVX512-NEXT: popcntq %rdx, %rcx
+; AVX512-NEXT: addl %eax, %ecx
+; AVX512-NEXT: xorl %edx, %edx
+; AVX512-NEXT: popcntq %rsi, %rdx
+; AVX512-NEXT: xorl %eax, %eax
+; AVX512-NEXT: popcntq %rdi, %rax
+; AVX512-NEXT: addl %edx, %eax
+; AVX512-NEXT: addl %ecx, %eax
+; AVX512-NEXT: addl %r8d, %eax
+; AVX512-NEXT: addl %r10d, %eax
+; AVX512-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX512-NEXT: popq %rbx
+; AVX512-NEXT: popq %r14
+; AVX512-NEXT: retq
+ %cnt = call i1024 @llvm.ctpop.i1024(i1024 %a0)
+ %res = trunc i1024 %cnt to i32
+ ret i32 %res
+}
+
+define i32 @load_ctpop_i1024(ptr %p0) nounwind {
+; SSE-LABEL: load_ctpop_i1024:
+; SSE: # %bb.0:
+; SSE-NEXT: popcntq 120(%rdi), %rax
+; SSE-NEXT: popcntq 112(%rdi), %rcx
+; SSE-NEXT: popcntq 104(%rdi), %rdx
+; SSE-NEXT: popcntq 96(%rdi), %rsi
+; SSE-NEXT: addl %eax, %ecx
+; SSE-NEXT: addl %edx, %esi
+; SSE-NEXT: xorl %eax, %eax
+; SSE-NEXT: popcntq 88(%rdi), %rax
+; SSE-NEXT: addl %ecx, %esi
+; SSE-NEXT: xorl %edx, %edx
+; SSE-NEXT: popcntq 80(%rdi), %rdx
+; SSE-NEXT: addl %eax, %edx
+; SSE-NEXT: xorl %eax, %eax
+; SSE-NEXT: popcntq 72(%rdi), %rax
+; SSE-NEXT: xorl %ecx, %ecx
+; SSE-NEXT: popcntq 64(%rdi), %rcx
+; SSE-NEXT: addl %eax, %ecx
+; SSE-NEXT: addl %edx, %ecx
+; SSE-NEXT: xorl %eax, %eax
+; SSE-NEXT: popcntq 56(%rdi), %rax
+; SSE-NEXT: addl %esi, %ecx
+; SSE-NEXT: xorl %edx, %edx
+; SSE-NEXT: popcntq 48(%rdi), %rdx
+; SSE-NEXT: addl %eax, %edx
+; SSE-NEXT: xorl %eax, %eax
+; SSE-NEXT: popcntq 40(%rdi), %rax
+; SSE-NEXT: xorl %esi, %esi
+; SSE-NEXT: popcntq 32(%rdi), %rsi
+; SSE-NEXT: addl %eax, %esi
+; SSE-NEXT: addl %edx, %esi
+; SSE-NEXT: xorl %eax, %eax
+; SSE-NEXT: popcntq 24(%rdi), %rax
+; SSE-NEXT: xorl %edx, %edx
+; SSE-NEXT: popcntq 16(%rdi), %rdx
+; SSE-NEXT: popcntq 8(%rdi), %r8
+; SSE-NEXT: addl %eax, %edx
+; SSE-NEXT: xorl %eax, %eax
+; SSE-NEXT: popcntq (%rdi), %rax
+; SSE-NEXT: addl %r8d, %eax
+; SSE-NEXT: addl %edx, %eax
+; SSE-NEXT: addl %esi, %eax
+; SSE-NEXT: addl %ecx, %eax
+; SSE-NEXT: # kill: def $eax killed $eax killed $rax
+; SSE-NEXT: retq
+;
+; AVX2-LABEL: load_ctpop_i1024:
+; AVX2: # %bb.0:
+; AVX2-NEXT: popcntq 120(%rdi), %rax
+; AVX2-NEXT: popcntq 112(%rdi), %rcx
+; AVX2-NEXT: addl %eax, %ecx
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: popcntq 104(%rdi), %rax
+; AVX2-NEXT: popcntq 96(%rdi), %rdx
+; AVX2-NEXT: addl %eax, %edx
+; AVX2-NEXT: addl %ecx, %edx
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: popcntq 88(%rdi), %rax
+; AVX2-NEXT: popcntq 80(%rdi), %rsi
+; AVX2-NEXT: popcntq 72(%rdi), %r8
+; AVX2-NEXT: xorl %ecx, %ecx
+; AVX2-NEXT: popcntq 64(%rdi), %rcx
+; AVX2-NEXT: addl %eax, %esi
+; AVX2-NEXT: addl %r8d, %ecx
+; AVX2-NEXT: addl %esi, %ecx
+; AVX2-NEXT: addl %edx, %ecx
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: popcntq 56(%rdi), %rax
+; AVX2-NEXT: xorl %edx, %edx
+; AVX2-NEXT: popcntq 48(%rdi), %rdx
+; AVX2-NEXT: xorl %esi, %esi
+; AVX2-NEXT: popcntq 40(%rdi), %rsi
+; AVX2-NEXT: xorl %r8d, %r8d
+; AVX2-NEXT: popcntq 32(%rdi), %r8
+; AVX2-NEXT: addl %eax, %edx
+; AVX2-NEXT: addl %esi, %r8d
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: popcntq 24(%rdi), %rax
+; AVX2-NEXT: addl %edx, %r8d
+; AVX2-NEXT: xorl %edx, %edx
+; AVX2-NEXT: popcntq 16(%rdi), %rdx
+; AVX2-NEXT: addl %eax, %edx
+; AVX2-NEXT: xorl %esi, %esi
+; AVX2-NEXT: popcntq 8(%rdi), %rsi
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: popcntq (%rdi), %rax
+; AVX2-NEXT: addl %esi, %eax
+; AVX2-NEXT: addl %edx, %eax
+; AVX2-NEXT: addl %r8d, %eax
+; AVX2-NEXT: addl %ecx, %eax
+; AVX2-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: load_ctpop_i1024:
+; AVX512: # %bb.0:
+; AVX512-NEXT: popcntq 120(%rdi), %rax
+; AVX512-NEXT: popcntq 112(%rdi), %rcx
+; AVX512-NEXT: addl %eax, %ecx
+; AVX512-NEXT: xorl %eax, %eax
+; AVX512-NEXT: popcntq 104(%rdi), %rax
+; AVX512-NEXT: popcntq 96(%rdi), %rdx
+; AVX512-NEXT: addl %eax, %edx
+; AVX512-NEXT: addl %ecx, %edx
+; AVX512-NEXT: xorl %eax, %eax
+; AVX512-NEXT: popcntq 88(%rdi), %rax
+; AVX512-NEXT: popcntq 80(%rdi), %rsi
+; AVX512-NEXT: popcntq 72(%rdi), %r8
+; AVX512-NEXT: addl %eax, %esi
+; AVX512-NEXT: xorl %ecx, %ecx
+; AVX512-NEXT: popcntq 64(%rdi), %rcx
+; AVX512-NEXT: addl %r8d, %ecx
+; AVX512-NEXT: addl %esi, %ecx
+; AVX512-NEXT: addl %edx, %ecx
+; AVX512-NEXT: xorl %eax, %eax
+; AVX512-NEXT: popcntq 56(%rdi), %rax
+; AVX512-NEXT: xorl %edx, %edx
+; AVX512-NEXT: popcntq 48(%rdi), %rdx
+; AVX512-NEXT: xorl %esi, %esi
+; AVX512-NEXT: popcntq 40(%rdi), %rsi
+; AVX512-NEXT: addl %eax, %edx
+; AVX512-NEXT: xorl %r8d, %r8d
+; AVX512-NEXT: popcntq 32(%rdi), %r8
+; AVX512-NEXT: addl %esi, %r8d
+; AVX512-NEXT: xorl %eax, %eax
+; AVX512-NEXT: popcntq 24(%rdi), %rax
+; AVX512-NEXT: addl %edx, %r8d
+; AVX512-NEXT: xorl %edx, %edx
+; AVX512-NEXT: popcntq 16(%rdi), %rdx
+; AVX512-NEXT: addl %eax, %edx
+; AVX512-NEXT: xorl %esi, %esi
+; AVX512-NEXT: popcntq 8(%rdi), %rsi
+; AVX512-NEXT: xorl %eax, %eax
+; AVX512-NEXT: popcntq (%rdi), %rax
+; AVX512-NEXT: addl %esi, %eax
+; AVX512-NEXT: addl %edx, %eax
+; AVX512-NEXT: addl %r8d, %eax
+; AVX512-NEXT: addl %ecx, %eax
+; AVX512-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX512-NEXT: retq
+ %a0 = load i1024, ptr %p0
+ %cnt = call i1024 @llvm.ctpop.i1024(i1024 %a0)
+ %res = trunc i1024 %cnt to i32
+ ret i32 %res
+}
+
+;
+; CTLZ
+;
+
+define i32 @test_ctlz_i128(i128 %a0) nounwind {
+; SSE-LABEL: test_ctlz_i128:
+; SSE: # %bb.0:
+; SSE-NEXT: bsrq %rsi, %rcx
+; SSE-NEXT: xorl $63, %ecx
+; SSE-NEXT: movl $127, %eax
+; SSE-NEXT: bsrq %rdi, %rax
+; SSE-NEXT: xorl $63, %eax
+; SSE-NEXT: addl $64, %eax
+; SSE-NEXT: testq %rsi, %rsi
+; SSE-NEXT: cmovnel %ecx, %eax
+; SSE-NEXT: # kill: def $eax killed $eax killed $rax
+; SSE-NEXT: retq
+;
+; AVX2-LABEL: test_ctlz_i128:
+; AVX2: # %bb.0:
+; AVX2-NEXT: lzcntq %rsi, %rcx
+; AVX2-NEXT: lzcntq %rdi, %rax
+; AVX2-NEXT: addl $64, %eax
+; AVX2-NEXT: testq %rsi, %rsi
+; AVX2-NEXT: cmovnel %ecx, %eax
+; AVX2-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_ctlz_i128:
+; AVX512: # %bb.0:
+; AVX512-NEXT: lzcntq %rsi, %rcx
+; AVX512-NEXT: lzcntq %rdi, %rax
+; AVX512-NEXT: addl $64, %eax
+; AVX512-NEXT: testq %rsi, %rsi
+; AVX512-NEXT: cmovnel %ecx, %eax
+; AVX512-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX512-NEXT: retq
+ %cnt = call i128 @llvm.ctlz.i128(i128 %a0, i1 0)
+ %res = trunc i128 %cnt to i32
+ ret i32 %res
+}
+
+define i32 @load_ctlz_i128(ptr %p0) nounwind {
+; SSE-LABEL: load_ctlz_i128:
+; SSE: # %bb.0:
+; SSE-NEXT: movq 8(%rdi), %rcx
+; SSE-NEXT: bsrq %rcx, %rdx
+; SSE-NEXT: xorl $63, %edx
+; SSE-NEXT: movl $127, %eax
+; SSE-NEXT: bsrq (%rdi), %rax
+; SSE-NEXT: xorl $63, %eax
+; SSE-NEXT: addl $64, %eax
+; SSE-NEXT: testq %rcx, %rcx
+; SSE-NEXT: cmovnel %edx, %eax
+; SSE-NEXT: # kill: def $eax killed $eax killed $rax
+; SSE-NEXT: retq
+;
+; AVX2-LABEL: load_ctlz_i128:
+; AVX2: # %bb.0:
+; AVX2-NEXT: movq 8(%rdi), %rcx
+; AVX2-NEXT: lzcntq %rcx, %rdx
+; AVX2-NEXT: lzcntq (%rdi), %rax
+; AVX2-NEXT: addl $64, %eax
+; AVX2-NEXT: testq %rcx, %rcx
+; AVX2-NEXT: cmovnel %edx, %eax
+; AVX2-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: load_ctlz_i128:
+; AVX512: # %bb.0:
+; AVX512-NEXT: movq 8(%rdi), %rcx
+; AVX512-NEXT: lzcntq %rcx, %rdx
+; AVX512-NEXT: lzcntq (%rdi), %rax
+; AVX512-NEXT: addl $64, %eax
+; AVX512-NEXT: testq %rcx, %rcx
+; AVX512-NEXT: cmovnel %edx, %eax
+; AVX512-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX512-NEXT: retq
+ %a0 = load i128, ptr %p0
+ %cnt = call i128 @llvm.ctlz.i128(i128 %a0, i1 0)
+ %res = trunc i128 %cnt to i32
+ ret i32 %res
+}
+
+define i32 @test_ctlz_i256(i256 %a0) nounwind {
+; SSE-LABEL: test_ctlz_i256:
+; SSE: # %bb.0:
+; SSE-NEXT: bsrq %rcx, %rax
+; SSE-NEXT: xorl $63, %eax
+; SSE-NEXT: bsrq %rdx, %r8
+; SSE-NEXT: xorl $63, %r8d
+; SSE-NEXT: orl $64, %r8d
+; SSE-NEXT: testq %rcx, %rcx
+; SSE-NEXT: cmovnel %eax, %r8d
+; SSE-NEXT: bsrq %rsi, %r9
+; SSE-NEXT: xorl $63, %r9d
+; SSE-NEXT: movl $127, %eax
+; SSE-NEXT: bsrq %rdi, %rax
+; SSE-NEXT: xorl $63, %eax
+; SSE-NEXT: addl $64, %eax
+; SSE-NEXT: testq %rsi, %rsi
+; SSE-NEXT: cmovnel %r9d, %eax
+; SSE-NEXT: subl $-128, %eax
+; SSE-NEXT: orq %rcx, %rdx
+; SSE-NEXT: cmovnel %r8d, %eax
+; SSE-NEXT: # kill: def $eax killed $eax killed $rax
+; SSE-NEXT: retq
+;
+; AVX2-LABEL: test_ctlz_i256:
+; AVX2: # %bb.0:
+; AVX2-NEXT: lzcntq %rcx, %rax
+; AVX2-NEXT: lzcntq %rdx, %r8
+; AVX2-NEXT: addl $64, %r8d
+; AVX2-NEXT: testq %rcx, %rcx
+; AVX2-NEXT: cmovnel %eax, %r8d
+; AVX2-NEXT: lzcntq %rsi, %r9
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: lzcntq %rdi, %rax
+; AVX2-NEXT: addl $64, %eax
+; AVX2-NEXT: testq %rsi, %rsi
+; AVX2-NEXT: cmovnel %r9d, %eax
+; AVX2-NEXT: subl $-128, %eax
+; AVX2-NEXT: orq %rcx, %rdx
+; AVX2-NEXT: cmovnel %r8d, %eax
+; AVX2-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_ctlz_i256:
+; AVX512: # %bb.0:
+; AVX512-NEXT: lzcntq %rcx, %rax
+; AVX512-NEXT: lzcntq %rdx, %r8
+; AVX512-NEXT: addl $64, %r8d
+; AVX512-NEXT: testq %rcx, %rcx
+; AVX512-NEXT: cmovnel %eax, %r8d
+; AVX512-NEXT: lzcntq %rsi, %r9
+; AVX512-NEXT: lzcntq %rdi, %rax
+; AVX512-NEXT: addl $64, %eax
+; AVX512-NEXT: testq %rsi, %rsi
+; AVX512-NEXT: cmovnel %r9d, %eax
+; AVX512-NEXT: subl $-128, %eax
+; AVX512-NEXT: orq %rcx, %rdx
+; AVX512-NEXT: cmovnel %r8d, %eax
+; AVX512-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX512-NEXT: retq
+ %cnt = call i256 @llvm.ctlz.i256(i256 %a0, i1 0)
+ %res = trunc i256 %cnt to i32
+ ret i32 %res
+}
+
+define i32 @load_ctlz_i256(ptr %p0) nounwind {
+; SSE-LABEL: load_ctlz_i256:
+; SSE: # %bb.0:
+; SSE-NEXT: movq 16(%rdi), %rcx
+; SSE-NEXT: movq 24(%rdi), %rdx
+; SSE-NEXT: bsrq %rdx, %rax
+; SSE-NEXT: xorl $63, %eax
+; SSE-NEXT: bsrq %rcx, %rsi
+; SSE-NEXT: xorl $63, %esi
+; SSE-NEXT: orl $64, %esi
+; SSE-NEXT: testq %rdx, %rdx
+; SSE-NEXT: cmovnel %eax, %esi
+; SSE-NEXT: movq 8(%rdi), %r8
+; SSE-NEXT: bsrq %r8, %r9
+; SSE-NEXT: xorl $63, %r9d
+; SSE-NEXT: movl $127, %eax
+; SSE-NEXT: bsrq (%rdi), %rax
+; SSE-NEXT: xorl $63, %eax
+; SSE-NEXT: addl $64, %eax
+; SSE-NEXT: testq %r8, %r8
+; SSE-NEXT: cmovnel %r9d, %eax
+; SSE-NEXT: subl $-128, %eax
+; SSE-NEXT: orq %rdx, %rcx
+; SSE-NEXT: cmovnel %esi, %eax
+; SSE-NEXT: # kill: def $eax killed $eax killed $rax
+; SSE-NEXT: retq
+;
+; AVX2-LABEL: load_ctlz_i256:
+; AVX2: # %bb.0:
+; AVX2-NEXT: movq 16(%rdi), %rcx
+; AVX2-NEXT: movq 24(%rdi), %rdx
+; AVX2-NEXT: lzcntq %rdx, %rax
+; AVX2-NEXT: lzcntq %rcx, %rsi
+; AVX2-NEXT: addl $64, %esi
+; AVX2-NEXT: testq %rdx, %rdx
+; AVX2-NEXT: cmovnel %eax, %esi
+; AVX2-NEXT: movq 8(%rdi), %r8
+; AVX2-NEXT: lzcntq %r8, %r9
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: lzcntq (%rdi), %rax
+; AVX2-NEXT: addl $64, %eax
+; AVX2-NEXT: testq %r8, %r8
+; AVX2-NEXT: cmovnel %r9d, %eax
+; AVX2-NEXT: subl $-128, %eax
+; AVX2-NEXT: orq %rdx, %rcx
+; AVX2-NEXT: cmovnel %esi, %eax
+; AVX2-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: load_ctlz_i256:
+; AVX512: # %bb.0:
+; AVX512-NEXT: movq 8(%rdi), %rcx
+; AVX512-NEXT: movq 16(%rdi), %rdx
+; AVX512-NEXT: movq 24(%rdi), %rsi
+; AVX512-NEXT: lzcntq %rsi, %rax
+; AVX512-NEXT: lzcntq %rdx, %r8
+; AVX512-NEXT: addl $64, %r8d
+; AVX512-NEXT: testq %rsi, %rsi
+; AVX512-NEXT: cmovnel %eax, %r8d
+; AVX512-NEXT: lzcntq %rcx, %r9
+; AVX512-NEXT: lzcntq (%rdi), %rax
+; AVX512-NEXT: addl $64, %eax
+; AVX512-NEXT: testq %rcx, %rcx
+; AVX512-NEXT: cmovnel %r9d, %eax
+; AVX512-NEXT: subl $-128, %eax
+; AVX512-NEXT: orq %rsi, %rdx
+; AVX512-NEXT: cmovnel %r8d, %eax
+; AVX512-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX512-NEXT: retq
+ %a0 = load i256, ptr %p0
+ %cnt = call i256 @llvm.ctlz.i256(i256 %a0, i1 0)
+ %res = trunc i256 %cnt to i32
+ ret i32 %res
+}
+
+define i32 @test_ctlz_i512(i512 %a0) nounwind {
+; SSE-LABEL: test_ctlz_i512:
+; SSE: # %bb.0:
+; SSE-NEXT: pushq %r15
+; SSE-NEXT: pushq %r14
+; SSE-NEXT: pushq %rbx
+; SSE-NEXT: movq {{[0-9]+}}(%rsp), %r10
+; SSE-NEXT: movq {{[0-9]+}}(%rsp), %r11
+; SSE-NEXT: bsrq %r11, %rax
+; SSE-NEXT: xorl $63, %eax
+; SSE-NEXT: bsrq %r10, %r14
+; SSE-NEXT: xorl $63, %r14d
+; SSE-NEXT: orl $64, %r14d
+; SSE-NEXT: testq %r11, %r11
+; SSE-NEXT: cmovnel %eax, %r14d
+; SSE-NEXT: bsrq %r9, %rax
+; SSE-NEXT: xorl $63, %eax
+; SSE-NEXT: bsrq %r8, %rbx
+; SSE-NEXT: xorl $63, %ebx
+; SSE-NEXT: orl $64, %ebx
+; SSE-NEXT: testq %r9, %r9
+; SSE-NEXT: cmovnel %eax, %ebx
+; SSE-NEXT: subl $-128, %ebx
+; SSE-NEXT: movq %r10, %rax
+; SSE-NEXT: orq %r11, %rax
+; SSE-NEXT: cmovnel %r14d, %ebx
+; SSE-NEXT: bsrq %rcx, %rax
+; SSE-NEXT: xorl $63, %eax
+; SSE-NEXT: bsrq %rdx, %r14
+; SSE-NEXT: xorl $63, %r14d
+; SSE-NEXT: orl $64, %r14d
+; SSE-NEXT: testq %rcx, %rcx
+; SSE-NEXT: cmovnel %eax, %r14d
+; SSE-NEXT: bsrq %rsi, %r15
+; SSE-NEXT: xorl $63, %r15d
+; SSE-NEXT: movl $127, %eax
+; SSE-NEXT: bsrq %rdi, %rax
+; SSE-NEXT: xorl $63, %eax
+; SSE-NEXT: addl $64, %eax
+; SSE-NEXT: testq %rsi, %rsi
+; SSE-NEXT: cmovnel %r15d, %eax
+; SSE-NEXT: subl $-128, %eax
+; SSE-NEXT: orq %rcx, %rdx
+; SSE-NEXT: cmovnel %r14d, %eax
+; SSE-NEXT: addl $256, %eax # imm = 0x100
+; SSE-NEXT: orq %r11, %r9
+; SSE-NEXT: orq %r10, %r8
+; SSE-NEXT: orq %r9, %r8
+; SSE-NEXT: cmovnel %ebx, %eax
+; SSE-NEXT: # kill: def $eax killed $eax killed $rax
+; SSE-NEXT: popq %rbx
+; SSE-NEXT: popq %r14
+; SSE-NEXT: popq %r15
+; SSE-NEXT: retq
+;
+; AVX2-LABEL: test_ctlz_i512:
+; AVX2: # %bb.0:
+; AVX2-NEXT: pushq %r15
+; AVX2-NEXT: pushq %r14
+; AVX2-NEXT: pushq %rbx
+; AVX2-NEXT: movq {{[0-9]+}}(%rsp), %r10
+; AVX2-NEXT: movq {{[0-9]+}}(%rsp), %r11
+; AVX2-NEXT: lzcntq %r11, %rax
+; AVX2-NEXT: xorl %r14d, %r14d
+; AVX2-NEXT: lzcntq %r10, %r14
+; AVX2-NEXT: addl $64, %r14d
+; AVX2-NEXT: testq %r11, %r11
+; AVX2-NEXT: cmovnel %eax, %r14d
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: lzcntq %r9, %rax
+; AVX2-NEXT: xorl %ebx, %ebx
+; AVX2-NEXT: lzcntq %r8, %rbx
+; AVX2-NEXT: addl $64, %ebx
+; AVX2-NEXT: testq %r9, %r9
+; AVX2-NEXT: cmovnel %eax, %ebx
+; AVX2-NEXT: subl $-128, %ebx
+; AVX2-NEXT: movq %r10, %rax
+; AVX2-NEXT: orq %r11, %rax
+; AVX2-NEXT: cmovnel %r14d, %ebx
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: lzcntq %rcx, %rax
+; AVX2-NEXT: xorl %r14d, %r14d
+; AVX2-NEXT: lzcntq %rdx, %r14
+; AVX2-NEXT: addl $64, %r14d
+; AVX2-NEXT: testq %rcx, %rcx
+; AVX2-NEXT: cmovnel %eax, %r14d
+; AVX2-NEXT: xorl %r15d, %r15d
+; AVX2-NEXT: lzcntq %rsi, %r15
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: lzcntq %rdi, %rax
+; AVX2-NEXT: addl $64, %eax
+; AVX2-NEXT: testq %rsi, %rsi
+; AVX2-NEXT: cmovnel %r15d, %eax
+; AVX2-NEXT: subl $-128, %eax
+; AVX2-NEXT: orq %rcx, %rdx
+; AVX2-NEXT: cmovnel %r14d, %eax
+; AVX2-NEXT: addl $256, %eax # imm = 0x100
+; AVX2-NEXT: orq %r11, %r9
+; AVX2-NEXT: orq %r10, %r8
+; AVX2-NEXT: orq %r9, %r8
+; AVX2-NEXT: cmovnel %ebx, %eax
+; AVX2-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX2-NEXT: popq %rbx
+; AVX2-NEXT: popq %r14
+; AVX2-NEXT: popq %r15
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_ctlz_i512:
+; AVX512: # %bb.0:
+; AVX512-NEXT: pushq %r15
+; AVX512-NEXT: pushq %r14
+; AVX512-NEXT: pushq %rbx
+; AVX512-NEXT: movq {{[0-9]+}}(%rsp), %r10
+; AVX512-NEXT: movq {{[0-9]+}}(%rsp), %r11
+; AVX512-NEXT: lzcntq %r11, %rax
+; AVX512-NEXT: lzcntq %r10, %r14
+; AVX512-NEXT: addl $64, %r14d
+; AVX512-NEXT: testq %r11, %r11
+; AVX512-NEXT: cmovnel %eax, %r14d
+; AVX512-NEXT: lzcntq %r9, %rax
+; AVX512-NEXT: lzcntq %r8, %rbx
+; AVX512-NEXT: addl $64, %ebx
+; AVX512-NEXT: testq %r9, %r9
+; AVX512-NEXT: cmovnel %eax, %ebx
+; AVX512-NEXT: subl $-128, %ebx
+; AVX512-NEXT: movq %r10, %rax
+; AVX512-NEXT: orq %r11, %rax
+; AVX512-NEXT: cmovnel %r14d, %ebx
+; AVX512-NEXT: lzcntq %rcx, %rax
+; AVX512-NEXT: lzcntq %rdx, %r14
+; AVX512-NEXT: addl $64, %r14d
+; AVX512-NEXT: testq %rcx, %rcx
+; AVX512-NEXT: cmovnel %eax, %r14d
+; AVX512-NEXT: lzcntq %rsi, %r15
+; AVX512-NEXT: lzcntq %rdi, %rax
+; AVX512-NEXT: addl $64, %eax
+; AVX512-NEXT: testq %rsi, %rsi
+; AVX512-NEXT: cmovnel %r15d, %eax
+; AVX512-NEXT: subl $-128, %eax
+; AVX512-NEXT: orq %rcx, %rdx
+; AVX512-NEXT: cmovnel %r14d, %eax
+; AVX512-NEXT: addl $256, %eax # imm = 0x100
+; AVX512-NEXT: orq %r11, %r9
+; AVX512-NEXT: orq %r10, %r8
+; AVX512-NEXT: orq %r9, %r8
+; AVX512-NEXT: cmovnel %ebx, %eax
+; AVX512-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX512-NEXT: popq %rbx
+; AVX512-NEXT: popq %r14
+; AVX512-NEXT: popq %r15
+; AVX512-NEXT: retq
+ %cnt = call i512 @llvm.ctlz.i512(i512 %a0, i1 0)
+ %res = trunc i512 %cnt to i32
+ ret i32 %res
+}
+
+define i32 @load_ctlz_i512(ptr %p0) nounwind {
+; SSE-LABEL: load_ctlz_i512:
+; SSE: # %bb.0:
+; SSE-NEXT: pushq %r15
+; SSE-NEXT: pushq %r14
+; SSE-NEXT: pushq %rbx
+; SSE-NEXT: movq 8(%rdi), %r10
+; SSE-NEXT: movq 16(%rdi), %r9
+; SSE-NEXT: movq 32(%rdi), %rcx
+; SSE-NEXT: movq 40(%rdi), %rdx
+; SSE-NEXT: movq 48(%rdi), %rsi
+; SSE-NEXT: movq 56(%rdi), %r8
+; SSE-NEXT: bsrq %r8, %rax
+; SSE-NEXT: xorl $63, %eax
+; SSE-NEXT: bsrq %rsi, %r14
+; SSE-NEXT: xorl $63, %r14d
+; SSE-NEXT: orl $64, %r14d
+; SSE-NEXT: testq %r8, %r8
+; SSE-NEXT: cmovnel %eax, %r14d
+; SSE-NEXT: bsrq %rdx, %rax
+; SSE-NEXT: xorl $63, %eax
+; SSE-NEXT: bsrq %rcx, %r11
+; SSE-NEXT: xorl $63, %r11d
+; SSE-NEXT: orl $64, %r11d
+; SSE-NEXT: testq %rdx, %rdx
+; SSE-NEXT: cmovnel %eax, %r11d
+; SSE-NEXT: movq 24(%rdi), %rbx
+; SSE-NEXT: subl $-128, %r11d
+; SSE-NEXT: movq %rsi, %rax
+; SSE-NEXT: orq %r8, %rax
+; SSE-NEXT: cmovnel %r14d, %r11d
+; SSE-NEXT: bsrq %rbx, %rax
+; SSE-NEXT: xorl $63, %eax
+; SSE-NEXT: bsrq %r9, %r14
+; SSE-NEXT: xorl $63, %r14d
+; SSE-NEXT: orl $64, %r14d
+; SSE-NEXT: testq %rbx, %rbx
+; SSE-NEXT: cmovnel %eax, %r14d
+; SSE-NEXT: bsrq %r10, %r15
+; SSE-NEXT: xorl $63, %r15d
+; SSE-NEXT: movl $127, %eax
+; SSE-NEXT: bsrq (%rdi), %rax
+; SSE-NEXT: xorl $63, %eax
+; SSE-NEXT: addl $64, %eax
+; SSE-NEXT: testq %r10, %r10
+; SSE-NEXT: cmovnel %r15d, %eax
+; SSE-NEXT: subl $-128, %eax
+; SSE-NEXT: orq %rbx, %r9
+; SSE-NEXT: cmovnel %r14d, %eax
+; SSE-NEXT: addl $256, %eax # imm = 0x100
+; SSE-NEXT: orq %r8, %rdx
+; SSE-NEXT: orq %rsi, %rcx
+; SSE-NEXT: orq %rdx, %rcx
+; SSE-NEXT: cmovnel %r11d, %eax
+; SSE-NEXT: # kill: def $eax killed $eax killed $rax
+; SSE-NEXT: popq %rbx
+; SSE-NEXT: popq %r14
+; SSE-NEXT: popq %r15
+; SSE-NEXT: retq
+;
+; AVX2-LABEL: load_ctlz_i512:
+; AVX2: # %bb.0:
+; AVX2-NEXT: pushq %r15
+; AVX2-NEXT: pushq %r14
+; AVX2-NEXT: pushq %rbx
+; AVX2-NEXT: movq 8(%rdi), %r10
+; AVX2-NEXT: movq 16(%rdi), %r9
+; AVX2-NEXT: movq 32(%rdi), %rcx
+; AVX2-NEXT: movq 40(%rdi), %rdx
+; AVX2-NEXT: movq 48(%rdi), %rsi
+; AVX2-NEXT: movq 56(%rdi), %r8
+; AVX2-NEXT: lzcntq %r8, %rax
+; AVX2-NEXT: xorl %ebx, %ebx
+; AVX2-NEXT: lzcntq %rsi, %rbx
+; AVX2-NEXT: addl $64, %ebx
+; AVX2-NEXT: testq %r8, %r8
+; AVX2-NEXT: cmovnel %eax, %ebx
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: lzcntq %rdx, %rax
+; AVX2-NEXT: lzcntq %rcx, %r11
+; AVX2-NEXT: addl $64, %r11d
+; AVX2-NEXT: testq %rdx, %rdx
+; AVX2-NEXT: cmovnel %eax, %r11d
+; AVX2-NEXT: subl $-128, %r11d
+; AVX2-NEXT: movq %rsi, %rax
+; AVX2-NEXT: orq %r8, %rax
+; AVX2-NEXT: cmovnel %ebx, %r11d
+; AVX2-NEXT: movq 24(%rdi), %rbx
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: lzcntq %rbx, %rax
+; AVX2-NEXT: xorl %r14d, %r14d
+; AVX2-NEXT: lzcntq %r9, %r14
+; AVX2-NEXT: addl $64, %r14d
+; AVX2-NEXT: testq %rbx, %rbx
+; AVX2-NEXT: cmovnel %eax, %r14d
+; AVX2-NEXT: xorl %r15d, %r15d
+; AVX2-NEXT: lzcntq %r10, %r15
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: lzcntq (%rdi), %rax
+; AVX2-NEXT: addl $64, %eax
+; AVX2-NEXT: testq %r10, %r10
+; AVX2-NEXT: cmovnel %r15d, %eax
+; AVX2-NEXT: subl $-128, %eax
+; AVX2-NEXT: orq %rbx, %r9
+; AVX2-NEXT: cmovnel %r14d, %eax
+; AVX2-NEXT: addl $256, %eax # imm = 0x100
+; AVX2-NEXT: orq %r8, %rdx
+; AVX2-NEXT: orq %rsi, %rcx
+; AVX2-NEXT: orq %rdx, %rcx
+; AVX2-NEXT: cmovnel %r11d, %eax
+; AVX2-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX2-NEXT: popq %rbx
+; AVX2-NEXT: popq %r14
+; AVX2-NEXT: popq %r15
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: load_ctlz_i512:
+; AVX512: # %bb.0:
+; AVX512-NEXT: pushq %r14
+; AVX512-NEXT: pushq %rbx
+; AVX512-NEXT: movq 8(%rdi), %r11
+; AVX512-NEXT: movq 16(%rdi), %r9
+; AVX512-NEXT: movq 24(%rdi), %r10
+; AVX512-NEXT: movq 32(%rdi), %rcx
+; AVX512-NEXT: movq 40(%rdi), %rdx
+; AVX512-NEXT: movq 48(%rdi), %rsi
+; AVX512-NEXT: movq 56(%rdi), %r8
+; AVX512-NEXT: lzcntq %r8, %rax
+; AVX512-NEXT: lzcntq %rsi, %r14
+; AVX512-NEXT: addl $64, %r14d
+; AVX512-NEXT: testq %r8, %r8
+; AVX512-NEXT: cmovnel %eax, %r14d
+; AVX512-NEXT: lzcntq %rdx, %rax
+; AVX512-NEXT: lzcntq %rcx, %rbx
+; AVX512-NEXT: addl $64, %ebx
+; AVX512-NEXT: testq %rdx, %rdx
+; AVX512-NEXT: cmovnel %eax, %ebx
+; AVX512-NEXT: subl $-128, %ebx
+; AVX512-NEXT: movq %rsi, %rax
+; AVX512-NEXT: orq %r8, %rax
+; AVX512-NEXT: cmovnel %r14d, %ebx
+; AVX512-NEXT: lzcntq %r10, %rax
+; AVX512-NEXT: lzcntq %r9, %r14
+; AVX512-NEXT: addl $64, %r14d
+; AVX512-NEXT: testq %r10, %r10
+; AVX512-NEXT: cmovnel %eax, %r14d
+; AVX512-NEXT: lzcntq (%rdi), %rax
+; AVX512-NEXT: lzcntq %r11, %rdi
+; AVX512-NEXT: addl $64, %eax
+; AVX512-NEXT: testq %r11, %r11
+; AVX512-NEXT: cmovnel %edi, %eax
+; AVX512-NEXT: subl $-128, %eax
+; AVX512-NEXT: orq %r10, %r9
+; AVX512-NEXT: cmovnel %r14d, %eax
+; AVX512-NEXT: addl $256, %eax # imm = 0x100
+; AVX512-NEXT: orq %r8, %rdx
+; AVX512-NEXT: orq %rsi, %rcx
+; AVX512-NEXT: orq %rdx, %rcx
+; AVX512-NEXT: cmovnel %ebx, %eax
+; AVX512-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX512-NEXT: popq %rbx
+; AVX512-NEXT: popq %r14
+; AVX512-NEXT: retq
+ %a0 = load i512, ptr %p0
+ %cnt = call i512 @llvm.ctlz.i512(i512 %a0, i1 0)
+ %res = trunc i512 %cnt to i32
+ ret i32 %res
+}
+
+define i32 @test_ctlz_i1024(i1024 %a0) nounwind {
+; SSE-LABEL: test_ctlz_i1024:
+; SSE: # %bb.0:
+; SSE-NEXT: pushq %rbp
+; SSE-NEXT: pushq %r15
+; SSE-NEXT: pushq %r14
+; SSE-NEXT: pushq %r13
+; SSE-NEXT: pushq %r12
+; SSE-NEXT: pushq %rbx
+; SSE-NEXT: movq %r9, %r11
+; SSE-NEXT: movq %r8, %r9
+; SSE-NEXT: movq %rcx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; SSE-NEXT: movq %rdx, %r12
+; SSE-NEXT: movq {{[0-9]+}}(%rsp), %r10
+; SSE-NEXT: movq {{[0-9]+}}(%rsp), %rbx
+; SSE-NEXT: movq {{[0-9]+}}(%rsp), %r14
+; SSE-NEXT: movq {{[0-9]+}}(%rsp), %r15
+; SSE-NEXT: movq {{[0-9]+}}(%rsp), %r8
+; SSE-NEXT: bsrq %r8, %rax
+; SSE-NEXT: xorl $63, %eax
+; SSE-NEXT: bsrq %r15, %rcx
+; SSE-NEXT: xorl $63, %ecx
+; SSE-NEXT: orl $64, %ecx
+; SSE-NEXT: testq %r8, %r8
+; SSE-NEXT: cmovnel %eax, %ecx
+; SSE-NEXT: bsrq %r14, %rdx
+; SSE-NEXT: xorl $63, %edx
+; SSE-NEXT: bsrq {{[0-9]+}}(%rsp), %rax
+; SSE-NEXT: xorl $63, %eax
+; SSE-NEXT: orl $64, %eax
+; SSE-NEXT: testq %r14, %r14
+; SSE-NEXT: cmovnel %edx, %eax
+; SSE-NEXT: movq {{[0-9]+}}(%rsp), %r13
+; SSE-NEXT: subl $-128, %eax
+; SSE-NEXT: movq %r15, %rdx
+; SSE-NEXT: orq %r8, %rdx
+; SSE-NEXT: movq %r8, %r14
+; SSE-NEXT: cmovnel %ecx, %eax
+; SSE-NEXT: bsrq %r13, %rcx
+; SSE-NEXT: xorl $63, %ecx
+; SSE-NEXT: bsrq %rbx, %rdx
+; SSE-NEXT: xorl $63, %edx
+; SSE-NEXT: orl $64, %edx
+; SSE-NEXT: testq %r13, %r13
+; SSE-NEXT: cmovnel %ecx, %edx
+; SSE-NEXT: bsrq %r10, %rcx
+; SSE-NEXT: xorl $63, %ecx
+; SSE-NEXT: movq {{[0-9]+}}(%rsp), %r8
+; SSE-NEXT: bsrq %r8, %rbp
+; SSE-NEXT: xorl $63, %ebp
+; SSE-NEXT: orl $64, %ebp
+; SSE-NEXT: testq %r10, %r10
+; SSE-NEXT: cmovnel %ecx, %ebp
+; SSE-NEXT: subl $-128, %ebp
+; SSE-NEXT: movq %rbx, %rcx
+; SSE-NEXT: orq %r13, %rcx
+; SSE-NEXT: cmovnel %edx, %ebp
+; SSE-NEXT: addl $256, %ebp # imm = 0x100
+; SSE-NEXT: movq {{[0-9]+}}(%rsp), %rcx
+; SSE-NEXT: orq %r14, %rcx
+; SSE-NEXT: movq {{[0-9]+}}(%rsp), %rdx
+; SSE-NEXT: orq %r15, %rdx
+; SSE-NEXT: orq %rcx, %rdx
+; SSE-NEXT: cmovnel %eax, %ebp
+; SSE-NEXT: movq {{[0-9]+}}(%rsp), %r14
+; SSE-NEXT: bsrq %r14, %rax
+; SSE-NEXT: xorl $63, %eax
+; SSE-NEXT: movq {{[0-9]+}}(%rsp), %r15
+; SSE-NEXT: bsrq %r15, %rcx
+; SSE-NEXT: xorl $63, %ecx
+; SSE-NEXT: orl $64, %ecx
+; SSE-NEXT: testq %r14, %r14
+; SSE-NEXT: cmovnel %eax, %ecx
+; SSE-NEXT: bsrq %r11, %rax
+; SSE-NEXT: xorl $63, %eax
+; SSE-NEXT: bsrq %r9, %rdx
+; SSE-NEXT: xorl $63, %edx
+; SSE-NEXT: orl $64, %edx
+; SSE-NEXT: testq %r11, %r11
+; SSE-NEXT: cmovnel %eax, %edx
+; SSE-NEXT: subl $-128, %edx
+; SSE-NEXT: movq %r15, %rax
+; SSE-NEXT: orq %r14, %rax
+; SSE-NEXT: cmovnel %ecx, %edx
+; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload
+; SSE-NEXT: bsrq %r15, %rax
+; SSE-NEXT: xorl $63, %eax
+; SSE-NEXT: bsrq %r12, %rcx
+; SSE-NEXT: xorl $63, %ecx
+; SSE-NEXT: orl $64, %ecx
+; SSE-NEXT: testq %r15, %r15
+; SSE-NEXT: cmovnel %eax, %ecx
+; SSE-NEXT: movl $127, %eax
+; SSE-NEXT: bsrq %rdi, %rax
+; SSE-NEXT: bsrq %rsi, %rdi
+; SSE-NEXT: xorl $63, %edi
+; SSE-NEXT: xorl $63, %eax
+; SSE-NEXT: addl $64, %eax
+; SSE-NEXT: testq %rsi, %rsi
+; SSE-NEXT: cmovnel %edi, %eax
+; SSE-NEXT: subl $-128, %eax
+; SSE-NEXT: orq %r15, %r12
+; SSE-NEXT: cmovnel %ecx, %eax
+; SSE-NEXT: orq %r14, %r11
+; SSE-NEXT: orq {{[0-9]+}}(%rsp), %r9
+; SSE-NEXT: addl $256, %eax # imm = 0x100
+; SSE-NEXT: orq %r11, %r9
+; SSE-NEXT: cmovnel %edx, %eax
+; SSE-NEXT: orq {{[0-9]+}}(%rsp), %r13
+; SSE-NEXT: orq {{[0-9]+}}(%rsp), %r10
+; SSE-NEXT: orq %r13, %r10
+; SSE-NEXT: orq {{[0-9]+}}(%rsp), %rbx
+; SSE-NEXT: orq {{[0-9]+}}(%rsp), %r8
+; SSE-NEXT: orq %rbx, %r8
+; SSE-NEXT: addl $512, %eax # imm = 0x200
+; SSE-NEXT: orq %r10, %r8
+; SSE-NEXT: cmovnel %ebp, %eax
+; SSE-NEXT: # kill: def $eax killed $eax killed $rax
+; SSE-NEXT: popq %rbx
+; SSE-NEXT: popq %r12
+; SSE-NEXT: popq %r13
+; SSE-NEXT: popq %r14
+; SSE-NEXT: popq %r15
+; SSE-NEXT: popq %rbp
+; SSE-NEXT: retq
+;
+; AVX2-LABEL: test_ctlz_i1024:
+; AVX2: # %bb.0:
+; AVX2-NEXT: pushq %rbp
+; AVX2-NEXT: pushq %r15
+; AVX2-NEXT: pushq %r14
+; AVX2-NEXT: pushq %r13
+; AVX2-NEXT: pushq %r12
+; AVX2-NEXT: pushq %rbx
+; AVX2-NEXT: movq %r9, %r14
+; AVX2-NEXT: movq %r8, %r11
+; AVX2-NEXT: movq %rcx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX2-NEXT: movq %rsi, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX2-NEXT: movq %rdi, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX2-NEXT: movq {{[0-9]+}}(%rsp), %r15
+; AVX2-NEXT: movq {{[0-9]+}}(%rsp), %rbx
+; AVX2-NEXT: movq {{[0-9]+}}(%rsp), %rax
+; AVX2-NEXT: movq {{[0-9]+}}(%rsp), %r10
+; AVX2-NEXT: movq {{[0-9]+}}(%rsp), %r8
+; AVX2-NEXT: movq {{[0-9]+}}(%rsp), %r12
+; AVX2-NEXT: xorl %ecx, %ecx
+; AVX2-NEXT: lzcntq %r12, %rcx
+; AVX2-NEXT: xorl %r9d, %r9d
+; AVX2-NEXT: lzcntq %r8, %r9
+; AVX2-NEXT: addl $64, %r9d
+; AVX2-NEXT: testq %r12, %r12
+; AVX2-NEXT: cmovnel %ecx, %r9d
+; AVX2-NEXT: xorl %esi, %esi
+; AVX2-NEXT: lzcntq %r10, %rsi
+; AVX2-NEXT: xorl %ecx, %ecx
+; AVX2-NEXT: lzcntq %rax, %rcx
+; AVX2-NEXT: addl $64, %ecx
+; AVX2-NEXT: testq %r10, %r10
+; AVX2-NEXT: cmovnel %esi, %ecx
+; AVX2-NEXT: subl $-128, %ecx
+; AVX2-NEXT: movq %r8, %rsi
+; AVX2-NEXT: orq %r12, %rsi
+; AVX2-NEXT: cmovnel %r9d, %ecx
+; AVX2-NEXT: xorl %edi, %edi
+; AVX2-NEXT: lzcntq %rbx, %rdi
+; AVX2-NEXT: xorl %esi, %esi
+; AVX2-NEXT: lzcntq %r15, %rsi
+; AVX2-NEXT: addl $64, %esi
+; AVX2-NEXT: testq %rbx, %rbx
+; AVX2-NEXT: cmovnel %edi, %esi
+; AVX2-NEXT: movq {{[0-9]+}}(%rsp), %r13
+; AVX2-NEXT: xorl %ebp, %ebp
+; AVX2-NEXT: lzcntq %r13, %rbp
+; AVX2-NEXT: addl $64, %ebp
+; AVX2-NEXT: movq {{[0-9]+}}(%rsp), %r9
+; AVX2-NEXT: xorl %edi, %edi
+; AVX2-NEXT: lzcntq %r9, %rdi
+; AVX2-NEXT: testq %r9, %r9
+; AVX2-NEXT: cmovnel %edi, %ebp
+; AVX2-NEXT: subl $-128, %ebp
+; AVX2-NEXT: movq %r15, %rdi
+; AVX2-NEXT: orq %rbx, %rdi
+; AVX2-NEXT: cmovnel %esi, %ebp
+; AVX2-NEXT: addl $256, %ebp # imm = 0x100
+; AVX2-NEXT: movq %r10, %rdi
+; AVX2-NEXT: orq %r12, %rdi
+; AVX2-NEXT: movq %rax, %rsi
+; AVX2-NEXT: orq %r8, %rsi
+; AVX2-NEXT: orq %rdi, %rsi
+; AVX2-NEXT: cmovnel %ecx, %ebp
+; AVX2-NEXT: movq {{[0-9]+}}(%rsp), %rdi
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: lzcntq %rdi, %rax
+; AVX2-NEXT: addl $64, %eax
+; AVX2-NEXT: movq {{[0-9]+}}(%rsp), %r12
+; AVX2-NEXT: xorl %ecx, %ecx
+; AVX2-NEXT: lzcntq %r12, %rcx
+; AVX2-NEXT: testq %r12, %r12
+; AVX2-NEXT: cmovnel %ecx, %eax
+; AVX2-NEXT: xorl %ecx, %ecx
+; AVX2-NEXT: lzcntq %r11, %rcx
+; AVX2-NEXT: addl $64, %ecx
+; AVX2-NEXT: xorl %esi, %esi
+; AVX2-NEXT: lzcntq %r14, %rsi
+; AVX2-NEXT: testq %r14, %r14
+; AVX2-NEXT: cmovnel %esi, %ecx
+; AVX2-NEXT: subl $-128, %ecx
+; AVX2-NEXT: movq %rdi, %rsi
+; AVX2-NEXT: orq %r12, %rsi
+; AVX2-NEXT: cmovnel %eax, %ecx
+; AVX2-NEXT: movq %rdx, %rdi
+; AVX2-NEXT: lzcntq %rdx, %rdx
+; AVX2-NEXT: addl $64, %edx
+; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r10 # 8-byte Reload
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: lzcntq %r10, %rax
+; AVX2-NEXT: testq %r10, %r10
+; AVX2-NEXT: cmovnel %eax, %edx
+; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload
+; AVX2-NEXT: lzcntq %rax, %rax
+; AVX2-NEXT: addl $64, %eax
+; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rsi # 8-byte Reload
+; AVX2-NEXT: lzcntq %rsi, %r8
+; AVX2-NEXT: testq %rsi, %rsi
+; AVX2-NEXT: cmovnel %r8d, %eax
+; AVX2-NEXT: subl $-128, %eax
+; AVX2-NEXT: orq %r10, %rdi
+; AVX2-NEXT: cmovnel %edx, %eax
+; AVX2-NEXT: orq %r12, %r14
+; AVX2-NEXT: orq {{[0-9]+}}(%rsp), %r11
+; AVX2-NEXT: addl $256, %eax # imm = 0x100
+; AVX2-NEXT: orq %r14, %r11
+; AVX2-NEXT: cmovnel %ecx, %eax
+; AVX2-NEXT: orq {{[0-9]+}}(%rsp), %rbx
+; AVX2-NEXT: orq {{[0-9]+}}(%rsp), %r9
+; AVX2-NEXT: orq %rbx, %r9
+; AVX2-NEXT: orq {{[0-9]+}}(%rsp), %r15
+; AVX2-NEXT: orq {{[0-9]+}}(%rsp), %r13
+; AVX2-NEXT: orq %r15, %r13
+; AVX2-NEXT: addl $512, %eax # imm = 0x200
+; AVX2-NEXT: orq %r9, %r13
+; AVX2-NEXT: cmovnel %ebp, %eax
+; AVX2-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX2-NEXT: popq %rbx
+; AVX2-NEXT: popq %r12
+; AVX2-NEXT: popq %r13
+; AVX2-NEXT: popq %r14
+; AVX2-NEXT: popq %r15
+; AVX2-NEXT: popq %rbp
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_ctlz_i1024:
+; AVX512: # %bb.0:
+; AVX512-NEXT: pushq %rbp
+; AVX512-NEXT: pushq %r15
+; AVX512-NEXT: pushq %r14
+; AVX512-NEXT: pushq %r13
+; AVX512-NEXT: pushq %r12
+; AVX512-NEXT: pushq %rbx
+; AVX512-NEXT: movq %r9, %r14
+; AVX512-NEXT: movq %r8, %r11
+; AVX512-NEXT: movq %rcx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX512-NEXT: movq %rsi, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX512-NEXT: movq %rdi, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX512-NEXT: movq {{[0-9]+}}(%rsp), %r15
+; AVX512-NEXT: movq {{[0-9]+}}(%rsp), %rbx
+; AVX512-NEXT: movq {{[0-9]+}}(%rsp), %rax
+; AVX512-NEXT: movq {{[0-9]+}}(%rsp), %r10
+; AVX512-NEXT: movq {{[0-9]+}}(%rsp), %r8
+; AVX512-NEXT: movq {{[0-9]+}}(%rsp), %r12
+; AVX512-NEXT: lzcntq %r12, %rcx
+; AVX512-NEXT: lzcntq %r8, %r9
+; AVX512-NEXT: addl $64, %r9d
+; AVX512-NEXT: testq %r12, %r12
+; AVX512-NEXT: cmovnel %ecx, %r9d
+; AVX512-NEXT: lzcntq %r10, %rsi
+; AVX512-NEXT: lzcntq %rax, %rcx
+; AVX512-NEXT: addl $64, %ecx
+; AVX512-NEXT: testq %r10, %r10
+; AVX512-NEXT: cmovnel %esi, %ecx
+; AVX512-NEXT: subl $-128, %ecx
+; AVX512-NEXT: movq %r8, %rsi
+; AVX512-NEXT: orq %r12, %rsi
+; AVX512-NEXT: cmovnel %r9d, %ecx
+; AVX512-NEXT: lzcntq %rbx, %rdi
+; AVX512-NEXT: lzcntq %r15, %rsi
+; AVX512-NEXT: addl $64, %esi
+; AVX512-NEXT: testq %rbx, %rbx
+; AVX512-NEXT: cmovnel %edi, %esi
+; AVX512-NEXT: movq {{[0-9]+}}(%rsp), %r13
+; AVX512-NEXT: lzcntq %r13, %rbp
+; AVX512-NEXT: addl $64, %ebp
+; AVX512-NEXT: movq {{[0-9]+}}(%rsp), %r9
+; AVX512-NEXT: lzcntq %r9, %rdi
+; AVX512-NEXT: testq %r9, %r9
+; AVX512-NEXT: cmovnel %edi, %ebp
+; AVX512-NEXT: subl $-128, %ebp
+; AVX512-NEXT: movq %r15, %rdi
+; AVX512-NEXT: orq %rbx, %rdi
+; AVX512-NEXT: cmovnel %esi, %ebp
+; AVX512-NEXT: addl $256, %ebp # imm = 0x100
+; AVX512-NEXT: movq %r10, %rdi
+; AVX512-NEXT: orq %r12, %rdi
+; AVX512-NEXT: movq %rax, %rsi
+; AVX512-NEXT: orq %r8, %rsi
+; AVX512-NEXT: orq %rdi, %rsi
+; AVX512-NEXT: cmovnel %ecx, %ebp
+; AVX512-NEXT: movq {{[0-9]+}}(%rsp), %rdi
+; AVX512-NEXT: lzcntq %rdi, %rax
+; AVX512-NEXT: addl $64, %eax
+; AVX512-NEXT: movq {{[0-9]+}}(%rsp), %r12
+; AVX512-NEXT: lzcntq %r12, %rcx
+; AVX512-NEXT: testq %r12, %r12
+; AVX512-NEXT: cmovnel %ecx, %eax
+; AVX512-NEXT: lzcntq %r11, %rcx
+; AVX512-NEXT: addl $64, %ecx
+; AVX512-NEXT: lzcntq %r14, %rsi
+; AVX512-NEXT: testq %r14, %r14
+; AVX512-NEXT: cmovnel %esi, %ecx
+; AVX512-NEXT: subl $-128, %ecx
+; AVX512-NEXT: movq %rdi, %rsi
+; AVX512-NEXT: orq %r12, %rsi
+; AVX512-NEXT: cmovnel %eax, %ecx
+; AVX512-NEXT: movq %rdx, %rdi
+; AVX512-NEXT: lzcntq %rdx, %rdx
+; AVX512-NEXT: addl $64, %edx
+; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r10 # 8-byte Reload
+; AVX512-NEXT: lzcntq %r10, %rax
+; AVX512-NEXT: testq %r10, %r10
+; AVX512-NEXT: cmovnel %eax, %edx
+; AVX512-NEXT: lzcntq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Folded Reload
+; AVX512-NEXT: addl $64, %eax
+; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rsi # 8-byte Reload
+; AVX512-NEXT: lzcntq %rsi, %r8
+; AVX512-NEXT: testq %rsi, %rsi
+; AVX512-NEXT: cmovnel %r8d, %eax
+; AVX512-NEXT: subl $-128, %eax
+; AVX512-NEXT: orq %r10, %rdi
+; AVX512-NEXT: cmovnel %edx, %eax
+; AVX512-NEXT: orq %r12, %r14
+; AVX512-NEXT: orq {{[0-9]+}}(%rsp), %r11
+; AVX512-NEXT: addl $256, %eax # imm = 0x100
+; AVX512-NEXT: orq %r14, %r11
+; AVX512-NEXT: cmovnel %ecx, %eax
+; AVX512-NEXT: orq {{[0-9]+}}(%rsp), %rbx
+; AVX512-NEXT: orq {{[0-9]+}}(%rsp), %r9
+; AVX512-NEXT: orq %rbx, %r9
+; AVX512-NEXT: orq {{[0-9]+}}(%rsp), %r15
+; AVX512-NEXT: orq {{[0-9]+}}(%rsp), %r13
+; AVX512-NEXT: orq %r15, %r13
+; AVX512-NEXT: addl $512, %eax # imm = 0x200
+; AVX512-NEXT: orq %r9, %r13
+; AVX512-NEXT: cmovnel %ebp, %eax
+; AVX512-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX512-NEXT: popq %rbx
+; AVX512-NEXT: popq %r12
+; AVX512-NEXT: popq %r13
+; AVX512-NEXT: popq %r14
+; AVX512-NEXT: popq %r15
+; AVX512-NEXT: popq %rbp
+; AVX512-NEXT: retq
+ %cnt = call i1024 @llvm.ctlz.i1024(i1024 %a0, i1 0)
+ %res = trunc i1024 %cnt to i32
+ ret i32 %res
+}
+
+define i32 @load_ctlz_i1024(ptr %p0) nounwind {
+; SSE-LABEL: load_ctlz_i1024:
+; SSE: # %bb.0:
+; SSE-NEXT: pushq %rbp
+; SSE-NEXT: pushq %r15
+; SSE-NEXT: pushq %r14
+; SSE-NEXT: pushq %r13
+; SSE-NEXT: pushq %r12
+; SSE-NEXT: pushq %rbx
+; SSE-NEXT: movq 40(%rdi), %rbp
+; SSE-NEXT: movq 64(%rdi), %rbx
+; SSE-NEXT: movq 72(%rdi), %r11
+; SSE-NEXT: movq 80(%rdi), %r12
+; SSE-NEXT: movq 88(%rdi), %r14
+; SSE-NEXT: movq 96(%rdi), %rsi
+; SSE-NEXT: movq 104(%rdi), %r9
+; SSE-NEXT: movq 112(%rdi), %r10
+; SSE-NEXT: movq 120(%rdi), %r8
+; SSE-NEXT: bsrq %r8, %rax
+; SSE-NEXT: xorl $63, %eax
+; SSE-NEXT: bsrq %r10, %rcx
+; SSE-NEXT: xorl $63, %ecx
+; SSE-NEXT: orl $64, %ecx
+; SSE-NEXT: testq %r8, %r8
+; SSE-NEXT: cmovnel %eax, %ecx
+; SSE-NEXT: bsrq %r9, %rdx
+; SSE-NEXT: xorl $63, %edx
+; SSE-NEXT: bsrq %rsi, %rax
+; SSE-NEXT: movq %rsi, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; SSE-NEXT: xorl $63, %eax
+; SSE-NEXT: orl $64, %eax
+; SSE-NEXT: testq %r9, %r9
+; SSE-NEXT: movq %r9, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; SSE-NEXT: cmovnel %edx, %eax
+; SSE-NEXT: subl $-128, %eax
+; SSE-NEXT: movq %r10, %rdx
+; SSE-NEXT: movq %r10, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; SSE-NEXT: orq %r8, %rdx
+; SSE-NEXT: movq %r8, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; SSE-NEXT: cmovnel %ecx, %eax
+; SSE-NEXT: bsrq %r14, %rcx
+; SSE-NEXT: xorl $63, %ecx
+; SSE-NEXT: movq %r12, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; SSE-NEXT: bsrq %r12, %rdx
+; SSE-NEXT: xorl $63, %edx
+; SSE-NEXT: orl $64, %edx
+; SSE-NEXT: testq %r14, %r14
+; SSE-NEXT: cmovnel %ecx, %edx
+; SSE-NEXT: bsrq %r11, %rcx
+; SSE-NEXT: xorl $63, %ecx
+; SSE-NEXT: bsrq %rbx, %r15
+; SSE-NEXT: xorl $63, %r15d
+; SSE-NEXT: orl $64, %r15d
+; SSE-NEXT: testq %r11, %r11
+; SSE-NEXT: cmovnel %ecx, %r15d
+; SSE-NEXT: subl $-128, %r15d
+; SSE-NEXT: movq %r12, %rcx
+; SSE-NEXT: orq %r14, %rcx
+; SSE-NEXT: cmovnel %edx, %r15d
+; SSE-NEXT: movq 48(%rdi), %r12
+; SSE-NEXT: addl $256, %r15d # imm = 0x100
+; SSE-NEXT: movq %r9, %rcx
+; SSE-NEXT: orq %r8, %rcx
+; SSE-NEXT: movq %rsi, %rdx
+; SSE-NEXT: orq %r10, %rdx
+; SSE-NEXT: orq %rcx, %rdx
+; SSE-NEXT: movq 56(%rdi), %r13
+; SSE-NEXT: cmovnel %eax, %r15d
+; SSE-NEXT: bsrq %r13, %rax
+; SSE-NEXT: xorl $63, %eax
+; SSE-NEXT: bsrq %r12, %rcx
+; SSE-NEXT: xorl $63, %ecx
+; SSE-NEXT: orl $64, %ecx
+; SSE-NEXT: testq %r13, %r13
+; SSE-NEXT: cmovnel %eax, %ecx
+; SSE-NEXT: movq %rbp, %r10
+; SSE-NEXT: bsrq %rbp, %rax
+; SSE-NEXT: xorl $63, %eax
+; SSE-NEXT: movq 32(%rdi), %r8
+; SSE-NEXT: bsrq %r8, %rbp
+; SSE-NEXT: xorl $63, %ebp
+; SSE-NEXT: orl $64, %ebp
+; SSE-NEXT: testq %r10, %r10
+; SSE-NEXT: cmovnel %eax, %ebp
+; SSE-NEXT: subl $-128, %ebp
+; SSE-NEXT: movq %r12, %rax
+; SSE-NEXT: orq %r13, %rax
+; SSE-NEXT: cmovnel %ecx, %ebp
+; SSE-NEXT: movq 24(%rdi), %r9
+; SSE-NEXT: bsrq %r9, %rax
+; SSE-NEXT: xorl $63, %eax
+; SSE-NEXT: movq 16(%rdi), %rsi
+; SSE-NEXT: bsrq %rsi, %rcx
+; SSE-NEXT: xorl $63, %ecx
+; SSE-NEXT: orl $64, %ecx
+; SSE-NEXT: testq %r9, %r9
+; SSE-NEXT: cmovnel %eax, %ecx
+; SSE-NEXT: movl $127, %eax
+; SSE-NEXT: bsrq (%rdi), %rax
+; SSE-NEXT: movq 8(%rdi), %rdi
+; SSE-NEXT: bsrq %rdi, %rdx
+; SSE-NEXT: xorl $63, %edx
+; SSE-NEXT: xorl $63, %eax
+; SSE-NEXT: addl $64, %eax
+; SSE-NEXT: testq %rdi, %rdi
+; SSE-NEXT: cmovnel %edx, %eax
+; SSE-NEXT: subl $-128, %eax
+; SSE-NEXT: orq %r9, %rsi
+; SSE-NEXT: cmovnel %ecx, %eax
+; SSE-NEXT: orq %r13, %r10
+; SSE-NEXT: orq %r12, %r8
+; SSE-NEXT: addl $256, %eax # imm = 0x100
+; SSE-NEXT: orq %r10, %r8
+; SSE-NEXT: cmovnel %ebp, %eax
+; SSE-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %r14 # 8-byte Folded Reload
+; SSE-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %r11 # 8-byte Folded Reload
+; SSE-NEXT: orq %r14, %r11
+; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rcx # 8-byte Reload
+; SSE-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %rcx # 8-byte Folded Reload
+; SSE-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %rbx # 8-byte Folded Reload
+; SSE-NEXT: orq %rcx, %rbx
+; SSE-NEXT: addl $512, %eax # imm = 0x200
+; SSE-NEXT: orq %r11, %rbx
+; SSE-NEXT: cmovnel %r15d, %eax
+; SSE-NEXT: # kill: def $eax killed $eax killed $rax
+; SSE-NEXT: popq %rbx
+; SSE-NEXT: popq %r12
+; SSE-NEXT: popq %r13
+; SSE-NEXT: popq %r14
+; SSE-NEXT: popq %r15
+; SSE-NEXT: popq %rbp
+; SSE-NEXT: retq
+;
+; AVX2-LABEL: load_ctlz_i1024:
+; AVX2: # %bb.0:
+; AVX2-NEXT: pushq %rbp
+; AVX2-NEXT: pushq %r15
+; AVX2-NEXT: pushq %r14
+; AVX2-NEXT: pushq %r13
+; AVX2-NEXT: pushq %r12
+; AVX2-NEXT: pushq %rbx
+; AVX2-NEXT: movq 48(%rdi), %r9
+; AVX2-NEXT: movq 56(%rdi), %rbp
+; AVX2-NEXT: movq 64(%rdi), %r11
+; AVX2-NEXT: movq 72(%rdi), %r10
+; AVX2-NEXT: movq 80(%rdi), %r14
+; AVX2-NEXT: movq 88(%rdi), %rbx
+; AVX2-NEXT: movq 96(%rdi), %rdx
+; AVX2-NEXT: movq 104(%rdi), %r8
+; AVX2-NEXT: movq 112(%rdi), %rsi
+; AVX2-NEXT: movq 120(%rdi), %r15
+; AVX2-NEXT: lzcntq %r15, %rax
+; AVX2-NEXT: lzcntq %rsi, %rcx
+; AVX2-NEXT: addl $64, %ecx
+; AVX2-NEXT: testq %r15, %r15
+; AVX2-NEXT: cmovnel %eax, %ecx
+; AVX2-NEXT: xorl %r12d, %r12d
+; AVX2-NEXT: lzcntq %r8, %r12
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: lzcntq %rdx, %rax
+; AVX2-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX2-NEXT: addl $64, %eax
+; AVX2-NEXT: testq %r8, %r8
+; AVX2-NEXT: movq %r8, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX2-NEXT: cmovnel %r12d, %eax
+; AVX2-NEXT: subl $-128, %eax
+; AVX2-NEXT: movq %rsi, %r12
+; AVX2-NEXT: movq %rsi, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX2-NEXT: orq %r15, %r12
+; AVX2-NEXT: cmovnel %ecx, %eax
+; AVX2-NEXT: xorl %ecx, %ecx
+; AVX2-NEXT: lzcntq %rbx, %rcx
+; AVX2-NEXT: movq %r14, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX2-NEXT: xorl %r13d, %r13d
+; AVX2-NEXT: lzcntq %r14, %r13
+; AVX2-NEXT: addl $64, %r13d
+; AVX2-NEXT: testq %rbx, %rbx
+; AVX2-NEXT: cmovnel %ecx, %r13d
+; AVX2-NEXT: xorl %ecx, %ecx
+; AVX2-NEXT: lzcntq %r10, %rcx
+; AVX2-NEXT: xorl %r12d, %r12d
+; AVX2-NEXT: lzcntq %r11, %r12
+; AVX2-NEXT: addl $64, %r12d
+; AVX2-NEXT: testq %r10, %r10
+; AVX2-NEXT: cmovnel %ecx, %r12d
+; AVX2-NEXT: subl $-128, %r12d
+; AVX2-NEXT: movq %r14, %rcx
+; AVX2-NEXT: orq %rbx, %rcx
+; AVX2-NEXT: cmovnel %r13d, %r12d
+; AVX2-NEXT: addl $256, %r12d # imm = 0x100
+; AVX2-NEXT: movq %r8, %rcx
+; AVX2-NEXT: orq %r15, %rcx
+; AVX2-NEXT: orq %rsi, %rdx
+; AVX2-NEXT: orq %rcx, %rdx
+; AVX2-NEXT: cmovnel %eax, %r12d
+; AVX2-NEXT: movq %rbp, %r14
+; AVX2-NEXT: xorl %ecx, %ecx
+; AVX2-NEXT: lzcntq %rbp, %rcx
+; AVX2-NEXT: movq %r9, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: lzcntq %r9, %rax
+; AVX2-NEXT: addl $64, %eax
+; AVX2-NEXT: testq %rbp, %rbp
+; AVX2-NEXT: cmovnel %ecx, %eax
+; AVX2-NEXT: movq 32(%rdi), %r13
+; AVX2-NEXT: xorl %ebp, %ebp
+; AVX2-NEXT: lzcntq %r13, %rbp
+; AVX2-NEXT: addl $64, %ebp
+; AVX2-NEXT: movq 40(%rdi), %r8
+; AVX2-NEXT: xorl %edx, %edx
+; AVX2-NEXT: lzcntq %r8, %rdx
+; AVX2-NEXT: testq %r8, %r8
+; AVX2-NEXT: cmovnel %edx, %ebp
+; AVX2-NEXT: subl $-128, %ebp
+; AVX2-NEXT: movq %r9, %rdx
+; AVX2-NEXT: orq %r14, %rdx
+; AVX2-NEXT: cmovnel %eax, %ebp
+; AVX2-NEXT: movq 16(%rdi), %r9
+; AVX2-NEXT: xorl %ecx, %ecx
+; AVX2-NEXT: lzcntq %r9, %rcx
+; AVX2-NEXT: addl $64, %ecx
+; AVX2-NEXT: movq 24(%rdi), %rdx
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: lzcntq %rdx, %rax
+; AVX2-NEXT: testq %rdx, %rdx
+; AVX2-NEXT: cmovnel %eax, %ecx
+; AVX2-NEXT: movq 8(%rdi), %rsi
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: lzcntq (%rdi), %rax
+; AVX2-NEXT: addl $64, %eax
+; AVX2-NEXT: lzcntq %rsi, %rdi
+; AVX2-NEXT: testq %rsi, %rsi
+; AVX2-NEXT: cmovnel %edi, %eax
+; AVX2-NEXT: subl $-128, %eax
+; AVX2-NEXT: orq %rdx, %r9
+; AVX2-NEXT: cmovnel %ecx, %eax
+; AVX2-NEXT: orq %r14, %r8
+; AVX2-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %r13 # 8-byte Folded Reload
+; AVX2-NEXT: addl $256, %eax # imm = 0x100
+; AVX2-NEXT: orq %r8, %r13
+; AVX2-NEXT: cmovnel %ebp, %eax
+; AVX2-NEXT: orq %r15, %rbx
+; AVX2-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %r10 # 8-byte Folded Reload
+; AVX2-NEXT: orq %rbx, %r10
+; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rcx # 8-byte Reload
+; AVX2-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %rcx # 8-byte Folded Reload
+; AVX2-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %r11 # 8-byte Folded Reload
+; AVX2-NEXT: orq %rcx, %r11
+; AVX2-NEXT: addl $512, %eax # imm = 0x200
+; AVX2-NEXT: orq %r10, %r11
+; AVX2-NEXT: cmovnel %r12d, %eax
+; AVX2-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX2-NEXT: popq %rbx
+; AVX2-NEXT: popq %r12
+; AVX2-NEXT: popq %r13
+; AVX2-NEXT: popq %r14
+; AVX2-NEXT: popq %r15
+; AVX2-NEXT: popq %rbp
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: load_ctlz_i1024:
+; AVX512: # %bb.0:
+; AVX512-NEXT: pushq %rbp
+; AVX512-NEXT: pushq %r15
+; AVX512-NEXT: pushq %r14
+; AVX512-NEXT: pushq %r13
+; AVX512-NEXT: pushq %r12
+; AVX512-NEXT: pushq %rbx
+; AVX512-NEXT: movq 32(%rdi), %r14
+; AVX512-NEXT: movq 48(%rdi), %rbp
+; AVX512-NEXT: movq 64(%rdi), %r11
+; AVX512-NEXT: movq 72(%rdi), %r10
+; AVX512-NEXT: movq 80(%rdi), %rdx
+; AVX512-NEXT: movq 88(%rdi), %rbx
+; AVX512-NEXT: movq 96(%rdi), %rsi
+; AVX512-NEXT: movq 104(%rdi), %r9
+; AVX512-NEXT: movq 112(%rdi), %r8
+; AVX512-NEXT: movq 120(%rdi), %r15
+; AVX512-NEXT: lzcntq %r15, %rax
+; AVX512-NEXT: lzcntq %r8, %rcx
+; AVX512-NEXT: addl $64, %ecx
+; AVX512-NEXT: testq %r15, %r15
+; AVX512-NEXT: cmovnel %eax, %ecx
+; AVX512-NEXT: lzcntq %r9, %r12
+; AVX512-NEXT: lzcntq %rsi, %rax
+; AVX512-NEXT: movq %rsi, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX512-NEXT: addl $64, %eax
+; AVX512-NEXT: testq %r9, %r9
+; AVX512-NEXT: movq %r9, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX512-NEXT: cmovnel %r12d, %eax
+; AVX512-NEXT: subl $-128, %eax
+; AVX512-NEXT: movq %r8, %r12
+; AVX512-NEXT: movq %r8, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX512-NEXT: orq %r15, %r12
+; AVX512-NEXT: cmovnel %ecx, %eax
+; AVX512-NEXT: lzcntq %rbx, %rcx
+; AVX512-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX512-NEXT: lzcntq %rdx, %r13
+; AVX512-NEXT: addl $64, %r13d
+; AVX512-NEXT: testq %rbx, %rbx
+; AVX512-NEXT: cmovnel %ecx, %r13d
+; AVX512-NEXT: lzcntq %r10, %rcx
+; AVX512-NEXT: lzcntq %r11, %r12
+; AVX512-NEXT: addl $64, %r12d
+; AVX512-NEXT: testq %r10, %r10
+; AVX512-NEXT: cmovnel %ecx, %r12d
+; AVX512-NEXT: subl $-128, %r12d
+; AVX512-NEXT: movq %rdx, %rcx
+; AVX512-NEXT: orq %rbx, %rcx
+; AVX512-NEXT: cmovnel %r13d, %r12d
+; AVX512-NEXT: addl $256, %r12d # imm = 0x100
+; AVX512-NEXT: movq %r9, %rcx
+; AVX512-NEXT: orq %r15, %rcx
+; AVX512-NEXT: orq %r8, %rsi
+; AVX512-NEXT: orq %rcx, %rsi
+; AVX512-NEXT: movq 56(%rdi), %r13
+; AVX512-NEXT: cmovnel %eax, %r12d
+; AVX512-NEXT: lzcntq %r13, %rcx
+; AVX512-NEXT: movq %rbp, %rsi
+; AVX512-NEXT: movq %rbp, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX512-NEXT: lzcntq %rbp, %rax
+; AVX512-NEXT: addl $64, %eax
+; AVX512-NEXT: testq %r13, %r13
+; AVX512-NEXT: cmovnel %ecx, %eax
+; AVX512-NEXT: lzcntq %r14, %rbp
+; AVX512-NEXT: addl $64, %ebp
+; AVX512-NEXT: movq 40(%rdi), %r8
+; AVX512-NEXT: lzcntq %r8, %rdx
+; AVX512-NEXT: testq %r8, %r8
+; AVX512-NEXT: cmovnel %edx, %ebp
+; AVX512-NEXT: subl $-128, %ebp
+; AVX512-NEXT: movq %rsi, %rdx
+; AVX512-NEXT: orq %r13, %rdx
+; AVX512-NEXT: cmovnel %eax, %ebp
+; AVX512-NEXT: movq 16(%rdi), %r9
+; AVX512-NEXT: lzcntq %r9, %rcx
+; AVX512-NEXT: addl $64, %ecx
+; AVX512-NEXT: movq 24(%rdi), %rdx
+; AVX512-NEXT: lzcntq %rdx, %rax
+; AVX512-NEXT: testq %rdx, %rdx
+; AVX512-NEXT: cmovnel %eax, %ecx
+; AVX512-NEXT: movq 8(%rdi), %rsi
+; AVX512-NEXT: lzcntq (%rdi), %rax
+; AVX512-NEXT: addl $64, %eax
+; AVX512-NEXT: lzcntq %rsi, %rdi
+; AVX512-NEXT: testq %rsi, %rsi
+; AVX512-NEXT: cmovnel %edi, %eax
+; AVX512-NEXT: subl $-128, %eax
+; AVX512-NEXT: orq %rdx, %r9
+; AVX512-NEXT: cmovnel %ecx, %eax
+; AVX512-NEXT: orq %r13, %r8
+; AVX512-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %r14 # 8-byte Folded Reload
+; AVX512-NEXT: addl $256, %eax # imm = 0x100
+; AVX512-NEXT: orq %r8, %r14
+; AVX512-NEXT: cmovnel %ebp, %eax
+; AVX512-NEXT: orq %r15, %rbx
+; AVX512-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %r10 # 8-byte Folded Reload
+; AVX512-NEXT: orq %rbx, %r10
+; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rcx # 8-byte Reload
+; AVX512-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %rcx # 8-byte Folded Reload
+; AVX512-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %r11 # 8-byte Folded Reload
+; AVX512-NEXT: orq %rcx, %r11
+; AVX512-NEXT: addl $512, %eax # imm = 0x200
+; AVX512-NEXT: orq %r10, %r11
+; AVX512-NEXT: cmovnel %r12d, %eax
+; AVX512-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX512-NEXT: popq %rbx
+; AVX512-NEXT: popq %r12
+; AVX512-NEXT: popq %r13
+; AVX512-NEXT: popq %r14
+; AVX512-NEXT: popq %r15
+; AVX512-NEXT: popq %rbp
+; AVX512-NEXT: retq
+ %a0 = load i1024, ptr %p0
+ %cnt = call i1024 @llvm.ctlz.i1024(i1024 %a0, i1 0)
+ %res = trunc i1024 %cnt to i32
+ ret i32 %res
+}
+
+;
+; CTTZ
+;
+
+define i32 @test_cttz_i128(i128 %a0) nounwind {
+; SSE-LABEL: test_cttz_i128:
+; SSE: # %bb.0:
+; SSE-NEXT: rep bsfq %rdi, %rcx
+; SSE-NEXT: movl $64, %eax
+; SSE-NEXT: rep bsfq %rsi, %rax
+; SSE-NEXT: addl $64, %eax
+; SSE-NEXT: testq %rdi, %rdi
+; SSE-NEXT: cmovnel %ecx, %eax
+; SSE-NEXT: # kill: def $eax killed $eax killed $rax
+; SSE-NEXT: retq
+;
+; AVX2-LABEL: test_cttz_i128:
+; AVX2: # %bb.0:
+; AVX2-NEXT: tzcntq %rdi, %rcx
+; AVX2-NEXT: tzcntq %rsi, %rax
+; AVX2-NEXT: addl $64, %eax
+; AVX2-NEXT: testq %rdi, %rdi
+; AVX2-NEXT: cmovnel %ecx, %eax
+; AVX2-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_cttz_i128:
+; AVX512: # %bb.0:
+; AVX512-NEXT: tzcntq %rdi, %rcx
+; AVX512-NEXT: tzcntq %rsi, %rax
+; AVX512-NEXT: addl $64, %eax
+; AVX512-NEXT: testq %rdi, %rdi
+; AVX512-NEXT: cmovnel %ecx, %eax
+; AVX512-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX512-NEXT: retq
+ %cnt = call i128 @llvm.cttz.i128(i128 %a0, i1 0)
+ %res = trunc i128 %cnt to i32
+ ret i32 %res
+}
+
+define i32 @load_cttz_i128(ptr %p0) nounwind {
+; SSE-LABEL: load_cttz_i128:
+; SSE: # %bb.0:
+; SSE-NEXT: movq (%rdi), %rcx
+; SSE-NEXT: rep bsfq %rcx, %rdx
+; SSE-NEXT: movl $64, %eax
+; SSE-NEXT: rep bsfq 8(%rdi), %rax
+; SSE-NEXT: addl $64, %eax
+; SSE-NEXT: testq %rcx, %rcx
+; SSE-NEXT: cmovnel %edx, %eax
+; SSE-NEXT: # kill: def $eax killed $eax killed $rax
+; SSE-NEXT: retq
+;
+; AVX2-LABEL: load_cttz_i128:
+; AVX2: # %bb.0:
+; AVX2-NEXT: movq (%rdi), %rcx
+; AVX2-NEXT: tzcntq %rcx, %rdx
+; AVX2-NEXT: tzcntq 8(%rdi), %rax
+; AVX2-NEXT: addl $64, %eax
+; AVX2-NEXT: testq %rcx, %rcx
+; AVX2-NEXT: cmovnel %edx, %eax
+; AVX2-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: load_cttz_i128:
+; AVX512: # %bb.0:
+; AVX512-NEXT: movq (%rdi), %rcx
+; AVX512-NEXT: tzcntq %rcx, %rdx
+; AVX512-NEXT: tzcntq 8(%rdi), %rax
+; AVX512-NEXT: addl $64, %eax
+; AVX512-NEXT: testq %rcx, %rcx
+; AVX512-NEXT: cmovnel %edx, %eax
+; AVX512-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX512-NEXT: retq
+ %a0 = load i128, ptr %p0
+ %cnt = call i128 @llvm.cttz.i128(i128 %a0, i1 0)
+ %res = trunc i128 %cnt to i32
+ ret i32 %res
+}
+
+define i32 @test_cttz_i256(i256 %a0) nounwind {
+; SSE-LABEL: test_cttz_i256:
+; SSE: # %bb.0:
+; SSE-NEXT: rep bsfq %rdi, %rax
+; SSE-NEXT: rep bsfq %rsi, %r8
+; SSE-NEXT: addl $64, %r8d
+; SSE-NEXT: testq %rdi, %rdi
+; SSE-NEXT: cmovnel %eax, %r8d
+; SSE-NEXT: rep bsfq %rdx, %r9
+; SSE-NEXT: movl $64, %eax
+; SSE-NEXT: rep bsfq %rcx, %rax
+; SSE-NEXT: addl $64, %eax
+; SSE-NEXT: testq %rdx, %rdx
+; SSE-NEXT: cmovnel %r9d, %eax
+; SSE-NEXT: subl $-128, %eax
+; SSE-NEXT: orq %rsi, %rdi
+; SSE-NEXT: cmovnel %r8d, %eax
+; SSE-NEXT: # kill: def $eax killed $eax killed $rax
+; SSE-NEXT: retq
+;
+; AVX2-LABEL: test_cttz_i256:
+; AVX2: # %bb.0:
+; AVX2-NEXT: tzcntq %rdi, %rax
+; AVX2-NEXT: tzcntq %rsi, %r8
+; AVX2-NEXT: addl $64, %r8d
+; AVX2-NEXT: testq %rdi, %rdi
+; AVX2-NEXT: cmovnel %eax, %r8d
+; AVX2-NEXT: tzcntq %rdx, %r9
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: tzcntq %rcx, %rax
+; AVX2-NEXT: addl $64, %eax
+; AVX2-NEXT: testq %rdx, %rdx
+; AVX2-NEXT: cmovnel %r9d, %eax
+; AVX2-NEXT: subl $-128, %eax
+; AVX2-NEXT: orq %rsi, %rdi
+; AVX2-NEXT: cmovnel %r8d, %eax
+; AVX2-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_cttz_i256:
+; AVX512: # %bb.0:
+; AVX512-NEXT: tzcntq %rdi, %rax
+; AVX512-NEXT: tzcntq %rsi, %r8
+; AVX512-NEXT: addl $64, %r8d
+; AVX512-NEXT: testq %rdi, %rdi
+; AVX512-NEXT: cmovnel %eax, %r8d
+; AVX512-NEXT: tzcntq %rdx, %r9
+; AVX512-NEXT: tzcntq %rcx, %rax
+; AVX512-NEXT: addl $64, %eax
+; AVX512-NEXT: testq %rdx, %rdx
+; AVX512-NEXT: cmovnel %r9d, %eax
+; AVX512-NEXT: subl $-128, %eax
+; AVX512-NEXT: orq %rsi, %rdi
+; AVX512-NEXT: cmovnel %r8d, %eax
+; AVX512-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX512-NEXT: retq
+ %cnt = call i256 @llvm.cttz.i256(i256 %a0, i1 0)
+ %res = trunc i256 %cnt to i32
+ ret i32 %res
+}
+
+define i32 @load_cttz_i256(ptr %p0) nounwind {
+; SSE-LABEL: load_cttz_i256:
+; SSE: # %bb.0:
+; SSE-NEXT: movq 16(%rdi), %rcx
+; SSE-NEXT: movq (%rdi), %rdx
+; SSE-NEXT: movq 8(%rdi), %rsi
+; SSE-NEXT: rep bsfq %rdx, %rax
+; SSE-NEXT: rep bsfq %rsi, %r8
+; SSE-NEXT: addl $64, %r8d
+; SSE-NEXT: testq %rdx, %rdx
+; SSE-NEXT: cmovnel %eax, %r8d
+; SSE-NEXT: rep bsfq %rcx, %r9
+; SSE-NEXT: movl $64, %eax
+; SSE-NEXT: rep bsfq 24(%rdi), %rax
+; SSE-NEXT: addl $64, %eax
+; SSE-NEXT: testq %rcx, %rcx
+; SSE-NEXT: cmovnel %r9d, %eax
+; SSE-NEXT: subl $-128, %eax
+; SSE-NEXT: orq %rsi, %rdx
+; SSE-NEXT: cmovnel %r8d, %eax
+; SSE-NEXT: # kill: def $eax killed $eax killed $rax
+; SSE-NEXT: retq
+;
+; AVX2-LABEL: load_cttz_i256:
+; AVX2: # %bb.0:
+; AVX2-NEXT: movq (%rdi), %rcx
+; AVX2-NEXT: movq 8(%rdi), %rdx
+; AVX2-NEXT: tzcntq %rcx, %rax
+; AVX2-NEXT: tzcntq %rdx, %rsi
+; AVX2-NEXT: addl $64, %esi
+; AVX2-NEXT: testq %rcx, %rcx
+; AVX2-NEXT: cmovnel %eax, %esi
+; AVX2-NEXT: movq 16(%rdi), %r8
+; AVX2-NEXT: tzcntq %r8, %r9
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: tzcntq 24(%rdi), %rax
+; AVX2-NEXT: addl $64, %eax
+; AVX2-NEXT: testq %r8, %r8
+; AVX2-NEXT: cmovnel %r9d, %eax
+; AVX2-NEXT: subl $-128, %eax
+; AVX2-NEXT: orq %rdx, %rcx
+; AVX2-NEXT: cmovnel %esi, %eax
+; AVX2-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: load_cttz_i256:
+; AVX512: # %bb.0:
+; AVX512-NEXT: movq 16(%rdi), %rcx
+; AVX512-NEXT: movq (%rdi), %rdx
+; AVX512-NEXT: movq 8(%rdi), %rsi
+; AVX512-NEXT: tzcntq %rdx, %rax
+; AVX512-NEXT: tzcntq %rsi, %r8
+; AVX512-NEXT: addl $64, %r8d
+; AVX512-NEXT: testq %rdx, %rdx
+; AVX512-NEXT: cmovnel %eax, %r8d
+; AVX512-NEXT: tzcntq %rcx, %r9
+; AVX512-NEXT: tzcntq 24(%rdi), %rax
+; AVX512-NEXT: addl $64, %eax
+; AVX512-NEXT: testq %rcx, %rcx
+; AVX512-NEXT: cmovnel %r9d, %eax
+; AVX512-NEXT: subl $-128, %eax
+; AVX512-NEXT: orq %rsi, %rdx
+; AVX512-NEXT: cmovnel %r8d, %eax
+; AVX512-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX512-NEXT: retq
+ %a0 = load i256, ptr %p0
+ %cnt = call i256 @llvm.cttz.i256(i256 %a0, i1 0)
+ %res = trunc i256 %cnt to i32
+ ret i32 %res
+}
+
+define i32 @test_cttz_i512(i512 %a0) nounwind {
+; SSE-LABEL: test_cttz_i512:
+; SSE: # %bb.0:
+; SSE-NEXT: pushq %r14
+; SSE-NEXT: pushq %rbx
+; SSE-NEXT: rep bsfq %rdi, %rax
+; SSE-NEXT: rep bsfq %rsi, %r11
+; SSE-NEXT: addl $64, %r11d
+; SSE-NEXT: testq %rdi, %rdi
+; SSE-NEXT: cmovnel %eax, %r11d
+; SSE-NEXT: rep bsfq %rdx, %rax
+; SSE-NEXT: rep bsfq %rcx, %r10
+; SSE-NEXT: addl $64, %r10d
+; SSE-NEXT: testq %rdx, %rdx
+; SSE-NEXT: cmovnel %eax, %r10d
+; SSE-NEXT: movq {{[0-9]+}}(%rsp), %rbx
+; SSE-NEXT: subl $-128, %r10d
+; SSE-NEXT: movq %rdi, %rax
+; SSE-NEXT: orq %rsi, %rax
+; SSE-NEXT: cmovnel %r11d, %r10d
+; SSE-NEXT: rep bsfq %r8, %rax
+; SSE-NEXT: rep bsfq %r9, %r11
+; SSE-NEXT: addl $64, %r11d
+; SSE-NEXT: testq %r8, %r8
+; SSE-NEXT: cmovnel %eax, %r11d
+; SSE-NEXT: rep bsfq %rbx, %r14
+; SSE-NEXT: movl $64, %eax
+; SSE-NEXT: rep bsfq {{[0-9]+}}(%rsp), %rax
+; SSE-NEXT: addl $64, %eax
+; SSE-NEXT: testq %rbx, %rbx
+; SSE-NEXT: cmovnel %r14d, %eax
+; SSE-NEXT: subl $-128, %eax
+; SSE-NEXT: orq %r9, %r8
+; SSE-NEXT: cmovnel %r11d, %eax
+; SSE-NEXT: addl $256, %eax # imm = 0x100
+; SSE-NEXT: orq %rcx, %rsi
+; SSE-NEXT: orq %rdx, %rdi
+; SSE-NEXT: orq %rsi, %rdi
+; SSE-NEXT: cmovnel %r10d, %eax
+; SSE-NEXT: # kill: def $eax killed $eax killed $rax
+; SSE-NEXT: popq %rbx
+; SSE-NEXT: popq %r14
+; SSE-NEXT: retq
+;
+; AVX2-LABEL: test_cttz_i512:
+; AVX2: # %bb.0:
+; AVX2-NEXT: pushq %r14
+; AVX2-NEXT: pushq %rbx
+; AVX2-NEXT: tzcntq %rdi, %rax
+; AVX2-NEXT: tzcntq %rsi, %r11
+; AVX2-NEXT: addl $64, %r11d
+; AVX2-NEXT: testq %rdi, %rdi
+; AVX2-NEXT: cmovnel %eax, %r11d
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: tzcntq %rdx, %rax
+; AVX2-NEXT: tzcntq %rcx, %r10
+; AVX2-NEXT: addl $64, %r10d
+; AVX2-NEXT: testq %rdx, %rdx
+; AVX2-NEXT: cmovnel %eax, %r10d
+; AVX2-NEXT: subl $-128, %r10d
+; AVX2-NEXT: movq %rdi, %rax
+; AVX2-NEXT: orq %rsi, %rax
+; AVX2-NEXT: cmovnel %r11d, %r10d
+; AVX2-NEXT: movq {{[0-9]+}}(%rsp), %r11
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: tzcntq %r8, %rax
+; AVX2-NEXT: xorl %ebx, %ebx
+; AVX2-NEXT: tzcntq %r9, %rbx
+; AVX2-NEXT: addl $64, %ebx
+; AVX2-NEXT: testq %r8, %r8
+; AVX2-NEXT: cmovnel %eax, %ebx
+; AVX2-NEXT: xorl %r14d, %r14d
+; AVX2-NEXT: tzcntq %r11, %r14
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: tzcntq {{[0-9]+}}(%rsp), %rax
+; AVX2-NEXT: addl $64, %eax
+; AVX2-NEXT: testq %r11, %r11
+; AVX2-NEXT: cmovnel %r14d, %eax
+; AVX2-NEXT: subl $-128, %eax
+; AVX2-NEXT: orq %r9, %r8
+; AVX2-NEXT: cmovnel %ebx, %eax
+; AVX2-NEXT: addl $256, %eax # imm = 0x100
+; AVX2-NEXT: orq %rcx, %rsi
+; AVX2-NEXT: orq %rdx, %rdi
+; AVX2-NEXT: orq %rsi, %rdi
+; AVX2-NEXT: cmovnel %r10d, %eax
+; AVX2-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX2-NEXT: popq %rbx
+; AVX2-NEXT: popq %r14
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_cttz_i512:
+; AVX512: # %bb.0:
+; AVX512-NEXT: pushq %r14
+; AVX512-NEXT: pushq %rbx
+; AVX512-NEXT: movq {{[0-9]+}}(%rsp), %r11
+; AVX512-NEXT: tzcntq %rdi, %rax
+; AVX512-NEXT: tzcntq %rsi, %rbx
+; AVX512-NEXT: addl $64, %ebx
+; AVX512-NEXT: testq %rdi, %rdi
+; AVX512-NEXT: cmovnel %eax, %ebx
+; AVX512-NEXT: tzcntq %rdx, %rax
+; AVX512-NEXT: tzcntq %rcx, %r10
+; AVX512-NEXT: addl $64, %r10d
+; AVX512-NEXT: testq %rdx, %rdx
+; AVX512-NEXT: cmovnel %eax, %r10d
+; AVX512-NEXT: subl $-128, %r10d
+; AVX512-NEXT: movq %rdi, %rax
+; AVX512-NEXT: orq %rsi, %rax
+; AVX512-NEXT: cmovnel %ebx, %r10d
+; AVX512-NEXT: tzcntq %r8, %rax
+; AVX512-NEXT: tzcntq %r9, %rbx
+; AVX512-NEXT: addl $64, %ebx
+; AVX512-NEXT: testq %r8, %r8
+; AVX512-NEXT: cmovnel %eax, %ebx
+; AVX512-NEXT: tzcntq {{[0-9]+}}(%rsp), %rax
+; AVX512-NEXT: tzcntq %r11, %r14
+; AVX512-NEXT: addl $64, %eax
+; AVX512-NEXT: testq %r11, %r11
+; AVX512-NEXT: cmovnel %r14d, %eax
+; AVX512-NEXT: subl $-128, %eax
+; AVX512-NEXT: orq %r9, %r8
+; AVX512-NEXT: cmovnel %ebx, %eax
+; AVX512-NEXT: addl $256, %eax # imm = 0x100
+; AVX512-NEXT: orq %rcx, %rsi
+; AVX512-NEXT: orq %rdx, %rdi
+; AVX512-NEXT: orq %rsi, %rdi
+; AVX512-NEXT: cmovnel %r10d, %eax
+; AVX512-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX512-NEXT: popq %rbx
+; AVX512-NEXT: popq %r14
+; AVX512-NEXT: retq
+ %cnt = call i512 @llvm.cttz.i512(i512 %a0, i1 0)
+ %res = trunc i512 %cnt to i32
+ ret i32 %res
+}
+
+define i32 @load_cttz_i512(ptr %p0) nounwind {
+; SSE-LABEL: load_cttz_i512:
+; SSE: # %bb.0:
+; SSE-NEXT: pushq %r15
+; SSE-NEXT: pushq %r14
+; SSE-NEXT: pushq %rbx
+; SSE-NEXT: movq 48(%rdi), %r10
+; SSE-NEXT: movq 40(%rdi), %r9
+; SSE-NEXT: movq 24(%rdi), %r8
+; SSE-NEXT: movq 16(%rdi), %rdx
+; SSE-NEXT: movq (%rdi), %rcx
+; SSE-NEXT: movq 8(%rdi), %rsi
+; SSE-NEXT: rep bsfq %rcx, %rax
+; SSE-NEXT: rep bsfq %rsi, %rbx
+; SSE-NEXT: addl $64, %ebx
+; SSE-NEXT: testq %rcx, %rcx
+; SSE-NEXT: cmovnel %eax, %ebx
+; SSE-NEXT: rep bsfq %rdx, %rax
+; SSE-NEXT: rep bsfq %r8, %r11
+; SSE-NEXT: addl $64, %r11d
+; SSE-NEXT: testq %rdx, %rdx
+; SSE-NEXT: cmovnel %eax, %r11d
+; SSE-NEXT: movq 32(%rdi), %r14
+; SSE-NEXT: subl $-128, %r11d
+; SSE-NEXT: movq %rcx, %rax
+; SSE-NEXT: orq %rsi, %rax
+; SSE-NEXT: cmovnel %ebx, %r11d
+; SSE-NEXT: rep bsfq %r14, %rax
+; SSE-NEXT: rep bsfq %r9, %rbx
+; SSE-NEXT: addl $64, %ebx
+; SSE-NEXT: testq %r14, %r14
+; SSE-NEXT: cmovnel %eax, %ebx
+; SSE-NEXT: rep bsfq %r10, %r15
+; SSE-NEXT: movl $64, %eax
+; SSE-NEXT: rep bsfq 56(%rdi), %rax
+; SSE-NEXT: addl $64, %eax
+; SSE-NEXT: testq %r10, %r10
+; SSE-NEXT: cmovnel %r15d, %eax
+; SSE-NEXT: subl $-128, %eax
+; SSE-NEXT: orq %r9, %r14
+; SSE-NEXT: cmovnel %ebx, %eax
+; SSE-NEXT: addl $256, %eax # imm = 0x100
+; SSE-NEXT: orq %r8, %rsi
+; SSE-NEXT: orq %rdx, %rcx
+; SSE-NEXT: orq %rsi, %rcx
+; SSE-NEXT: cmovnel %r11d, %eax
+; SSE-NEXT: # kill: def $eax killed $eax killed $rax
+; SSE-NEXT: popq %rbx
+; SSE-NEXT: popq %r14
+; SSE-NEXT: popq %r15
+; SSE-NEXT: retq
+;
+; AVX2-LABEL: load_cttz_i512:
+; AVX2: # %bb.0:
+; AVX2-NEXT: pushq %r15
+; AVX2-NEXT: pushq %r14
+; AVX2-NEXT: pushq %rbx
+; AVX2-NEXT: movq 48(%rdi), %r10
+; AVX2-NEXT: movq 40(%rdi), %r9
+; AVX2-NEXT: movq 24(%rdi), %r8
+; AVX2-NEXT: movq 16(%rdi), %rdx
+; AVX2-NEXT: movq (%rdi), %rcx
+; AVX2-NEXT: movq 8(%rdi), %rsi
+; AVX2-NEXT: tzcntq %rcx, %rax
+; AVX2-NEXT: xorl %ebx, %ebx
+; AVX2-NEXT: tzcntq %rsi, %rbx
+; AVX2-NEXT: addl $64, %ebx
+; AVX2-NEXT: testq %rcx, %rcx
+; AVX2-NEXT: cmovnel %eax, %ebx
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: tzcntq %rdx, %rax
+; AVX2-NEXT: tzcntq %r8, %r11
+; AVX2-NEXT: addl $64, %r11d
+; AVX2-NEXT: testq %rdx, %rdx
+; AVX2-NEXT: cmovnel %eax, %r11d
+; AVX2-NEXT: subl $-128, %r11d
+; AVX2-NEXT: movq %rcx, %rax
+; AVX2-NEXT: orq %rsi, %rax
+; AVX2-NEXT: cmovnel %ebx, %r11d
+; AVX2-NEXT: movq 32(%rdi), %rbx
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: tzcntq %rbx, %rax
+; AVX2-NEXT: xorl %r14d, %r14d
+; AVX2-NEXT: tzcntq %r9, %r14
+; AVX2-NEXT: addl $64, %r14d
+; AVX2-NEXT: testq %rbx, %rbx
+; AVX2-NEXT: cmovnel %eax, %r14d
+; AVX2-NEXT: xorl %r15d, %r15d
+; AVX2-NEXT: tzcntq %r10, %r15
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: tzcntq 56(%rdi), %rax
+; AVX2-NEXT: addl $64, %eax
+; AVX2-NEXT: testq %r10, %r10
+; AVX2-NEXT: cmovnel %r15d, %eax
+; AVX2-NEXT: subl $-128, %eax
+; AVX2-NEXT: orq %r9, %rbx
+; AVX2-NEXT: cmovnel %r14d, %eax
+; AVX2-NEXT: addl $256, %eax # imm = 0x100
+; AVX2-NEXT: orq %r8, %rsi
+; AVX2-NEXT: orq %rdx, %rcx
+; AVX2-NEXT: orq %rsi, %rcx
+; AVX2-NEXT: cmovnel %r11d, %eax
+; AVX2-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX2-NEXT: popq %rbx
+; AVX2-NEXT: popq %r14
+; AVX2-NEXT: popq %r15
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: load_cttz_i512:
+; AVX512: # %bb.0:
+; AVX512-NEXT: pushq %r14
+; AVX512-NEXT: pushq %rbx
+; AVX512-NEXT: movq 48(%rdi), %r11
+; AVX512-NEXT: movq 40(%rdi), %r9
+; AVX512-NEXT: movq 32(%rdi), %r10
+; AVX512-NEXT: movq 24(%rdi), %r8
+; AVX512-NEXT: movq 16(%rdi), %rdx
+; AVX512-NEXT: movq (%rdi), %rcx
+; AVX512-NEXT: movq 8(%rdi), %rsi
+; AVX512-NEXT: tzcntq %rcx, %rax
+; AVX512-NEXT: tzcntq %rsi, %r14
+; AVX512-NEXT: addl $64, %r14d
+; AVX512-NEXT: testq %rcx, %rcx
+; AVX512-NEXT: cmovnel %eax, %r14d
+; AVX512-NEXT: tzcntq %rdx, %rax
+; AVX512-NEXT: tzcntq %r8, %rbx
+; AVX512-NEXT: addl $64, %ebx
+; AVX512-NEXT: testq %rdx, %rdx
+; AVX512-NEXT: cmovnel %eax, %ebx
+; AVX512-NEXT: subl $-128, %ebx
+; AVX512-NEXT: movq %rcx, %rax
+; AVX512-NEXT: orq %rsi, %rax
+; AVX512-NEXT: cmovnel %r14d, %ebx
+; AVX512-NEXT: tzcntq %r10, %rax
+; AVX512-NEXT: tzcntq %r9, %r14
+; AVX512-NEXT: addl $64, %r14d
+; AVX512-NEXT: testq %r10, %r10
+; AVX512-NEXT: cmovnel %eax, %r14d
+; AVX512-NEXT: tzcntq 56(%rdi), %rax
+; AVX512-NEXT: tzcntq %r11, %rdi
+; AVX512-NEXT: addl $64, %eax
+; AVX512-NEXT: testq %r11, %r11
+; AVX512-NEXT: cmovnel %edi, %eax
+; AVX512-NEXT: subl $-128, %eax
+; AVX512-NEXT: orq %r9, %r10
+; AVX512-NEXT: cmovnel %r14d, %eax
+; AVX512-NEXT: addl $256, %eax # imm = 0x100
+; AVX512-NEXT: orq %r8, %rsi
+; AVX512-NEXT: orq %rdx, %rcx
+; AVX512-NEXT: orq %rsi, %rcx
+; AVX512-NEXT: cmovnel %ebx, %eax
+; AVX512-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX512-NEXT: popq %rbx
+; AVX512-NEXT: popq %r14
+; AVX512-NEXT: retq
+ %a0 = load i512, ptr %p0
+ %cnt = call i512 @llvm.cttz.i512(i512 %a0, i1 0)
+ %res = trunc i512 %cnt to i32
+ ret i32 %res
+}
+
+define i32 @test_cttz_i1024(i1024 %a0) nounwind {
+; SSE-LABEL: test_cttz_i1024:
+; SSE: # %bb.0:
+; SSE-NEXT: pushq %rbp
+; SSE-NEXT: pushq %r15
+; SSE-NEXT: pushq %r14
+; SSE-NEXT: pushq %r13
+; SSE-NEXT: pushq %r12
+; SSE-NEXT: pushq %rbx
+; SSE-NEXT: movq %r9, %r13
+; SSE-NEXT: movq %r8, %r14
+; SSE-NEXT: movq %rcx, %rbx
+; SSE-NEXT: movq %rdx, %r10
+; SSE-NEXT: movq %rsi, %r9
+; SSE-NEXT: movq {{[0-9]+}}(%rsp), %rsi
+; SSE-NEXT: movq {{[0-9]+}}(%rsp), %r11
+; SSE-NEXT: movq {{[0-9]+}}(%rsp), %rdx
+; SSE-NEXT: rep bsfq %rdi, %rax
+; SSE-NEXT: rep bsfq %r9, %r15
+; SSE-NEXT: addl $64, %r15d
+; SSE-NEXT: testq %rdi, %rdi
+; SSE-NEXT: cmovnel %eax, %r15d
+; SSE-NEXT: rep bsfq %r10, %r12
+; SSE-NEXT: rep bsfq %rcx, %rax
+; SSE-NEXT: addl $64, %eax
+; SSE-NEXT: testq %r10, %r10
+; SSE-NEXT: cmovnel %r12d, %eax
+; SSE-NEXT: subl $-128, %eax
+; SSE-NEXT: movq %rdi, %r12
+; SSE-NEXT: orq %r9, %r12
+; SSE-NEXT: cmovnel %r15d, %eax
+; SSE-NEXT: rep bsfq %r8, %r15
+; SSE-NEXT: movq %r13, %rcx
+; SSE-NEXT: movq %r13, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; SSE-NEXT: rep bsfq %r13, %r13
+; SSE-NEXT: addl $64, %r13d
+; SSE-NEXT: testq %r8, %r8
+; SSE-NEXT: cmovnel %r15d, %r13d
+; SSE-NEXT: rep bsfq %rdx, %r12
+; SSE-NEXT: rep bsfq {{[0-9]+}}(%rsp), %r15
+; SSE-NEXT: addl $64, %r15d
+; SSE-NEXT: testq %rdx, %rdx
+; SSE-NEXT: cmovnel %r12d, %r15d
+; SSE-NEXT: movq {{[0-9]+}}(%rsp), %r12
+; SSE-NEXT: subl $-128, %r15d
+; SSE-NEXT: movq %r8, %rbp
+; SSE-NEXT: orq %rcx, %rbp
+; SSE-NEXT: cmovnel %r13d, %r15d
+; SSE-NEXT: addl $256, %r15d # imm = 0x100
+; SSE-NEXT: movq %r9, %r13
+; SSE-NEXT: orq %rbx, %r13
+; SSE-NEXT: movq %rdi, %rbp
+; SSE-NEXT: orq %r10, %rbp
+; SSE-NEXT: orq %r13, %rbp
+; SSE-NEXT: cmovnel %eax, %r15d
+; SSE-NEXT: rep bsfq %r11, %r13
+; SSE-NEXT: rep bsfq %r12, %rax
+; SSE-NEXT: addl $64, %eax
+; SSE-NEXT: testq %r11, %r11
+; SSE-NEXT: cmovnel %r13d, %eax
+; SSE-NEXT: rep bsfq {{[0-9]+}}(%rsp), %r13
+; SSE-NEXT: addl $64, %r13d
+; SSE-NEXT: rep bsfq %rsi, %rcx
+; SSE-NEXT: testq %rsi, %rsi
+; SSE-NEXT: cmovnel %ecx, %r13d
+; SSE-NEXT: subl $-128, %r13d
+; SSE-NEXT: movq %r11, %rcx
+; SSE-NEXT: orq %r12, %rcx
+; SSE-NEXT: cmovnel %eax, %r13d
+; SSE-NEXT: movq {{[0-9]+}}(%rsp), %rbp
+; SSE-NEXT: rep bsfq %rbp, %rcx
+; SSE-NEXT: addl $64, %ecx
+; SSE-NEXT: movq {{[0-9]+}}(%rsp), %rdx
+; SSE-NEXT: rep bsfq %rdx, %rax
+; SSE-NEXT: testq %rdx, %rdx
+; SSE-NEXT: cmovnel %eax, %ecx
+; SSE-NEXT: movl $64, %eax
+; SSE-NEXT: rep bsfq {{[0-9]+}}(%rsp), %rax
+; SSE-NEXT: addl $64, %eax
+; SSE-NEXT: movq {{[0-9]+}}(%rsp), %r8
+; SSE-NEXT: rep bsfq %r8, %rsi
+; SSE-NEXT: testq %r8, %r8
+; SSE-NEXT: cmovnel %esi, %eax
+; SSE-NEXT: subl $-128, %eax
+; SSE-NEXT: orq %rbp, %rdx
+; SSE-NEXT: cmovnel %ecx, %eax
+; SSE-NEXT: orq {{[0-9]+}}(%rsp), %r12
+; SSE-NEXT: orq {{[0-9]+}}(%rsp), %r11
+; SSE-NEXT: addl $256, %eax # imm = 0x100
+; SSE-NEXT: orq %r12, %r11
+; SSE-NEXT: cmovnel %r13d, %eax
+; SSE-NEXT: orq {{[0-9]+}}(%rsp), %rbx
+; SSE-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %r9 # 8-byte Folded Reload
+; SSE-NEXT: orq %rbx, %r9
+; SSE-NEXT: orq {{[0-9]+}}(%rsp), %r10
+; SSE-NEXT: orq %r14, %rdi
+; SSE-NEXT: orq %r10, %rdi
+; SSE-NEXT: addl $512, %eax # imm = 0x200
+; SSE-NEXT: orq %r9, %rdi
+; SSE-NEXT: cmovnel %r15d, %eax
+; SSE-NEXT: # kill: def $eax killed $eax killed $rax
+; SSE-NEXT: popq %rbx
+; SSE-NEXT: popq %r12
+; SSE-NEXT: popq %r13
+; SSE-NEXT: popq %r14
+; SSE-NEXT: popq %r15
+; SSE-NEXT: popq %rbp
+; SSE-NEXT: retq
+;
+; AVX2-LABEL: test_cttz_i1024:
+; AVX2: # %bb.0:
+; AVX2-NEXT: pushq %rbp
+; AVX2-NEXT: pushq %r15
+; AVX2-NEXT: pushq %r14
+; AVX2-NEXT: pushq %r13
+; AVX2-NEXT: pushq %r12
+; AVX2-NEXT: pushq %rbx
+; AVX2-NEXT: movq %r9, %rbx
+; AVX2-NEXT: movq %r8, %r14
+; AVX2-NEXT: movq %rcx, %r11
+; AVX2-NEXT: movq %rdx, %r10
+; AVX2-NEXT: movq %rsi, %r9
+; AVX2-NEXT: movq {{[0-9]+}}(%rsp), %r8
+; AVX2-NEXT: movq {{[0-9]+}}(%rsp), %rsi
+; AVX2-NEXT: movq {{[0-9]+}}(%rsp), %rcx
+; AVX2-NEXT: movq {{[0-9]+}}(%rsp), %rdx
+; AVX2-NEXT: tzcntq %rdi, %rax
+; AVX2-NEXT: xorl %r15d, %r15d
+; AVX2-NEXT: tzcntq %r9, %r15
+; AVX2-NEXT: addl $64, %r15d
+; AVX2-NEXT: testq %rdi, %rdi
+; AVX2-NEXT: cmovnel %eax, %r15d
+; AVX2-NEXT: xorl %r12d, %r12d
+; AVX2-NEXT: tzcntq %r10, %r12
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: tzcntq %r11, %rax
+; AVX2-NEXT: addl $64, %eax
+; AVX2-NEXT: testq %r10, %r10
+; AVX2-NEXT: cmovnel %r12d, %eax
+; AVX2-NEXT: subl $-128, %eax
+; AVX2-NEXT: movq %rdi, %r12
+; AVX2-NEXT: orq %r9, %r12
+; AVX2-NEXT: cmovnel %r15d, %eax
+; AVX2-NEXT: xorl %r15d, %r15d
+; AVX2-NEXT: tzcntq %r14, %r15
+; AVX2-NEXT: movq %rbx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX2-NEXT: xorl %r12d, %r12d
+; AVX2-NEXT: tzcntq %rbx, %r12
+; AVX2-NEXT: addl $64, %r12d
+; AVX2-NEXT: testq %r14, %r14
+; AVX2-NEXT: cmovnel %r15d, %r12d
+; AVX2-NEXT: xorl %r13d, %r13d
+; AVX2-NEXT: tzcntq %rcx, %r13
+; AVX2-NEXT: xorl %r15d, %r15d
+; AVX2-NEXT: tzcntq %rdx, %r15
+; AVX2-NEXT: addl $64, %r15d
+; AVX2-NEXT: testq %rcx, %rcx
+; AVX2-NEXT: cmovnel %r13d, %r15d
+; AVX2-NEXT: subl $-128, %r15d
+; AVX2-NEXT: movq %r14, %r13
+; AVX2-NEXT: orq %rbx, %r13
+; AVX2-NEXT: cmovnel %r12d, %r15d
+; AVX2-NEXT: movq {{[0-9]+}}(%rsp), %r12
+; AVX2-NEXT: addl $256, %r15d # imm = 0x100
+; AVX2-NEXT: movq %r9, %r13
+; AVX2-NEXT: orq %r11, %r13
+; AVX2-NEXT: movq %rdi, %rbp
+; AVX2-NEXT: orq %r10, %rbp
+; AVX2-NEXT: orq %r13, %rbp
+; AVX2-NEXT: movq {{[0-9]+}}(%rsp), %r13
+; AVX2-NEXT: cmovnel %eax, %r15d
+; AVX2-NEXT: xorl %ebp, %ebp
+; AVX2-NEXT: tzcntq %r12, %rbp
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: tzcntq %r13, %rax
+; AVX2-NEXT: addl $64, %eax
+; AVX2-NEXT: testq %r12, %r12
+; AVX2-NEXT: cmovnel %ebp, %eax
+; AVX2-NEXT: xorl %ebp, %ebp
+; AVX2-NEXT: tzcntq %r8, %rbp
+; AVX2-NEXT: addl $64, %ebp
+; AVX2-NEXT: xorl %ecx, %ecx
+; AVX2-NEXT: tzcntq %rsi, %rcx
+; AVX2-NEXT: testq %rsi, %rsi
+; AVX2-NEXT: cmovnel %ecx, %ebp
+; AVX2-NEXT: subl $-128, %ebp
+; AVX2-NEXT: movq %r12, %rcx
+; AVX2-NEXT: orq %r13, %rcx
+; AVX2-NEXT: cmovnel %eax, %ebp
+; AVX2-NEXT: movq {{[0-9]+}}(%rsp), %rbx
+; AVX2-NEXT: xorl %ecx, %ecx
+; AVX2-NEXT: tzcntq %rbx, %rcx
+; AVX2-NEXT: addl $64, %ecx
+; AVX2-NEXT: movq {{[0-9]+}}(%rsp), %rdx
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: tzcntq %rdx, %rax
+; AVX2-NEXT: testq %rdx, %rdx
+; AVX2-NEXT: cmovnel %eax, %ecx
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: tzcntq {{[0-9]+}}(%rsp), %rax
+; AVX2-NEXT: addl $64, %eax
+; AVX2-NEXT: movq {{[0-9]+}}(%rsp), %r8
+; AVX2-NEXT: tzcntq %r8, %rsi
+; AVX2-NEXT: testq %r8, %r8
+; AVX2-NEXT: cmovnel %esi, %eax
+; AVX2-NEXT: subl $-128, %eax
+; AVX2-NEXT: orq %rbx, %rdx
+; AVX2-NEXT: cmovnel %ecx, %eax
+; AVX2-NEXT: orq {{[0-9]+}}(%rsp), %r13
+; AVX2-NEXT: orq {{[0-9]+}}(%rsp), %r12
+; AVX2-NEXT: addl $256, %eax # imm = 0x100
+; AVX2-NEXT: orq %r13, %r12
+; AVX2-NEXT: cmovnel %ebp, %eax
+; AVX2-NEXT: orq {{[0-9]+}}(%rsp), %r11
+; AVX2-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %r9 # 8-byte Folded Reload
+; AVX2-NEXT: orq %r11, %r9
+; AVX2-NEXT: orq {{[0-9]+}}(%rsp), %r10
+; AVX2-NEXT: orq %r14, %rdi
+; AVX2-NEXT: orq %r10, %rdi
+; AVX2-NEXT: addl $512, %eax # imm = 0x200
+; AVX2-NEXT: orq %r9, %rdi
+; AVX2-NEXT: cmovnel %r15d, %eax
+; AVX2-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX2-NEXT: popq %rbx
+; AVX2-NEXT: popq %r12
+; AVX2-NEXT: popq %r13
+; AVX2-NEXT: popq %r14
+; AVX2-NEXT: popq %r15
+; AVX2-NEXT: popq %rbp
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: test_cttz_i1024:
+; AVX512: # %bb.0:
+; AVX512-NEXT: pushq %rbp
+; AVX512-NEXT: pushq %r15
+; AVX512-NEXT: pushq %r14
+; AVX512-NEXT: pushq %r13
+; AVX512-NEXT: pushq %r12
+; AVX512-NEXT: pushq %rbx
+; AVX512-NEXT: movq %r9, %r14
+; AVX512-NEXT: movq %r8, %r15
+; AVX512-NEXT: movq %rcx, %r11
+; AVX512-NEXT: movq %rdx, %r10
+; AVX512-NEXT: movq %rsi, %r9
+; AVX512-NEXT: movq {{[0-9]+}}(%rsp), %rsi
+; AVX512-NEXT: movq {{[0-9]+}}(%rsp), %rbx
+; AVX512-NEXT: movq {{[0-9]+}}(%rsp), %rcx
+; AVX512-NEXT: tzcntq %rdi, %rax
+; AVX512-NEXT: tzcntq %r9, %r12
+; AVX512-NEXT: addl $64, %r12d
+; AVX512-NEXT: testq %rdi, %rdi
+; AVX512-NEXT: cmovnel %eax, %r12d
+; AVX512-NEXT: tzcntq %rdx, %r13
+; AVX512-NEXT: tzcntq %r11, %rax
+; AVX512-NEXT: addl $64, %eax
+; AVX512-NEXT: testq %rdx, %rdx
+; AVX512-NEXT: cmovnel %r13d, %eax
+; AVX512-NEXT: subl $-128, %eax
+; AVX512-NEXT: movq %rdi, %r13
+; AVX512-NEXT: orq %r9, %r13
+; AVX512-NEXT: cmovnel %r12d, %eax
+; AVX512-NEXT: tzcntq %r8, %r12
+; AVX512-NEXT: movq %r14, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX512-NEXT: tzcntq %r14, %r13
+; AVX512-NEXT: addl $64, %r13d
+; AVX512-NEXT: testq %r8, %r8
+; AVX512-NEXT: cmovnel %r12d, %r13d
+; AVX512-NEXT: tzcntq %rcx, %rbp
+; AVX512-NEXT: tzcntq {{[0-9]+}}(%rsp), %r12
+; AVX512-NEXT: addl $64, %r12d
+; AVX512-NEXT: testq %rcx, %rcx
+; AVX512-NEXT: cmovnel %ebp, %r12d
+; AVX512-NEXT: subl $-128, %r12d
+; AVX512-NEXT: movq %r8, %rbp
+; AVX512-NEXT: orq %r14, %rbp
+; AVX512-NEXT: cmovnel %r13d, %r12d
+; AVX512-NEXT: addl $256, %r12d # imm = 0x100
+; AVX512-NEXT: movq %r9, %r13
+; AVX512-NEXT: orq %r11, %r13
+; AVX512-NEXT: movq %rdi, %rbp
+; AVX512-NEXT: orq %rdx, %rbp
+; AVX512-NEXT: orq %r13, %rbp
+; AVX512-NEXT: movq {{[0-9]+}}(%rsp), %r13
+; AVX512-NEXT: cmovnel %eax, %r12d
+; AVX512-NEXT: tzcntq %rbx, %rbp
+; AVX512-NEXT: tzcntq %r13, %rax
+; AVX512-NEXT: addl $64, %eax
+; AVX512-NEXT: testq %rbx, %rbx
+; AVX512-NEXT: cmovnel %ebp, %eax
+; AVX512-NEXT: tzcntq {{[0-9]+}}(%rsp), %rbp
+; AVX512-NEXT: addl $64, %ebp
+; AVX512-NEXT: tzcntq %rsi, %rcx
+; AVX512-NEXT: testq %rsi, %rsi
+; AVX512-NEXT: cmovnel %ecx, %ebp
+; AVX512-NEXT: subl $-128, %ebp
+; AVX512-NEXT: movq %rbx, %rcx
+; AVX512-NEXT: orq %r13, %rcx
+; AVX512-NEXT: cmovnel %eax, %ebp
+; AVX512-NEXT: movq {{[0-9]+}}(%rsp), %r14
+; AVX512-NEXT: tzcntq %r14, %rcx
+; AVX512-NEXT: addl $64, %ecx
+; AVX512-NEXT: movq {{[0-9]+}}(%rsp), %rdx
+; AVX512-NEXT: tzcntq %rdx, %rax
+; AVX512-NEXT: testq %rdx, %rdx
+; AVX512-NEXT: cmovnel %eax, %ecx
+; AVX512-NEXT: tzcntq {{[0-9]+}}(%rsp), %rax
+; AVX512-NEXT: addl $64, %eax
+; AVX512-NEXT: movq {{[0-9]+}}(%rsp), %r8
+; AVX512-NEXT: tzcntq %r8, %rsi
+; AVX512-NEXT: testq %r8, %r8
+; AVX512-NEXT: cmovnel %esi, %eax
+; AVX512-NEXT: subl $-128, %eax
+; AVX512-NEXT: orq %r14, %rdx
+; AVX512-NEXT: cmovnel %ecx, %eax
+; AVX512-NEXT: orq {{[0-9]+}}(%rsp), %r13
+; AVX512-NEXT: orq {{[0-9]+}}(%rsp), %rbx
+; AVX512-NEXT: addl $256, %eax # imm = 0x100
+; AVX512-NEXT: orq %r13, %rbx
+; AVX512-NEXT: cmovnel %ebp, %eax
+; AVX512-NEXT: orq {{[0-9]+}}(%rsp), %r11
+; AVX512-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %r9 # 8-byte Folded Reload
+; AVX512-NEXT: orq %r11, %r9
+; AVX512-NEXT: orq {{[0-9]+}}(%rsp), %r10
+; AVX512-NEXT: orq %r15, %rdi
+; AVX512-NEXT: orq %r10, %rdi
+; AVX512-NEXT: addl $512, %eax # imm = 0x200
+; AVX512-NEXT: orq %r9, %rdi
+; AVX512-NEXT: cmovnel %r12d, %eax
+; AVX512-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX512-NEXT: popq %rbx
+; AVX512-NEXT: popq %r12
+; AVX512-NEXT: popq %r13
+; AVX512-NEXT: popq %r14
+; AVX512-NEXT: popq %r15
+; AVX512-NEXT: popq %rbp
+; AVX512-NEXT: retq
+ %cnt = call i1024 @llvm.cttz.i1024(i1024 %a0, i1 0)
+ %res = trunc i1024 %cnt to i32
+ ret i32 %res
+}
+
+define i32 @load_cttz_i1024(ptr %p0) nounwind {
+; SSE-LABEL: load_cttz_i1024:
+; SSE: # %bb.0:
+; SSE-NEXT: pushq %rbp
+; SSE-NEXT: pushq %r15
+; SSE-NEXT: pushq %r14
+; SSE-NEXT: pushq %r13
+; SSE-NEXT: pushq %r12
+; SSE-NEXT: pushq %rbx
+; SSE-NEXT: movq 88(%rdi), %r10
+; SSE-NEXT: movq %r10, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; SSE-NEXT: movq 56(%rdi), %rcx
+; SSE-NEXT: movq %rcx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; SSE-NEXT: movq 40(%rdi), %rsi
+; SSE-NEXT: movq %rsi, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; SSE-NEXT: movq 24(%rdi), %r9
+; SSE-NEXT: movq 16(%rdi), %r15
+; SSE-NEXT: movq (%rdi), %r8
+; SSE-NEXT: movq 8(%rdi), %r11
+; SSE-NEXT: rep bsfq %r8, %rax
+; SSE-NEXT: rep bsfq %r11, %rdx
+; SSE-NEXT: addl $64, %edx
+; SSE-NEXT: testq %r8, %r8
+; SSE-NEXT: cmovnel %eax, %edx
+; SSE-NEXT: rep bsfq %r15, %rbx
+; SSE-NEXT: rep bsfq %r9, %rax
+; SSE-NEXT: movq %r9, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; SSE-NEXT: addl $64, %eax
+; SSE-NEXT: testq %r15, %r15
+; SSE-NEXT: cmovnel %ebx, %eax
+; SSE-NEXT: movq 32(%rdi), %rbx
+; SSE-NEXT: subl $-128, %eax
+; SSE-NEXT: movq %r8, %r14
+; SSE-NEXT: orq %r11, %r14
+; SSE-NEXT: cmovnel %edx, %eax
+; SSE-NEXT: rep bsfq %rbx, %rdx
+; SSE-NEXT: rep bsfq %rsi, %r12
+; SSE-NEXT: addl $64, %r12d
+; SSE-NEXT: testq %rbx, %rbx
+; SSE-NEXT: cmovnel %edx, %r12d
+; SSE-NEXT: movq 48(%rdi), %r13
+; SSE-NEXT: movq %r13, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; SSE-NEXT: rep bsfq %r13, %rdx
+; SSE-NEXT: rep bsfq %rcx, %r14
+; SSE-NEXT: addl $64, %r14d
+; SSE-NEXT: testq %r13, %r13
+; SSE-NEXT: cmovnel %edx, %r14d
+; SSE-NEXT: subl $-128, %r14d
+; SSE-NEXT: movq %rbx, %rdx
+; SSE-NEXT: orq %rsi, %rdx
+; SSE-NEXT: cmovnel %r12d, %r14d
+; SSE-NEXT: movq 72(%rdi), %r12
+; SSE-NEXT: addl $256, %r14d # imm = 0x100
+; SSE-NEXT: movq %r11, %rdx
+; SSE-NEXT: orq %r9, %rdx
+; SSE-NEXT: movq %r8, %r13
+; SSE-NEXT: orq %r15, %r13
+; SSE-NEXT: orq %rdx, %r13
+; SSE-NEXT: movq 64(%rdi), %r13
+; SSE-NEXT: cmovnel %eax, %r14d
+; SSE-NEXT: rep bsfq %r13, %rdx
+; SSE-NEXT: rep bsfq %r12, %rax
+; SSE-NEXT: addl $64, %eax
+; SSE-NEXT: testq %r13, %r13
+; SSE-NEXT: cmovnel %edx, %eax
+; SSE-NEXT: rep bsfq %r10, %rbp
+; SSE-NEXT: addl $64, %ebp
+; SSE-NEXT: movq 80(%rdi), %r10
+; SSE-NEXT: rep bsfq %r10, %rcx
+; SSE-NEXT: testq %r10, %r10
+; SSE-NEXT: cmovnel %ecx, %ebp
+; SSE-NEXT: subl $-128, %ebp
+; SSE-NEXT: movq %r13, %rcx
+; SSE-NEXT: orq %r12, %rcx
+; SSE-NEXT: cmovnel %eax, %ebp
+; SSE-NEXT: movq 104(%rdi), %r9
+; SSE-NEXT: rep bsfq %r9, %rcx
+; SSE-NEXT: addl $64, %ecx
+; SSE-NEXT: movq 96(%rdi), %rdx
+; SSE-NEXT: rep bsfq %rdx, %rax
+; SSE-NEXT: testq %rdx, %rdx
+; SSE-NEXT: cmovnel %eax, %ecx
+; SSE-NEXT: movl $64, %eax
+; SSE-NEXT: rep bsfq 120(%rdi), %rax
+; SSE-NEXT: movq 112(%rdi), %rdi
+; SSE-NEXT: addl $64, %eax
+; SSE-NEXT: rep bsfq %rdi, %rsi
+; SSE-NEXT: testq %rdi, %rdi
+; SSE-NEXT: cmovnel %esi, %eax
+; SSE-NEXT: subl $-128, %eax
+; SSE-NEXT: orq %r9, %rdx
+; SSE-NEXT: cmovnel %ecx, %eax
+; SSE-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %r12 # 8-byte Folded Reload
+; SSE-NEXT: orq %r10, %r13
+; SSE-NEXT: addl $256, %eax # imm = 0x100
+; SSE-NEXT: orq %r12, %r13
+; SSE-NEXT: cmovnel %ebp, %eax
+; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rcx # 8-byte Reload
+; SSE-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %rcx # 8-byte Folded Reload
+; SSE-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %r11 # 8-byte Folded Reload
+; SSE-NEXT: orq %rcx, %r11
+; SSE-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Folded Reload
+; SSE-NEXT: orq %rbx, %r8
+; SSE-NEXT: orq %r15, %r8
+; SSE-NEXT: addl $512, %eax # imm = 0x200
+; SSE-NEXT: orq %r11, %r8
+; SSE-NEXT: cmovnel %r14d, %eax
+; SSE-NEXT: # kill: def $eax killed $eax killed $rax
+; SSE-NEXT: popq %rbx
+; SSE-NEXT: popq %r12
+; SSE-NEXT: popq %r13
+; SSE-NEXT: popq %r14
+; SSE-NEXT: popq %r15
+; SSE-NEXT: popq %rbp
+; SSE-NEXT: retq
+;
+; AVX2-LABEL: load_cttz_i1024:
+; AVX2: # %bb.0:
+; AVX2-NEXT: pushq %rbp
+; AVX2-NEXT: pushq %r15
+; AVX2-NEXT: pushq %r14
+; AVX2-NEXT: pushq %r13
+; AVX2-NEXT: pushq %r12
+; AVX2-NEXT: pushq %rbx
+; AVX2-NEXT: movq 72(%rdi), %r14
+; AVX2-NEXT: movq 64(%rdi), %r15
+; AVX2-NEXT: movq 56(%rdi), %r9
+; AVX2-NEXT: movq %r9, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX2-NEXT: movq 48(%rdi), %rcx
+; AVX2-NEXT: movq 40(%rdi), %r10
+; AVX2-NEXT: movq %r10, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX2-NEXT: movq 32(%rdi), %rsi
+; AVX2-NEXT: movq 24(%rdi), %rbp
+; AVX2-NEXT: movq 16(%rdi), %rbx
+; AVX2-NEXT: movq (%rdi), %r8
+; AVX2-NEXT: movq 8(%rdi), %r11
+; AVX2-NEXT: tzcntq %r8, %rax
+; AVX2-NEXT: tzcntq %r11, %rdx
+; AVX2-NEXT: addl $64, %edx
+; AVX2-NEXT: testq %r8, %r8
+; AVX2-NEXT: cmovnel %eax, %edx
+; AVX2-NEXT: xorl %r12d, %r12d
+; AVX2-NEXT: tzcntq %rbx, %r12
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: tzcntq %rbp, %rax
+; AVX2-NEXT: movq %rbp, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX2-NEXT: addl $64, %eax
+; AVX2-NEXT: testq %rbx, %rbx
+; AVX2-NEXT: cmovnel %r12d, %eax
+; AVX2-NEXT: subl $-128, %eax
+; AVX2-NEXT: movq %r8, %r12
+; AVX2-NEXT: orq %r11, %r12
+; AVX2-NEXT: cmovnel %edx, %eax
+; AVX2-NEXT: xorl %edx, %edx
+; AVX2-NEXT: tzcntq %rsi, %rdx
+; AVX2-NEXT: xorl %r13d, %r13d
+; AVX2-NEXT: tzcntq %r10, %r13
+; AVX2-NEXT: addl $64, %r13d
+; AVX2-NEXT: testq %rsi, %rsi
+; AVX2-NEXT: movq %rsi, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX2-NEXT: cmovnel %edx, %r13d
+; AVX2-NEXT: movq %rcx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX2-NEXT: xorl %edx, %edx
+; AVX2-NEXT: tzcntq %rcx, %rdx
+; AVX2-NEXT: xorl %r12d, %r12d
+; AVX2-NEXT: tzcntq %r9, %r12
+; AVX2-NEXT: addl $64, %r12d
+; AVX2-NEXT: testq %rcx, %rcx
+; AVX2-NEXT: cmovnel %edx, %r12d
+; AVX2-NEXT: subl $-128, %r12d
+; AVX2-NEXT: movq %rsi, %rdx
+; AVX2-NEXT: orq %r10, %rdx
+; AVX2-NEXT: cmovnel %r13d, %r12d
+; AVX2-NEXT: addl $256, %r12d # imm = 0x100
+; AVX2-NEXT: movq %r11, %rdx
+; AVX2-NEXT: orq %rbp, %rdx
+; AVX2-NEXT: movq %r8, %r13
+; AVX2-NEXT: orq %rbx, %r13
+; AVX2-NEXT: orq %rdx, %r13
+; AVX2-NEXT: cmovnel %eax, %r12d
+; AVX2-NEXT: xorl %edx, %edx
+; AVX2-NEXT: tzcntq %r15, %rdx
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: tzcntq %r14, %rax
+; AVX2-NEXT: addl $64, %eax
+; AVX2-NEXT: testq %r15, %r15
+; AVX2-NEXT: cmovnel %edx, %eax
+; AVX2-NEXT: movq 88(%rdi), %rbp
+; AVX2-NEXT: xorl %r13d, %r13d
+; AVX2-NEXT: tzcntq %rbp, %r13
+; AVX2-NEXT: addl $64, %r13d
+; AVX2-NEXT: movq 80(%rdi), %r10
+; AVX2-NEXT: xorl %ecx, %ecx
+; AVX2-NEXT: tzcntq %r10, %rcx
+; AVX2-NEXT: testq %r10, %r10
+; AVX2-NEXT: cmovnel %ecx, %r13d
+; AVX2-NEXT: subl $-128, %r13d
+; AVX2-NEXT: movq %r15, %rcx
+; AVX2-NEXT: orq %r14, %rcx
+; AVX2-NEXT: cmovnel %eax, %r13d
+; AVX2-NEXT: movq 104(%rdi), %r9
+; AVX2-NEXT: xorl %ecx, %ecx
+; AVX2-NEXT: tzcntq %r9, %rcx
+; AVX2-NEXT: addl $64, %ecx
+; AVX2-NEXT: movq 96(%rdi), %rdx
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: tzcntq %rdx, %rax
+; AVX2-NEXT: testq %rdx, %rdx
+; AVX2-NEXT: cmovnel %eax, %ecx
+; AVX2-NEXT: movq 112(%rdi), %rsi
+; AVX2-NEXT: xorl %eax, %eax
+; AVX2-NEXT: tzcntq 120(%rdi), %rax
+; AVX2-NEXT: addl $64, %eax
+; AVX2-NEXT: tzcntq %rsi, %rdi
+; AVX2-NEXT: testq %rsi, %rsi
+; AVX2-NEXT: cmovnel %edi, %eax
+; AVX2-NEXT: subl $-128, %eax
+; AVX2-NEXT: orq %r9, %rdx
+; AVX2-NEXT: cmovnel %ecx, %eax
+; AVX2-NEXT: orq %rbp, %r14
+; AVX2-NEXT: orq %r10, %r15
+; AVX2-NEXT: addl $256, %eax # imm = 0x100
+; AVX2-NEXT: orq %r14, %r15
+; AVX2-NEXT: cmovnel %r13d, %eax
+; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rcx # 8-byte Reload
+; AVX2-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %rcx # 8-byte Folded Reload
+; AVX2-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %r11 # 8-byte Folded Reload
+; AVX2-NEXT: orq %rcx, %r11
+; AVX2-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %rbx # 8-byte Folded Reload
+; AVX2-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Folded Reload
+; AVX2-NEXT: orq %rbx, %r8
+; AVX2-NEXT: addl $512, %eax # imm = 0x200
+; AVX2-NEXT: orq %r11, %r8
+; AVX2-NEXT: cmovnel %r12d, %eax
+; AVX2-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX2-NEXT: popq %rbx
+; AVX2-NEXT: popq %r12
+; AVX2-NEXT: popq %r13
+; AVX2-NEXT: popq %r14
+; AVX2-NEXT: popq %r15
+; AVX2-NEXT: popq %rbp
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: load_cttz_i1024:
+; AVX512: # %bb.0:
+; AVX512-NEXT: pushq %rbp
+; AVX512-NEXT: pushq %r15
+; AVX512-NEXT: pushq %r14
+; AVX512-NEXT: pushq %r13
+; AVX512-NEXT: pushq %r12
+; AVX512-NEXT: pushq %rbx
+; AVX512-NEXT: movq 88(%rdi), %rbp
+; AVX512-NEXT: movq 72(%rdi), %r15
+; AVX512-NEXT: movq 56(%rdi), %r9
+; AVX512-NEXT: movq %r9, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX512-NEXT: movq 48(%rdi), %rcx
+; AVX512-NEXT: movq 40(%rdi), %r10
+; AVX512-NEXT: movq %r10, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX512-NEXT: movq 32(%rdi), %rsi
+; AVX512-NEXT: movq 24(%rdi), %r14
+; AVX512-NEXT: movq 16(%rdi), %rbx
+; AVX512-NEXT: movq (%rdi), %r8
+; AVX512-NEXT: movq 8(%rdi), %r11
+; AVX512-NEXT: tzcntq %r8, %rax
+; AVX512-NEXT: tzcntq %r11, %rdx
+; AVX512-NEXT: addl $64, %edx
+; AVX512-NEXT: testq %r8, %r8
+; AVX512-NEXT: cmovnel %eax, %edx
+; AVX512-NEXT: tzcntq %rbx, %r12
+; AVX512-NEXT: tzcntq %r14, %rax
+; AVX512-NEXT: movq %r14, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX512-NEXT: addl $64, %eax
+; AVX512-NEXT: testq %rbx, %rbx
+; AVX512-NEXT: cmovnel %r12d, %eax
+; AVX512-NEXT: subl $-128, %eax
+; AVX512-NEXT: movq %r8, %r12
+; AVX512-NEXT: orq %r11, %r12
+; AVX512-NEXT: cmovnel %edx, %eax
+; AVX512-NEXT: tzcntq %rsi, %rdx
+; AVX512-NEXT: tzcntq %r10, %r13
+; AVX512-NEXT: addl $64, %r13d
+; AVX512-NEXT: testq %rsi, %rsi
+; AVX512-NEXT: movq %rsi, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX512-NEXT: cmovnel %edx, %r13d
+; AVX512-NEXT: movq %rcx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX512-NEXT: tzcntq %rcx, %rdx
+; AVX512-NEXT: tzcntq %r9, %r12
+; AVX512-NEXT: addl $64, %r12d
+; AVX512-NEXT: testq %rcx, %rcx
+; AVX512-NEXT: cmovnel %edx, %r12d
+; AVX512-NEXT: subl $-128, %r12d
+; AVX512-NEXT: movq %rsi, %rdx
+; AVX512-NEXT: orq %r10, %rdx
+; AVX512-NEXT: cmovnel %r13d, %r12d
+; AVX512-NEXT: addl $256, %r12d # imm = 0x100
+; AVX512-NEXT: movq %r11, %rdx
+; AVX512-NEXT: orq %r14, %rdx
+; AVX512-NEXT: movq %r8, %r13
+; AVX512-NEXT: orq %rbx, %r13
+; AVX512-NEXT: orq %rdx, %r13
+; AVX512-NEXT: movq 64(%rdi), %r13
+; AVX512-NEXT: cmovnel %eax, %r12d
+; AVX512-NEXT: tzcntq %r13, %rdx
+; AVX512-NEXT: tzcntq %r15, %rax
+; AVX512-NEXT: addl $64, %eax
+; AVX512-NEXT: testq %r13, %r13
+; AVX512-NEXT: cmovnel %edx, %eax
+; AVX512-NEXT: movq %rbp, %r14
+; AVX512-NEXT: tzcntq %rbp, %rbp
+; AVX512-NEXT: addl $64, %ebp
+; AVX512-NEXT: movq 80(%rdi), %r10
+; AVX512-NEXT: tzcntq %r10, %rcx
+; AVX512-NEXT: testq %r10, %r10
+; AVX512-NEXT: cmovnel %ecx, %ebp
+; AVX512-NEXT: subl $-128, %ebp
+; AVX512-NEXT: movq %r13, %rcx
+; AVX512-NEXT: orq %r15, %rcx
+; AVX512-NEXT: cmovnel %eax, %ebp
+; AVX512-NEXT: movq 104(%rdi), %r9
+; AVX512-NEXT: tzcntq %r9, %rcx
+; AVX512-NEXT: addl $64, %ecx
+; AVX512-NEXT: movq 96(%rdi), %rdx
+; AVX512-NEXT: tzcntq %rdx, %rax
+; AVX512-NEXT: testq %rdx, %rdx
+; AVX512-NEXT: cmovnel %eax, %ecx
+; AVX512-NEXT: movq 112(%rdi), %rsi
+; AVX512-NEXT: tzcntq 120(%rdi), %rax
+; AVX512-NEXT: addl $64, %eax
+; AVX512-NEXT: tzcntq %rsi, %rdi
+; AVX512-NEXT: testq %rsi, %rsi
+; AVX512-NEXT: cmovnel %edi, %eax
+; AVX512-NEXT: subl $-128, %eax
+; AVX512-NEXT: orq %r9, %rdx
+; AVX512-NEXT: cmovnel %ecx, %eax
+; AVX512-NEXT: orq %r14, %r15
+; AVX512-NEXT: orq %r10, %r13
+; AVX512-NEXT: addl $256, %eax # imm = 0x100
+; AVX512-NEXT: orq %r15, %r13
+; AVX512-NEXT: cmovnel %ebp, %eax
+; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rcx # 8-byte Reload
+; AVX512-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %rcx # 8-byte Folded Reload
+; AVX512-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %r11 # 8-byte Folded Reload
+; AVX512-NEXT: orq %rcx, %r11
+; AVX512-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %rbx # 8-byte Folded Reload
+; AVX512-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Folded Reload
+; AVX512-NEXT: orq %rbx, %r8
+; AVX512-NEXT: addl $512, %eax # imm = 0x200
+; AVX512-NEXT: orq %r11, %r8
+; AVX512-NEXT: cmovnel %r12d, %eax
+; AVX512-NEXT: # kill: def $eax killed $eax killed $rax
+; AVX512-NEXT: popq %rbx
+; AVX512-NEXT: popq %r12
+; AVX512-NEXT: popq %r13
+; AVX512-NEXT: popq %r14
+; AVX512-NEXT: popq %r15
+; AVX512-NEXT: popq %rbp
+; AVX512-NEXT: retq
+ %a0 = load i1024, ptr %p0
+ %cnt = call i1024 @llvm.cttz.i1024(i1024 %a0, i1 0)
+ %res = trunc i1024 %cnt to i32
+ ret i32 %res
+}
More information about the llvm-commits
mailing list