derek.holloway
7502bf64a4
CodeQL / Analyze (actions) (push) Canceled after 0s
CodeQL / Analyze (python) (push) Canceled after 0s
pre-commit / pre-commit (push) Canceled after 0s
Multi-Arch CI / setup (push) Canceled after 0s
Publish build_manylinux_x86_64 images / publish_build_manylinux_x86_64 (push) Canceled after 0s
Setup (Multi-Arch) / setup (push) Canceled after 0s
Multi-Arch CI / Linux::${{ fromJSON(needs.setup.outputs.linux_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubi10 image / publish_no_rocm_image_ubi10 (push) Canceled after 0s
Multi-Arch CI / Windows::${{ fromJSON(needs.setup.outputs.windows_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Multi-Arch CI / Manifest Diff (push) Canceled after 0s
TheRock Manifest Diff Report / Generate Manifest Diff Report (push) Canceled after 0s
Multi-Arch CI / CI Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_ocl_rt images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_rocgdb images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Unit Tests / Unit Tests :: ubuntu-24.04 (push) Canceled after 0s
Unit Tests / Unit Tests :: windows-2022 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_media (push) Canceled after 0s
Unit Tests / Unit Tests Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_openmpi (push) Canceled after 0s
215 lines
6.8 KiB
LLVM
215 lines
6.8 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
|
|
; RUN: llc < %s -mtriple=x86_64-- -mcpu=x86-64-v2 | FileCheck %s --check-prefixes=CHECK,NOBMI
|
|
; RUN: llc < %s -mtriple=x86_64-- -mcpu=x86-64-v3 | FileCheck %s --check-prefixes=CHECK,BMI
|
|
|
|
define i64 @test_i64(i64 %w, i64 %x, i64 %y, i64 %z) {
|
|
; NOBMI-LABEL: test_i64:
|
|
; NOBMI: # %bb.0: # %Entry
|
|
; NOBMI-NEXT: movq %rcx, %rax
|
|
; NOBMI-NEXT: andq %rdx, %rsi
|
|
; NOBMI-NEXT: notq %rsi
|
|
; NOBMI-NEXT: andq %rdi, %rsi
|
|
; NOBMI-NEXT: notq %rax
|
|
; NOBMI-NEXT: orq %rdx, %rax
|
|
; NOBMI-NEXT: andq %rsi, %rax
|
|
; NOBMI-NEXT: retq
|
|
;
|
|
; BMI-LABEL: test_i64:
|
|
; BMI: # %bb.0: # %Entry
|
|
; BMI-NEXT: andq %rdx, %rsi
|
|
; BMI-NEXT: andnq %rdi, %rsi, %rax
|
|
; BMI-NEXT: andnq %rcx, %rdx, %rcx
|
|
; BMI-NEXT: andnq %rax, %rcx, %rax
|
|
; BMI-NEXT: retq
|
|
Entry:
|
|
%and1 = and i64 %y, %x
|
|
%xor1 = xor i64 %and1, -1
|
|
%and2 = and i64 %xor1, %w
|
|
%.not = xor i64 %z, -1
|
|
%or1 = or i64 %.not, %y
|
|
%and3 = and i64 %and2, %or1
|
|
ret i64 %and3
|
|
}
|
|
|
|
define i32 @test_i32(i32 %w, i32 %x, i32 %y, i32 %z) {
|
|
; NOBMI-LABEL: test_i32:
|
|
; NOBMI: # %bb.0: # %Entry
|
|
; NOBMI-NEXT: movl %ecx, %eax
|
|
; NOBMI-NEXT: andl %edx, %esi
|
|
; NOBMI-NEXT: notl %esi
|
|
; NOBMI-NEXT: andl %edi, %esi
|
|
; NOBMI-NEXT: notl %eax
|
|
; NOBMI-NEXT: orl %edx, %eax
|
|
; NOBMI-NEXT: andl %esi, %eax
|
|
; NOBMI-NEXT: retq
|
|
;
|
|
; BMI-LABEL: test_i32:
|
|
; BMI: # %bb.0: # %Entry
|
|
; BMI-NEXT: andl %edx, %esi
|
|
; BMI-NEXT: andnl %edi, %esi, %eax
|
|
; BMI-NEXT: andnl %ecx, %edx, %ecx
|
|
; BMI-NEXT: andnl %eax, %ecx, %eax
|
|
; BMI-NEXT: retq
|
|
Entry:
|
|
%and1 = and i32 %y, %x
|
|
%xor1 = xor i32 %and1, -1
|
|
%and2 = and i32 %xor1, %w
|
|
%.not = xor i32 %z, -1
|
|
%or1 = or i32 %.not, %y
|
|
%and3 = and i32 %and2, %or1
|
|
ret i32 %and3
|
|
}
|
|
|
|
define i16 @test_i16(i16 %w, i16 %x, i16 %y, i16 %z) {
|
|
; NOBMI-LABEL: test_i16:
|
|
; NOBMI: # %bb.0: # %Entry
|
|
; NOBMI-NEXT: movl %ecx, %eax
|
|
; NOBMI-NEXT: andl %edx, %esi
|
|
; NOBMI-NEXT: notl %esi
|
|
; NOBMI-NEXT: andl %edi, %esi
|
|
; NOBMI-NEXT: notl %eax
|
|
; NOBMI-NEXT: orl %edx, %eax
|
|
; NOBMI-NEXT: andl %esi, %eax
|
|
; NOBMI-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; NOBMI-NEXT: retq
|
|
;
|
|
; BMI-LABEL: test_i16:
|
|
; BMI: # %bb.0: # %Entry
|
|
; BMI-NEXT: andl %edx, %esi
|
|
; BMI-NEXT: andnl %edi, %esi, %eax
|
|
; BMI-NEXT: notl %ecx
|
|
; BMI-NEXT: orl %edx, %ecx
|
|
; BMI-NEXT: andl %ecx, %eax
|
|
; BMI-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; BMI-NEXT: retq
|
|
Entry:
|
|
%and1 = and i16 %y, %x
|
|
%xor1 = xor i16 %and1, -1
|
|
%and2 = and i16 %xor1, %w
|
|
%.not = xor i16 %z, -1
|
|
%or1 = or i16 %.not, %y
|
|
%and3 = and i16 %and2, %or1
|
|
ret i16 %and3
|
|
}
|
|
|
|
define i8 @test_i8(i8 %w, i8 %x, i8 %y, i8 %z) {
|
|
; CHECK-LABEL: test_i8:
|
|
; CHECK: # %bb.0: # %Entry
|
|
; CHECK-NEXT: movl %edx, %eax
|
|
; CHECK-NEXT: andl %edx, %esi
|
|
; CHECK-NEXT: notb %sil
|
|
; CHECK-NEXT: andb %dil, %sil
|
|
; CHECK-NEXT: notb %cl
|
|
; CHECK-NEXT: orb %cl, %al
|
|
; CHECK-NEXT: andb %sil, %al
|
|
; CHECK-NEXT: # kill: def $al killed $al killed $eax
|
|
; CHECK-NEXT: retq
|
|
Entry:
|
|
%and1 = and i8 %y, %x
|
|
%xor1 = xor i8 %and1, -1
|
|
%and2 = and i8 %xor1, %w
|
|
%.not = xor i8 %z, -1
|
|
%or1 = or i8 %.not, %y
|
|
%and3 = and i8 %and2, %or1
|
|
ret i8 %and3
|
|
}
|
|
|
|
define <16 x i8> @test_v16i8(<16 x i8> %w, <16 x i8> %x, <16 x i8> %y, <16 x i8> %z) {
|
|
; NOBMI-LABEL: test_v16i8:
|
|
; NOBMI: # %bb.0: # %Entry
|
|
; NOBMI-NEXT: andps %xmm2, %xmm1
|
|
; NOBMI-NEXT: andnps %xmm0, %xmm1
|
|
; NOBMI-NEXT: andnps %xmm3, %xmm2
|
|
; NOBMI-NEXT: andnps %xmm1, %xmm2
|
|
; NOBMI-NEXT: movaps %xmm2, %xmm0
|
|
; NOBMI-NEXT: retq
|
|
;
|
|
; BMI-LABEL: test_v16i8:
|
|
; BMI: # %bb.0: # %Entry
|
|
; BMI-NEXT: vandps %xmm1, %xmm2, %xmm1
|
|
; BMI-NEXT: vandnps %xmm0, %xmm1, %xmm0
|
|
; BMI-NEXT: vandnps %xmm3, %xmm2, %xmm1
|
|
; BMI-NEXT: vandnps %xmm0, %xmm1, %xmm0
|
|
; BMI-NEXT: retq
|
|
Entry:
|
|
%and1 = and <16 x i8> %y, %x
|
|
%xor1 = xor <16 x i8> %and1, <i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1>
|
|
%and2 = and <16 x i8> %xor1, %w
|
|
%.not = xor <16 x i8> %z, <i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1>
|
|
%or1 = or <16 x i8> %.not, %y
|
|
%and3 = and <16 x i8> %and2, %or1
|
|
ret <16 x i8> %and3
|
|
}
|
|
|
|
define <32 x i8> @test_v32i8(<32 x i8> %w, <32 x i8> %x, <32 x i8> %y, <32 x i8> %z) {
|
|
; NOBMI-LABEL: test_v32i8:
|
|
; NOBMI: # %bb.0: # %Entry
|
|
; NOBMI-NEXT: andps %xmm4, %xmm2
|
|
; NOBMI-NEXT: andps %xmm5, %xmm3
|
|
; NOBMI-NEXT: andnps %xmm1, %xmm3
|
|
; NOBMI-NEXT: andnps %xmm0, %xmm2
|
|
; NOBMI-NEXT: andnps %xmm6, %xmm4
|
|
; NOBMI-NEXT: andnps %xmm2, %xmm4
|
|
; NOBMI-NEXT: andnps %xmm7, %xmm5
|
|
; NOBMI-NEXT: andnps %xmm3, %xmm5
|
|
; NOBMI-NEXT: movaps %xmm4, %xmm0
|
|
; NOBMI-NEXT: movaps %xmm5, %xmm1
|
|
; NOBMI-NEXT: retq
|
|
;
|
|
; BMI-LABEL: test_v32i8:
|
|
; BMI: # %bb.0: # %Entry
|
|
; BMI-NEXT: vandps %ymm1, %ymm2, %ymm1
|
|
; BMI-NEXT: vandnps %ymm0, %ymm1, %ymm0
|
|
; BMI-NEXT: vandnps %ymm3, %ymm2, %ymm1
|
|
; BMI-NEXT: vandnps %ymm0, %ymm1, %ymm0
|
|
; BMI-NEXT: retq
|
|
Entry:
|
|
%and1 = and <32 x i8> %y, %x
|
|
%xor1 = xor <32 x i8> %and1, <i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1>
|
|
%and2 = and <32 x i8> %xor1, %w
|
|
%.not = xor <32 x i8> %z, <i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1>
|
|
%or1 = or <32 x i8> %.not, %y
|
|
%and3 = and <32 x i8> %and2, %or1
|
|
ret <32 x i8> %and3
|
|
}
|
|
|
|
; PR112347 - don't fold if we'd be inverting a constant, as demorgan normalisation will invert it back again.
|
|
define void @PR112347(ptr %p0, ptr %p1, ptr %p2) {
|
|
; CHECK-LABEL: PR112347:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: movl (%rdi), %eax
|
|
; CHECK-NEXT: notl %eax
|
|
; CHECK-NEXT: orl $-16777204, %eax # imm = 0xFF00000C
|
|
; CHECK-NEXT: andl (%rsi), %eax
|
|
; CHECK-NEXT: movl %eax, (%rdx)
|
|
; CHECK-NEXT: retq
|
|
%load0 = load i32, ptr %p0, align 1
|
|
%load1 = load i32, ptr %p1, align 4
|
|
%not = xor i32 %load0, -1
|
|
%top = or i32 %not, -16777204
|
|
%mask = and i32 %load1, %top
|
|
store i32 %mask, ptr %p2, align 4
|
|
ret void
|
|
}
|
|
|
|
define void @PR113240(i64 %a) {
|
|
; CHECK-LABEL: PR113240:
|
|
; CHECK: # %bb.0: # %entry
|
|
; CHECK-NEXT: movq %rdi, %rax
|
|
; CHECK-NEXT: notq %rax
|
|
; CHECK-NEXT: movabsq $8796093022206, %rcx # imm = 0x7FFFFFFFFFE
|
|
; CHECK-NEXT: notq %rcx
|
|
; CHECK-NEXT: orq %rax, %rcx
|
|
; CHECK-NEXT: andq %rdi, %rcx
|
|
; CHECK-NEXT: movq %rcx, 0
|
|
; CHECK-NEXT: retq
|
|
entry:
|
|
%and = and i64 %a, 8796093022206
|
|
%bf.value = and i64 8796093022206, 0
|
|
%not = xor i64 %and, -1
|
|
%and4 = and i64 %a, %not
|
|
store i64 %and4, ptr null, align 8
|
|
ret void
|
|
}
|
|
|