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
143 lines
5.2 KiB
LLVM
143 lines
5.2 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
|
|
; RUN: opt -passes=expand-memcmp -mtriple=wasm32-unknown-unknown -mattr=+simd128 -S < %s | llc -disable-wasm-fallthrough-return-opt -wasm-disable-explicit-locals -wasm-keep-registers -mattr=+simd128 | FileCheck %s
|
|
|
|
target triple = "wasm32-unknown-unknown"
|
|
|
|
declare i32 @memcmp(ptr, ptr, i32)
|
|
|
|
define i1 @memcmp_expand_3(ptr %a, ptr %b) {
|
|
; CHECK-LABEL: memcmp_expand_3:
|
|
; CHECK: .functype memcmp_expand_3 (i32, i32) -> (i32)
|
|
; CHECK-NEXT: # %bb.0:
|
|
; CHECK-NEXT: i32.load16_u $push7=, 0($0):p2align=0
|
|
; CHECK-NEXT: i32.load16_u $push6=, 0($1):p2align=0
|
|
; CHECK-NEXT: i32.xor $push8=, $pop7, $pop6
|
|
; CHECK-NEXT: i32.const $push0=, 2
|
|
; CHECK-NEXT: i32.add $push3=, $0, $pop0
|
|
; CHECK-NEXT: i32.load8_u $push4=, 0($pop3)
|
|
; CHECK-NEXT: i32.const $push13=, 2
|
|
; CHECK-NEXT: i32.add $push1=, $1, $pop13
|
|
; CHECK-NEXT: i32.load8_u $push2=, 0($pop1)
|
|
; CHECK-NEXT: i32.xor $push5=, $pop4, $pop2
|
|
; CHECK-NEXT: i32.or $push9=, $pop8, $pop5
|
|
; CHECK-NEXT: i32.const $push10=, 65535
|
|
; CHECK-NEXT: i32.and $push11=, $pop9, $pop10
|
|
; CHECK-NEXT: i32.eqz $push12=, $pop11
|
|
; CHECK-NEXT: return $pop12
|
|
%cmp_3 = call i32 @memcmp(ptr %a, ptr %b, i32 3)
|
|
%res = icmp eq i32 %cmp_3, 0
|
|
ret i1 %res
|
|
}
|
|
|
|
define i1 @memcmp_expand_5(ptr %a, ptr %b) {
|
|
; CHECK-LABEL: memcmp_expand_5:
|
|
; CHECK: .functype memcmp_expand_5 (i32, i32) -> (i32)
|
|
; CHECK-NEXT: # %bb.0:
|
|
; CHECK-NEXT: i32.load $push7=, 0($0):p2align=0
|
|
; CHECK-NEXT: i32.load $push6=, 0($1):p2align=0
|
|
; CHECK-NEXT: i32.xor $push8=, $pop7, $pop6
|
|
; CHECK-NEXT: i32.const $push0=, 4
|
|
; CHECK-NEXT: i32.add $push3=, $0, $pop0
|
|
; CHECK-NEXT: i32.load8_u $push4=, 0($pop3)
|
|
; CHECK-NEXT: i32.const $push11=, 4
|
|
; CHECK-NEXT: i32.add $push1=, $1, $pop11
|
|
; CHECK-NEXT: i32.load8_u $push2=, 0($pop1)
|
|
; CHECK-NEXT: i32.xor $push5=, $pop4, $pop2
|
|
; CHECK-NEXT: i32.or $push9=, $pop8, $pop5
|
|
; CHECK-NEXT: i32.eqz $push10=, $pop9
|
|
; CHECK-NEXT: return $pop10
|
|
%cmp_5 = call i32 @memcmp(ptr %a, ptr %b, i32 5)
|
|
%res = icmp eq i32 %cmp_5, 0
|
|
ret i1 %res
|
|
}
|
|
|
|
define i1 @memcmp_expand_7(ptr %a, ptr %b) {
|
|
; CHECK-LABEL: memcmp_expand_7:
|
|
; CHECK: .functype memcmp_expand_7 (i32, i32) -> (i32)
|
|
; CHECK-NEXT: # %bb.0:
|
|
; CHECK-NEXT: i32.load $push7=, 0($0):p2align=0
|
|
; CHECK-NEXT: i32.load $push6=, 0($1):p2align=0
|
|
; CHECK-NEXT: i32.xor $push8=, $pop7, $pop6
|
|
; CHECK-NEXT: i32.const $push0=, 3
|
|
; CHECK-NEXT: i32.add $push3=, $0, $pop0
|
|
; CHECK-NEXT: i32.load $push4=, 0($pop3):p2align=0
|
|
; CHECK-NEXT: i32.const $push11=, 3
|
|
; CHECK-NEXT: i32.add $push1=, $1, $pop11
|
|
; CHECK-NEXT: i32.load $push2=, 0($pop1):p2align=0
|
|
; CHECK-NEXT: i32.xor $push5=, $pop4, $pop2
|
|
; CHECK-NEXT: i32.or $push9=, $pop8, $pop5
|
|
; CHECK-NEXT: i32.eqz $push10=, $pop9
|
|
; CHECK-NEXT: return $pop10
|
|
%cmp_7 = call i32 @memcmp(ptr %a, ptr %b, i32 7)
|
|
%res = icmp eq i32 %cmp_7, 0
|
|
ret i1 %res
|
|
}
|
|
|
|
; INFO: Negative test
|
|
; Should not expand even with simd128
|
|
define i1 @memcmp_expand_129(ptr %a, ptr %b) {
|
|
; CHECK-LABEL: memcmp_expand_129:
|
|
; CHECK: .functype memcmp_expand_129 (i32, i32) -> (i32)
|
|
; CHECK-NEXT: # %bb.0:
|
|
; CHECK-NEXT: i32.const $push0=, 129
|
|
; CHECK-NEXT: call $push1=, memcmp, $0, $1, $pop0
|
|
; CHECK-NEXT: i32.eqz $push2=, $pop1
|
|
; CHECK-NEXT: return $pop2
|
|
%cmp_129 = call i32 @memcmp(ptr %a, ptr %b, i32 129)
|
|
%res = icmp eq i32 %cmp_129, 0
|
|
ret i1 %res
|
|
}
|
|
|
|
define i1 @memcmp_expand_2(ptr %a, ptr %b) {
|
|
; CHECK-LABEL: memcmp_expand_2:
|
|
; CHECK: .functype memcmp_expand_2 (i32, i32) -> (i32)
|
|
; CHECK-NEXT: # %bb.0:
|
|
; CHECK-NEXT: i32.load16_u $push1=, 0($0):p2align=0
|
|
; CHECK-NEXT: i32.load16_u $push0=, 0($1):p2align=0
|
|
; CHECK-NEXT: i32.eq $push2=, $pop1, $pop0
|
|
; CHECK-NEXT: return $pop2
|
|
%cmp_2 = call i32 @memcmp(ptr %a, ptr %b, i32 2)
|
|
%res = icmp eq i32 %cmp_2, 0
|
|
ret i1 %res
|
|
}
|
|
|
|
define i1 @memcmp_expand_2_align(ptr align(2) %a, ptr align(2) %b) {
|
|
; CHECK-LABEL: memcmp_expand_2_align:
|
|
; CHECK: .functype memcmp_expand_2_align (i32, i32) -> (i32)
|
|
; CHECK-NEXT: # %bb.0:
|
|
; CHECK-NEXT: i32.load16_u $push1=, 0($0)
|
|
; CHECK-NEXT: i32.load16_u $push0=, 0($1)
|
|
; CHECK-NEXT: i32.eq $push2=, $pop1, $pop0
|
|
; CHECK-NEXT: return $pop2
|
|
%cmp_2 = call i32 @memcmp(ptr %a, ptr %b, i32 2)
|
|
%res = icmp eq i32 %cmp_2, 0
|
|
ret i1 %res
|
|
}
|
|
|
|
define i1 @memcmp_expand_8(ptr %a, ptr %b) {
|
|
; CHECK-LABEL: memcmp_expand_8:
|
|
; CHECK: .functype memcmp_expand_8 (i32, i32) -> (i32)
|
|
; CHECK-NEXT: # %bb.0:
|
|
; CHECK-NEXT: i64.load $push1=, 0($0):p2align=0
|
|
; CHECK-NEXT: i64.load $push0=, 0($1):p2align=0
|
|
; CHECK-NEXT: i64.eq $push2=, $pop1, $pop0
|
|
; CHECK-NEXT: return $pop2
|
|
%cmp_8 = call i32 @memcmp(ptr %a, ptr %b, i32 8)
|
|
%res = icmp eq i32 %cmp_8, 0
|
|
ret i1 %res
|
|
}
|
|
|
|
define i1 @memcmp_expand_16(ptr %a, ptr %b) {
|
|
; CHECK-LABEL: memcmp_expand_16:
|
|
; CHECK: .functype memcmp_expand_16 (i32, i32) -> (i32)
|
|
; CHECK-NEXT: # %bb.0:
|
|
; CHECK-NEXT: v128.load $push1=, 0($0):p2align=0
|
|
; CHECK-NEXT: v128.load $push0=, 0($1):p2align=0
|
|
; CHECK-NEXT: i8x16.eq $push2=, $pop1, $pop0
|
|
; CHECK-NEXT: i8x16.all_true $push3=, $pop2
|
|
; CHECK-NEXT: return $pop3
|
|
%cmp_16 = call i32 @memcmp(ptr %a, ptr %b, i32 16)
|
|
%res = icmp eq i32 %cmp_16, 0
|
|
ret i1 %res
|
|
}
|