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
188 lines
6.2 KiB
LLVM
188 lines
6.2 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+bmi,+bmi2 | FileCheck %s --check-prefixes=CHECK
|
|
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+bmi,+bmi2,egpr --show-mc-encoding | FileCheck %s --check-prefixes=EGPR
|
|
|
|
define i64 @bzhi64(i64 %x, i64 %y) {
|
|
; CHECK-LABEL: bzhi64:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: bzhiq %rsi, %rdi, %rax
|
|
; CHECK-NEXT: retq
|
|
;
|
|
; EGPR-LABEL: bzhi64:
|
|
; EGPR: # %bb.0:
|
|
; EGPR-NEXT: bzhiq %rsi, %rdi, %rax # EVEX TO VEX Compression encoding: [0xc4,0xe2,0xc8,0xf5,0xc7]
|
|
; EGPR-NEXT: retq # encoding: [0xc3]
|
|
%tmp = tail call i64 @llvm.x86.bmi.bzhi.64(i64 %x, i64 %y)
|
|
ret i64 %tmp
|
|
}
|
|
|
|
define i64 @bzhi64_load(ptr %x, i64 %y) {
|
|
; CHECK-LABEL: bzhi64_load:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: bzhiq %rsi, (%rdi), %rax
|
|
; CHECK-NEXT: retq
|
|
;
|
|
; EGPR-LABEL: bzhi64_load:
|
|
; EGPR: # %bb.0:
|
|
; EGPR-NEXT: bzhiq %rsi, (%rdi), %rax # EVEX TO VEX Compression encoding: [0xc4,0xe2,0xc8,0xf5,0x07]
|
|
; EGPR-NEXT: retq # encoding: [0xc3]
|
|
%x1 = load i64, ptr %x
|
|
%tmp = tail call i64 @llvm.x86.bmi.bzhi.64(i64 %x1, i64 %y)
|
|
ret i64 %tmp
|
|
}
|
|
|
|
declare i64 @llvm.x86.bmi.bzhi.64(i64, i64)
|
|
|
|
define i64 @pdep64(i64 %x, i64 %y) {
|
|
; CHECK-LABEL: pdep64:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: pdepq %rsi, %rdi, %rax
|
|
; CHECK-NEXT: retq
|
|
;
|
|
; EGPR-LABEL: pdep64:
|
|
; EGPR: # %bb.0:
|
|
; EGPR-NEXT: pdepq %rsi, %rdi, %rax # EVEX TO VEX Compression encoding: [0xc4,0xe2,0xc3,0xf5,0xc6]
|
|
; EGPR-NEXT: retq # encoding: [0xc3]
|
|
%tmp = tail call i64 @llvm.x86.bmi.pdep.64(i64 %x, i64 %y)
|
|
ret i64 %tmp
|
|
}
|
|
|
|
define i64 @pdep64_load(i64 %x, ptr %y) {
|
|
; CHECK-LABEL: pdep64_load:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: pdepq (%rsi), %rdi, %rax
|
|
; CHECK-NEXT: retq
|
|
;
|
|
; EGPR-LABEL: pdep64_load:
|
|
; EGPR: # %bb.0:
|
|
; EGPR-NEXT: pdepq (%rsi), %rdi, %rax # EVEX TO VEX Compression encoding: [0xc4,0xe2,0xc3,0xf5,0x06]
|
|
; EGPR-NEXT: retq # encoding: [0xc3]
|
|
%y1 = load i64, ptr %y
|
|
%tmp = tail call i64 @llvm.x86.bmi.pdep.64(i64 %x, i64 %y1)
|
|
ret i64 %tmp
|
|
}
|
|
|
|
define i64 @pdep64_anyext(i32 %x) {
|
|
; CHECK-LABEL: pdep64_anyext:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: # kill: def $edi killed $edi def $rdi
|
|
; CHECK-NEXT: movabsq $6148914691236517205, %rax # imm = 0x5555555555555555
|
|
; CHECK-NEXT: pdepq %rax, %rdi, %rax
|
|
; CHECK-NEXT: retq
|
|
;
|
|
; EGPR-LABEL: pdep64_anyext:
|
|
; EGPR: # %bb.0:
|
|
; EGPR-NEXT: # kill: def $edi killed $edi def $rdi
|
|
; EGPR-NEXT: movabsq $6148914691236517205, %rax # encoding: [0x48,0xb8,0x55,0x55,0x55,0x55,0x55,0x55,0x55,0x55]
|
|
; EGPR-NEXT: # imm = 0x5555555555555555
|
|
; EGPR-NEXT: pdepq %rax, %rdi, %rax # EVEX TO VEX Compression encoding: [0xc4,0xe2,0xc3,0xf5,0xc0]
|
|
; EGPR-NEXT: retq # encoding: [0xc3]
|
|
%x1 = sext i32 %x to i64
|
|
%tmp = tail call i64 @llvm.x86.bmi.pdep.64(i64 %x1, i64 6148914691236517205)
|
|
ret i64 %tmp
|
|
}
|
|
|
|
declare i64 @llvm.x86.bmi.pdep.64(i64, i64)
|
|
|
|
define i64 @pext64(i64 %x, i64 %y) {
|
|
; CHECK-LABEL: pext64:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: pextq %rsi, %rdi, %rax
|
|
; CHECK-NEXT: retq
|
|
;
|
|
; EGPR-LABEL: pext64:
|
|
; EGPR: # %bb.0:
|
|
; EGPR-NEXT: pextq %rsi, %rdi, %rax # EVEX TO VEX Compression encoding: [0xc4,0xe2,0xc2,0xf5,0xc6]
|
|
; EGPR-NEXT: retq # encoding: [0xc3]
|
|
%tmp = tail call i64 @llvm.x86.bmi.pext.64(i64 %x, i64 %y)
|
|
ret i64 %tmp
|
|
}
|
|
|
|
define i64 @pext64_load(i64 %x, ptr %y) {
|
|
; CHECK-LABEL: pext64_load:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: pextq (%rsi), %rdi, %rax
|
|
; CHECK-NEXT: retq
|
|
;
|
|
; EGPR-LABEL: pext64_load:
|
|
; EGPR: # %bb.0:
|
|
; EGPR-NEXT: pextq (%rsi), %rdi, %rax # EVEX TO VEX Compression encoding: [0xc4,0xe2,0xc2,0xf5,0x06]
|
|
; EGPR-NEXT: retq # encoding: [0xc3]
|
|
%y1 = load i64, ptr %y
|
|
%tmp = tail call i64 @llvm.x86.bmi.pext.64(i64 %x, i64 %y1)
|
|
ret i64 %tmp
|
|
}
|
|
|
|
define i64 @pext64_knownbits(i64 %x, i64 %y) {
|
|
; CHECK-LABEL: pext64_knownbits:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: movabsq $6148914691236517205, %rax # imm = 0x5555555555555555
|
|
; CHECK-NEXT: pextq %rax, %rdi, %rax
|
|
; CHECK-NEXT: retq
|
|
;
|
|
; EGPR-LABEL: pext64_knownbits:
|
|
; EGPR: # %bb.0:
|
|
; EGPR-NEXT: movabsq $6148914691236517205, %rax # encoding: [0x48,0xb8,0x55,0x55,0x55,0x55,0x55,0x55,0x55,0x55]
|
|
; EGPR-NEXT: # imm = 0x5555555555555555
|
|
; EGPR-NEXT: pextq %rax, %rdi, %rax # EVEX TO VEX Compression encoding: [0xc4,0xe2,0xc2,0xf5,0xc0]
|
|
; EGPR-NEXT: retq # encoding: [0xc3]
|
|
%tmp = tail call i64 @llvm.x86.bmi.pext.64(i64 %x, i64 6148914691236517205)
|
|
%tmp2 = and i64 %tmp, 4294967295
|
|
ret i64 %tmp2
|
|
}
|
|
|
|
declare i64 @llvm.x86.bmi.pext.64(i64, i64)
|
|
|
|
define i64 @mulx64(i64 %x, i64 %y, ptr %p) {
|
|
; CHECK-LABEL: mulx64:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: movq %rdx, %rcx
|
|
; CHECK-NEXT: movq %rdi, %rdx
|
|
; CHECK-NEXT: mulxq %rsi, %rax, %rdx
|
|
; CHECK-NEXT: movq %rdx, (%rcx)
|
|
; CHECK-NEXT: retq
|
|
;
|
|
; EGPR-LABEL: mulx64:
|
|
; EGPR: # %bb.0:
|
|
; EGPR-NEXT: movq %rdx, %rcx # encoding: [0x48,0x89,0xd1]
|
|
; EGPR-NEXT: movq %rdi, %rdx # encoding: [0x48,0x89,0xfa]
|
|
; EGPR-NEXT: mulxq %rsi, %rax, %rdx # EVEX TO VEX Compression encoding: [0xc4,0xe2,0xfb,0xf6,0xd6]
|
|
; EGPR-NEXT: movq %rdx, (%rcx) # encoding: [0x48,0x89,0x11]
|
|
; EGPR-NEXT: retq # encoding: [0xc3]
|
|
%x1 = zext i64 %x to i128
|
|
%y1 = zext i64 %y to i128
|
|
%r1 = mul i128 %x1, %y1
|
|
%h1 = lshr i128 %r1, 64
|
|
%h = trunc i128 %h1 to i64
|
|
%l = trunc i128 %r1 to i64
|
|
store i64 %h, ptr %p
|
|
ret i64 %l
|
|
}
|
|
|
|
define i64 @mulx64_load(i64 %x, ptr %y, ptr %p) {
|
|
; CHECK-LABEL: mulx64_load:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: movq %rdx, %rcx
|
|
; CHECK-NEXT: movq %rdi, %rdx
|
|
; CHECK-NEXT: mulxq (%rsi), %rax, %rdx
|
|
; CHECK-NEXT: movq %rdx, (%rcx)
|
|
; CHECK-NEXT: retq
|
|
;
|
|
; EGPR-LABEL: mulx64_load:
|
|
; EGPR: # %bb.0:
|
|
; EGPR-NEXT: movq %rdx, %rcx # encoding: [0x48,0x89,0xd1]
|
|
; EGPR-NEXT: movq %rdi, %rdx # encoding: [0x48,0x89,0xfa]
|
|
; EGPR-NEXT: mulxq (%rsi), %rax, %rdx # EVEX TO VEX Compression encoding: [0xc4,0xe2,0xfb,0xf6,0x16]
|
|
; EGPR-NEXT: movq %rdx, (%rcx) # encoding: [0x48,0x89,0x11]
|
|
; EGPR-NEXT: retq # encoding: [0xc3]
|
|
%y1 = load i64, ptr %y
|
|
%x2 = zext i64 %x to i128
|
|
%y2 = zext i64 %y1 to i128
|
|
%r1 = mul i128 %x2, %y2
|
|
%h1 = lshr i128 %r1, 64
|
|
%h = trunc i128 %h1 to i64
|
|
%l = trunc i128 %r1 to i64
|
|
store i64 %h, ptr %p
|
|
ret i64 %l
|
|
}
|