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
142 lines
4.6 KiB
LLVM
142 lines
4.6 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc < %s -mtriple=x86_64-unknown-linux-gnu | FileCheck %s --check-prefix=X64
|
|
; RUN: llc < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+bmi2 | FileCheck %s --check-prefix=X64-BMI2
|
|
; RUN: llc < %s -mtriple=i386-unknown-linux-gnu | FileCheck %s --check-prefix=X86
|
|
|
|
; Test optimization of 32-bit unsigned division by constants with 33-bit magic
|
|
; constants (IsAdd=true) on 64-bit targets. The optimization uses pre-shifted
|
|
; constants and 64x64->128 bit multiplication to reduce instruction count.
|
|
|
|
define i32 @udiv_by_7(i32 %x) nounwind {
|
|
; X64-LABEL: udiv_by_7:
|
|
; X64: # %bb.0:
|
|
; X64-NEXT: movl %edi, %eax
|
|
; X64-NEXT: movabsq $2635249153617166336, %rcx # imm = 0x24924924A0000000
|
|
; X64-NEXT: mulq %rcx
|
|
; X64-NEXT: movq %rdx, %rax
|
|
; X64-NEXT: # kill: def $eax killed $eax killed $rax
|
|
; X64-NEXT: retq
|
|
;
|
|
; X64-BMI2-LABEL: udiv_by_7:
|
|
; X64-BMI2: # %bb.0:
|
|
; X64-BMI2-NEXT: movl %edi, %edx
|
|
; X64-BMI2-NEXT: movabsq $2635249153617166336, %rax # imm = 0x24924924A0000000
|
|
; X64-BMI2-NEXT: mulxq %rax, %rax, %rax
|
|
; X64-BMI2-NEXT: # kill: def $eax killed $eax killed $rax
|
|
; X64-BMI2-NEXT: retq
|
|
;
|
|
; X86-LABEL: udiv_by_7:
|
|
; X86: # %bb.0:
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx
|
|
; X86-NEXT: movl $613566757, %edx # imm = 0x24924925
|
|
; X86-NEXT: movl %ecx, %eax
|
|
; X86-NEXT: mull %edx
|
|
; X86-NEXT: subl %edx, %ecx
|
|
; X86-NEXT: shrl %ecx
|
|
; X86-NEXT: addl %edx, %ecx
|
|
; X86-NEXT: shrl $2, %ecx
|
|
; X86-NEXT: movl %ecx, %eax
|
|
; X86-NEXT: retl
|
|
%div = udiv i32 %x, 7
|
|
ret i32 %div
|
|
}
|
|
|
|
define i32 @udiv_by_19(i32 %x) nounwind {
|
|
; X64-LABEL: udiv_by_19:
|
|
; X64: # %bb.0:
|
|
; X64-NEXT: movl %edi, %eax
|
|
; X64-NEXT: movabsq $970881267157434368, %rcx # imm = 0xD79435E58000000
|
|
; X64-NEXT: mulq %rcx
|
|
; X64-NEXT: movq %rdx, %rax
|
|
; X64-NEXT: # kill: def $eax killed $eax killed $rax
|
|
; X64-NEXT: retq
|
|
;
|
|
; X64-BMI2-LABEL: udiv_by_19:
|
|
; X64-BMI2: # %bb.0:
|
|
; X64-BMI2-NEXT: movl %edi, %edx
|
|
; X64-BMI2-NEXT: movabsq $970881267157434368, %rax # imm = 0xD79435E58000000
|
|
; X64-BMI2-NEXT: mulxq %rax, %rax, %rax
|
|
; X64-BMI2-NEXT: # kill: def $eax killed $eax killed $rax
|
|
; X64-BMI2-NEXT: retq
|
|
;
|
|
; X86-LABEL: udiv_by_19:
|
|
; X86: # %bb.0:
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx
|
|
; X86-NEXT: movl ${{-?[0-9]+}}, %edx # imm = 0xAF286BCB
|
|
; X86-NEXT: movl %ecx, %eax
|
|
; X86-NEXT: mull %edx
|
|
; X86-NEXT: subl %edx, %ecx
|
|
; X86-NEXT: shrl %ecx
|
|
; X86-NEXT: addl %edx, %ecx
|
|
; X86-NEXT: shrl $4, %ecx
|
|
; X86-NEXT: movl %ecx, %eax
|
|
; X86-NEXT: retl
|
|
%div = udiv i32 %x, 19
|
|
ret i32 %div
|
|
}
|
|
|
|
define i32 @udiv_by_21(i32 %x) nounwind {
|
|
; X64-LABEL: udiv_by_21:
|
|
; X64: # %bb.0:
|
|
; X64-NEXT: movl %edi, %eax
|
|
; X64-NEXT: movabsq $878416384583794688, %rcx # imm = 0xC30C30C38000000
|
|
; X64-NEXT: mulq %rcx
|
|
; X64-NEXT: movq %rdx, %rax
|
|
; X64-NEXT: # kill: def $eax killed $eax killed $rax
|
|
; X64-NEXT: retq
|
|
;
|
|
; X64-BMI2-LABEL: udiv_by_21:
|
|
; X64-BMI2: # %bb.0:
|
|
; X64-BMI2-NEXT: movl %edi, %edx
|
|
; X64-BMI2-NEXT: movabsq $878416384583794688, %rax # imm = 0xC30C30C38000000
|
|
; X64-BMI2-NEXT: mulxq %rax, %rax, %rax
|
|
; X64-BMI2-NEXT: # kill: def $eax killed $eax killed $rax
|
|
; X64-BMI2-NEXT: retq
|
|
;
|
|
; X86-LABEL: udiv_by_21:
|
|
; X86: # %bb.0:
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx
|
|
; X86-NEXT: movl ${{-?[0-9]+}}, %edx # imm = 0x86186187
|
|
; X86-NEXT: movl %ecx, %eax
|
|
; X86-NEXT: mull %edx
|
|
; X86-NEXT: subl %edx, %ecx
|
|
; X86-NEXT: shrl %ecx
|
|
; X86-NEXT: addl %edx, %ecx
|
|
; X86-NEXT: shrl $4, %ecx
|
|
; X86-NEXT: movl %ecx, %eax
|
|
; X86-NEXT: retl
|
|
%div = udiv i32 %x, 21
|
|
ret i32 %div
|
|
}
|
|
|
|
; Test non-optimized case: divisor that doesn't use IsAdd
|
|
define i32 @udiv_by_3(i32 %x) nounwind {
|
|
; X64-LABEL: udiv_by_3:
|
|
; X64: # %bb.0:
|
|
; X64-NEXT: movl %edi, %ecx
|
|
; X64-NEXT: movl $2863311531, %eax # imm = 0xAAAAAAAB
|
|
; X64-NEXT: imulq %rcx, %rax
|
|
; X64-NEXT: shrq $33, %rax
|
|
; X64-NEXT: # kill: def $eax killed $eax killed $rax
|
|
; X64-NEXT: retq
|
|
;
|
|
; X64-BMI2-LABEL: udiv_by_3:
|
|
; X64-BMI2: # %bb.0:
|
|
; X64-BMI2-NEXT: movl %edi, %ecx
|
|
; X64-BMI2-NEXT: movl $2863311531, %eax # imm = 0xAAAAAAAB
|
|
; X64-BMI2-NEXT: imulq %rcx, %rax
|
|
; X64-BMI2-NEXT: shrq $33, %rax
|
|
; X64-BMI2-NEXT: # kill: def $eax killed $eax killed $rax
|
|
; X64-BMI2-NEXT: retq
|
|
;
|
|
; X86-LABEL: udiv_by_3:
|
|
; X86: # %bb.0:
|
|
; X86-NEXT: movl ${{-?[0-9]+}}, %eax # imm = 0xAAAAAAAB
|
|
; X86-NEXT: mull {{[0-9]+}}(%esp)
|
|
; X86-NEXT: movl %edx, %eax
|
|
; X86-NEXT: shrl %eax
|
|
; X86-NEXT: retl
|
|
%div = udiv i32 %x, 3
|
|
ret i32 %div
|
|
}
|