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
210 lines
7.8 KiB
LLVM
210 lines
7.8 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py
|
|
; RUN: opt < %s -mtriple=aarch64 -passes="print<cost-model>" -cost-kind=all 2>&1 -disable-output | FileCheck %s
|
|
|
|
; Verify the cost of scalar ctlz instructions.
|
|
|
|
target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128"
|
|
|
|
define i64 @test_cttz_i64(i64 %a) {
|
|
;
|
|
; CHECK-LABEL: 'test_cttz_i64'
|
|
; CHECK-NEXT: Cost Model: Found costs of 1 for: %cttz = call i64 @llvm.cttz.i64(i64 %a, i1 false)
|
|
; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret i64 %cttz
|
|
;
|
|
%cttz = call i64 @llvm.cttz.i64(i64 %a)
|
|
ret i64 %cttz
|
|
}
|
|
|
|
define i32 @test_cttz_i32(i32 %a) {
|
|
;
|
|
; CHECK-LABEL: 'test_cttz_i32'
|
|
; CHECK-NEXT: Cost Model: Found costs of 1 for: %cttz = call i32 @llvm.cttz.i32(i32 %a, i1 false)
|
|
; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret i32 %cttz
|
|
;
|
|
%cttz = call i32 @llvm.cttz.i32(i32 %a)
|
|
ret i32 %cttz
|
|
}
|
|
|
|
define i16 @test_cttz_i16(i16 %a) {
|
|
;
|
|
; CHECK-LABEL: 'test_cttz_i16'
|
|
; CHECK-NEXT: Cost Model: Found costs of 1 for: %cttz = call i16 @llvm.cttz.i16(i16 %a, i1 false)
|
|
; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret i16 %cttz
|
|
;
|
|
%cttz = call i16 @llvm.cttz.i16(i16 %a)
|
|
ret i16 %cttz
|
|
}
|
|
|
|
define i8 @test_cttz_i8(i8 %a) {
|
|
;
|
|
; CHECK-LABEL: 'test_cttz_i8'
|
|
; CHECK-NEXT: Cost Model: Found costs of 1 for: %cttz = call i8 @llvm.cttz.i8(i8 %a, i1 false)
|
|
; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret i8 %cttz
|
|
;
|
|
%cttz = call i8 @llvm.cttz.i8(i8 %a)
|
|
ret i8 %cttz
|
|
}
|
|
|
|
declare i64 @llvm.cttz.i64(i64)
|
|
declare i32 @llvm.cttz.i32(i32)
|
|
declare i16 @llvm.cttz.i16(i16)
|
|
declare i8 @llvm.cttz.i8(i8)
|
|
|
|
; Verify the cost of vector cttz instructions.
|
|
|
|
define <2 x i64> @test_cttz_v2i64(<2 x i64> %a) {
|
|
;
|
|
; CHECK-LABEL: 'test_cttz_v2i64'
|
|
; CHECK-NEXT: Cost Model: Found costs of RThru:10 CodeSize:6 Lat:10 SizeLat:10 for: %cttz = call <2 x i64> @llvm.cttz.v2i64(<2 x i64> %a, i1 true)
|
|
; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret <2 x i64> %cttz
|
|
;
|
|
%cttz = call <2 x i64> @llvm.cttz.v2i64(<2 x i64> %a, i1 true)
|
|
ret <2 x i64> %cttz
|
|
}
|
|
|
|
define <2 x i32> @test_cttz_v2i32(<2 x i32> %a) {
|
|
;
|
|
; CHECK-LABEL: 'test_cttz_v2i32'
|
|
; CHECK-NEXT: Cost Model: Found costs of 3 for: %cttz = call <2 x i32> @llvm.cttz.v2i32(<2 x i32> %a, i1 true)
|
|
; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret <2 x i32> %cttz
|
|
;
|
|
%cttz = call <2 x i32> @llvm.cttz.v2i32(<2 x i32> %a, i1 true)
|
|
ret <2 x i32> %cttz
|
|
}
|
|
|
|
define <4 x i32> @test_cttz_v4i32(<4 x i32> %a) {
|
|
;
|
|
; CHECK-LABEL: 'test_cttz_v4i32'
|
|
; CHECK-NEXT: Cost Model: Found costs of 3 for: %cttz = call <4 x i32> @llvm.cttz.v4i32(<4 x i32> %a, i1 true)
|
|
; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret <4 x i32> %cttz
|
|
;
|
|
%cttz = call <4 x i32> @llvm.cttz.v4i32(<4 x i32> %a, i1 true)
|
|
ret <4 x i32> %cttz
|
|
}
|
|
|
|
define <2 x i16> @test_cttz_v2i16(<2 x i16> %a) {
|
|
;
|
|
; CHECK-LABEL: 'test_cttz_v2i16'
|
|
; CHECK-NEXT: Cost Model: Found costs of 3 for: %cttz = call <2 x i16> @llvm.cttz.v2i16(<2 x i16> %a, i1 true)
|
|
; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret <2 x i16> %cttz
|
|
;
|
|
%cttz = call <2 x i16> @llvm.cttz.v2i16(<2 x i16> %a, i1 true)
|
|
ret <2 x i16> %cttz
|
|
}
|
|
|
|
define <4 x i16> @test_cttz_v4i16(<4 x i16> %a) {
|
|
;
|
|
; CHECK-LABEL: 'test_cttz_v4i16'
|
|
; CHECK-NEXT: Cost Model: Found costs of 3 for: %cttz = call <4 x i16> @llvm.cttz.v4i16(<4 x i16> %a, i1 true)
|
|
; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret <4 x i16> %cttz
|
|
;
|
|
%cttz = call <4 x i16> @llvm.cttz.v4i16(<4 x i16> %a, i1 true)
|
|
ret <4 x i16> %cttz
|
|
}
|
|
|
|
define <8 x i16> @test_cttz_v8i16(<8 x i16> %a) {
|
|
;
|
|
; CHECK-LABEL: 'test_cttz_v8i16'
|
|
; CHECK-NEXT: Cost Model: Found costs of 3 for: %cttz = call <8 x i16> @llvm.cttz.v8i16(<8 x i16> %a, i1 true)
|
|
; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret <8 x i16> %cttz
|
|
;
|
|
%cttz = call <8 x i16> @llvm.cttz.v8i16(<8 x i16> %a, i1 true)
|
|
ret <8 x i16> %cttz
|
|
}
|
|
|
|
define <2 x i8> @test_cttz_v2i8(<2 x i8> %a) {
|
|
;
|
|
; CHECK-LABEL: 'test_cttz_v2i8'
|
|
; CHECK-NEXT: Cost Model: Found costs of 3 for: %cttz = call <2 x i8> @llvm.cttz.v2i8(<2 x i8> %a, i1 true)
|
|
; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret <2 x i8> %cttz
|
|
;
|
|
%cttz = call <2 x i8> @llvm.cttz.v2i8(<2 x i8> %a, i1 true)
|
|
ret <2 x i8> %cttz
|
|
}
|
|
|
|
define <4 x i8> @test_cttz_v4i8(<4 x i8> %a) {
|
|
;
|
|
; CHECK-LABEL: 'test_cttz_v4i8'
|
|
; CHECK-NEXT: Cost Model: Found costs of 3 for: %cttz = call <4 x i8> @llvm.cttz.v4i8(<4 x i8> %a, i1 true)
|
|
; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret <4 x i8> %cttz
|
|
;
|
|
%cttz = call <4 x i8> @llvm.cttz.v4i8(<4 x i8> %a, i1 true)
|
|
ret <4 x i8> %cttz
|
|
}
|
|
|
|
define <8 x i8> @test_cttz_v8i8(<8 x i8> %a) {
|
|
;
|
|
; CHECK-LABEL: 'test_cttz_v8i8'
|
|
; CHECK-NEXT: Cost Model: Found costs of 2 for: %cttz = call <8 x i8> @llvm.cttz.v8i8(<8 x i8> %a, i1 true)
|
|
; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret <8 x i8> %cttz
|
|
;
|
|
%cttz = call <8 x i8> @llvm.cttz.v8i8(<8 x i8> %a, i1 true)
|
|
ret <8 x i8> %cttz
|
|
}
|
|
|
|
define <16 x i8> @test_cttz_v16i8(<16 x i8> %a) {
|
|
;
|
|
; CHECK-LABEL: 'test_cttz_v16i8'
|
|
; CHECK-NEXT: Cost Model: Found costs of 2 for: %cttz = call <16 x i8> @llvm.cttz.v16i8(<16 x i8> %a, i1 true)
|
|
; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret <16 x i8> %cttz
|
|
;
|
|
%cttz = call <16 x i8> @llvm.cttz.v16i8(<16 x i8> %a, i1 true)
|
|
ret <16 x i8> %cttz
|
|
}
|
|
|
|
define <4 x i64> @test_cttz_v4i64(<4 x i64> %a) {
|
|
;
|
|
; CHECK-LABEL: 'test_cttz_v4i64'
|
|
; CHECK-NEXT: Cost Model: Found costs of RThru:20 CodeSize:12 Lat:20 SizeLat:20 for: %cttz = call <4 x i64> @llvm.cttz.v4i64(<4 x i64> %a, i1 true)
|
|
; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret <4 x i64> %cttz
|
|
;
|
|
%cttz = call <4 x i64> @llvm.cttz.v4i64(<4 x i64> %a, i1 true)
|
|
ret <4 x i64> %cttz
|
|
}
|
|
|
|
define <8 x i32> @test_cttz_v8i32(<8 x i32> %a) {
|
|
;
|
|
; CHECK-LABEL: 'test_cttz_v8i32'
|
|
; CHECK-NEXT: Cost Model: Found costs of 6 for: %cttz = call <8 x i32> @llvm.cttz.v8i32(<8 x i32> %a, i1 true)
|
|
; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret <8 x i32> %cttz
|
|
;
|
|
%cttz = call <8 x i32> @llvm.cttz.v8i32(<8 x i32> %a, i1 true)
|
|
ret <8 x i32> %cttz
|
|
}
|
|
|
|
define <16 x i16> @test_cttz_v16i16(<16 x i16> %a) {
|
|
;
|
|
; CHECK-LABEL: 'test_cttz_v16i16'
|
|
; CHECK-NEXT: Cost Model: Found costs of 6 for: %cttz = call <16 x i16> @llvm.cttz.v16i16(<16 x i16> %a, i1 true)
|
|
; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret <16 x i16> %cttz
|
|
;
|
|
%cttz = call <16 x i16> @llvm.cttz.v16i16(<16 x i16> %a, i1 true)
|
|
ret <16 x i16> %cttz
|
|
}
|
|
|
|
define <32 x i8> @test_cttz_v32i8(<32 x i8> %a) {
|
|
;
|
|
; CHECK-LABEL: 'test_cttz_v32i8'
|
|
; CHECK-NEXT: Cost Model: Found costs of 4 for: %cttz = call <32 x i8> @llvm.cttz.v32i8(<32 x i8> %a, i1 true)
|
|
; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret <32 x i8> %cttz
|
|
;
|
|
%cttz = call <32 x i8> @llvm.cttz.v32i8(<32 x i8> %a, i1 true)
|
|
ret <32 x i8> %cttz
|
|
}
|
|
|
|
declare <2 x i64> @llvm.cttz.v2i64(<2 x i64>, i1)
|
|
declare <2 x i32> @llvm.cttz.v2i32(<2 x i32>, i1)
|
|
declare <4 x i32> @llvm.cttz.v4i32(<4 x i32>, i1)
|
|
declare <2 x i16> @llvm.cttz.v2i16(<2 x i16>, i1)
|
|
declare <4 x i16> @llvm.cttz.v4i16(<4 x i16>, i1)
|
|
declare <8 x i16> @llvm.cttz.v8i16(<8 x i16>, i1)
|
|
declare <2 x i8> @llvm.cttz.v2i8(<2 x i8>, i1)
|
|
declare <4 x i8> @llvm.cttz.v4i8(<4 x i8>, i1)
|
|
declare <8 x i8> @llvm.cttz.v8i8(<8 x i8>, i1)
|
|
declare <16 x i8> @llvm.cttz.v16i8(<16 x i8>, i1)
|
|
|
|
declare <4 x i64> @llvm.cttz.v4i64(<4 x i64>, i1)
|
|
declare <8 x i32> @llvm.cttz.v8i32(<8 x i32>, i1)
|
|
declare <16 x i16> @llvm.cttz.v16i16(<16 x i16>, i1)
|
|
declare <32 x i8> @llvm.cttz.v32i8(<32 x i8>, i1)
|