83c1d00311
This PR is to address legacy issues with module analysis that currently uses a complicated and not so efficient approach to trace dependencies between SPIR-V id's via a duplicate tracker data structures and an explicitly built dependency graph. Even a quick performance check without any specialized benchmarks points to this part of the implementation as a biggest bottleneck. This PR specifically: * eliminates a need to build a dependency graph as a data structure, * updates the test suite (mainly, by fixing incorrect CHECK's referring to a hardcoded order of definitions, contradicting the spec requirement to allow certain definitions to go "in any order", see https://registry.khronos.org/SPIR-V/specs/unified1/SPIRV.html#_logical_layout_of_a_module), * improves function pointers implementation so that it now passes EXPENSIVE_CHECKS (thus removing 3 XFAIL's in the test suite). As a quick sanity check of whether goals of the PR are achieved, we can measure time of translation for any big LLVM IR. While testing the PR in the local development environment, improvements of the x5 order have been observed. For example, the SYCL test case "group barrier" that is a ~1Mb binary IR input shows the following values of the naive performance metric that we can nevertheless apply here to roughly estimate effects of the PR. before the PR: ``` $ time llc -O0 -mtriple=spirv64v1.6-unknown-unknown _group_barrier_phi.bc -o 1 --filetype=obj real 3m33.241s user 3m14.688s sys 0m18.530s ``` after the PR ``` $ time llc -O0 -mtriple=spirv64v1.6-unknown-unknown _group_barrier_phi.bc -o 1 --filetype=obj real 0m42.031s user 0m38.834s sys 0m3.193s ``` Next work should probably address Duplicate Tracker further, as it needs analysis now from the perspective of what parts of it are not necessary now, after changing the approach to implementation of the module analysis step.
89 lines
6.1 KiB
LLVM
89 lines
6.1 KiB
LLVM
; RUN: llc -verify-machineinstrs -O0 -mtriple=spirv64-unknown-unknown %s -o - | FileCheck %s --check-prefix=CHECK-SPIRV
|
|
; RUN: %if spirv-tools %{ llc -O0 -mtriple=spirv64-unknown-unknown %s -o - -filetype=obj | spirv-val %}
|
|
|
|
; RUN: llc -verify-machineinstrs -O0 -mtriple=spirv32-unknown-unknown %s -o - | FileCheck %s --check-prefix=CHECK-SPIRV
|
|
; RUN: %if spirv-tools %{ llc -O0 -mtriple=spirv32-unknown-unknown %s -o - -filetype=obj | spirv-val %}
|
|
|
|
;; constant sampler_t constSampl = CLK_FILTER_LINEAR;
|
|
;;
|
|
;; __kernel
|
|
;; void sample_kernel_float(image2d_t input, float2 coords, global float4 *results, sampler_t argSampl) {
|
|
;; *results = read_imagef(input, constSampl, coords);
|
|
;; *results = read_imagef(input, argSampl, coords);
|
|
;; *results = read_imagef(input, CLK_FILTER_NEAREST|CLK_ADDRESS_REPEAT, coords);
|
|
;; }
|
|
;;
|
|
;; __kernel
|
|
;; void sample_kernel_int(image2d_t input, float2 coords, global int4 *results, sampler_t argSampl) {
|
|
;; *results = read_imagei(input, constSampl, coords);
|
|
;; *results = read_imagei(input, argSampl, coords);
|
|
;; *results = read_imagei(input, CLK_FILTER_NEAREST|CLK_ADDRESS_REPEAT, coords);
|
|
;; }
|
|
|
|
; CHECK-SPIRV: OpCapability LiteralSampler
|
|
; CHECK-SPIRV: OpName %[[#sample_kernel_float:]] "sample_kernel_float"
|
|
; CHECK-SPIRV: OpName %[[#sample_kernel_int:]] "sample_kernel_int"
|
|
|
|
; CHECK-SPIRV-DAG: %[[#TypeSampler:]] = OpTypeSampler
|
|
; CHECK-SPIRV-DAG: %[[#SampledImageTy:]] = OpTypeSampledImage
|
|
; CHECK-SPIRV-DAG: %[[#ConstSampler1:]] = OpConstantSampler %[[#TypeSampler]] None 0 Linear
|
|
; CHECK-SPIRV-DAG: %[[#ConstSampler2:]] = OpConstantSampler %[[#TypeSampler]] Repeat 0 Nearest
|
|
|
|
; CHECK-SPIRV: %[[#sample_kernel_float]] = OpFunction %{{.*}}
|
|
; CHECK-SPIRV: %[[#InputImage:]] = OpFunctionParameter %{{.*}}
|
|
; CHECK-SPIRV: %[[#argSampl:]] = OpFunctionParameter %[[#TypeSampler]]
|
|
|
|
; CHECK-SPIRV: %[[#SampledImage1:]] = OpSampledImage %[[#SampledImageTy]] %[[#InputImage]] %[[#ConstSampler1]]
|
|
; CHECK-SPIRV: %[[#]] = OpImageSampleExplicitLod %[[#]] %[[#SampledImage1]]
|
|
|
|
; CHECK-SPIRV: %[[#SampledImage2:]] = OpSampledImage %[[#SampledImageTy]] %[[#InputImage]] %[[#argSampl]]
|
|
; CHECK-SPIRV: %[[#]] = OpImageSampleExplicitLod %[[#]] %[[#SampledImage2]]
|
|
|
|
; CHECK-SPIRV: %[[#SampledImage3:]] = OpSampledImage %[[#SampledImageTy]] %[[#InputImage]] %[[#ConstSampler2]]
|
|
; CHECK-SPIRV: %[[#]] = OpImageSampleExplicitLod %[[#]] %[[#SampledImage3]]
|
|
|
|
define dso_local spir_kernel void @sample_kernel_float(target("spirv.Image", void, 1, 0, 0, 0, 0, 0, 0) %input, <2 x float> noundef %coords, <4 x float> addrspace(1)* nocapture noundef writeonly %results, target("spirv.Sampler") %argSampl) local_unnamed_addr {
|
|
entry:
|
|
%0 = tail call spir_func target("spirv.Sampler") @__translate_sampler_initializer(i32 32)
|
|
%call = tail call spir_func <4 x float> @_Z11read_imagef14ocl_image2d_ro11ocl_samplerDv2_f(target("spirv.Image", void, 1, 0, 0, 0, 0, 0, 0) %input, target("spirv.Sampler") %0, <2 x float> noundef %coords)
|
|
store <4 x float> %call, <4 x float> addrspace(1)* %results, align 16
|
|
%call1 = tail call spir_func <4 x float> @_Z11read_imagef14ocl_image2d_ro11ocl_samplerDv2_f(target("spirv.Image", void, 1, 0, 0, 0, 0, 0, 0) %input, target("spirv.Sampler") %argSampl, <2 x float> noundef %coords)
|
|
store <4 x float> %call1, <4 x float> addrspace(1)* %results, align 16
|
|
%1 = tail call spir_func target("spirv.Sampler") @__translate_sampler_initializer(i32 22)
|
|
%call2 = tail call spir_func <4 x float> @_Z11read_imagef14ocl_image2d_ro11ocl_samplerDv2_f(target("spirv.Image", void, 1, 0, 0, 0, 0, 0, 0) %input, target("spirv.Sampler") %1, <2 x float> noundef %coords)
|
|
store <4 x float> %call2, <4 x float> addrspace(1)* %results, align 16
|
|
ret void
|
|
}
|
|
|
|
declare spir_func <4 x float> @_Z11read_imagef14ocl_image2d_ro11ocl_samplerDv2_f(target("spirv.Image", void, 1, 0, 0, 0, 0, 0, 0), target("spirv.Sampler"), <2 x float> noundef) local_unnamed_addr
|
|
|
|
declare spir_func target("spirv.Sampler") @__translate_sampler_initializer(i32) local_unnamed_addr
|
|
|
|
; CHECK-SPIRV: %[[#sample_kernel_int]] = OpFunction %{{.*}}
|
|
; CHECK-SPIRV: %[[#InputImage:]] = OpFunctionParameter %{{.*}}
|
|
; CHECK-SPIRV: %[[#argSampl:]] = OpFunctionParameter %[[#TypeSampler]]
|
|
|
|
; CHECK-SPIRV: %[[#SampledImage4:]] = OpSampledImage %[[#SampledImageTy]] %[[#InputImage]] %[[#ConstSampler1]]
|
|
; CHECK-SPIRV: %[[#]] = OpImageSampleExplicitLod %[[#]] %[[#SampledImage4]]
|
|
|
|
; CHECK-SPIRV: %[[#SampledImage5:]] = OpSampledImage %[[#SampledImageTy]] %[[#InputImage]] %[[#argSampl]]
|
|
; CHECK-SPIRV: %[[#]] = OpImageSampleExplicitLod %[[#]] %[[#SampledImage5]]
|
|
|
|
; CHECK-SPIRV: %[[#SampledImage6:]] = OpSampledImage %[[#SampledImageTy]] %[[#InputImage]] %[[#ConstSampler2]]
|
|
; CHECK-SPIRV: %[[#]] = OpImageSampleExplicitLod %[[#]] %[[#SampledImage6]]
|
|
|
|
define dso_local spir_kernel void @sample_kernel_int(target("spirv.Image", void, 1, 0, 0, 0, 0, 0, 0) %input, <2 x float> noundef %coords, <4 x i32> addrspace(1)* nocapture noundef writeonly %results, target("spirv.Sampler") %argSampl) local_unnamed_addr {
|
|
entry:
|
|
%0 = tail call spir_func target("spirv.Sampler") @__translate_sampler_initializer(i32 32)
|
|
%call = tail call spir_func <4 x i32> @_Z11read_imagei14ocl_image2d_ro11ocl_samplerDv2_f(target("spirv.Image", void, 1, 0, 0, 0, 0, 0, 0) %input, target("spirv.Sampler") %0, <2 x float> noundef %coords)
|
|
store <4 x i32> %call, <4 x i32> addrspace(1)* %results, align 16
|
|
%call1 = tail call spir_func <4 x i32> @_Z11read_imagei14ocl_image2d_ro11ocl_samplerDv2_f(target("spirv.Image", void, 1, 0, 0, 0, 0, 0, 0) %input, target("spirv.Sampler") %argSampl, <2 x float> noundef %coords)
|
|
store <4 x i32> %call1, <4 x i32> addrspace(1)* %results, align 16
|
|
%1 = tail call spir_func target("spirv.Sampler") @__translate_sampler_initializer(i32 22)
|
|
%call2 = tail call spir_func <4 x i32> @_Z11read_imagei14ocl_image2d_ro11ocl_samplerDv2_f(target("spirv.Image", void, 1, 0, 0, 0, 0, 0, 0) %input, target("spirv.Sampler") %1, <2 x float> noundef %coords)
|
|
store <4 x i32> %call2, <4 x i32> addrspace(1)* %results, align 16
|
|
ret void
|
|
}
|
|
|
|
declare spir_func <4 x i32> @_Z11read_imagei14ocl_image2d_ro11ocl_samplerDv2_f(target("spirv.Image", void, 1, 0, 0, 0, 0, 0, 0), target("spirv.Sampler"), <2 x float> noundef) local_unnamed_addr
|