Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

EXSWHTEC-106 - Reimplement tests for hipOccupancyMaxActiveBlocksPerMultiprocessor and hipOccupancyMaxPotentialBlockSize APIs #46

Merged
merged 12 commits into from
Jun 28, 2023
Merged
1 change: 1 addition & 0 deletions catch/hipTestMain/config/config_amd_linux_common.json
Original file line number Diff line number Diff line change
Expand Up @@ -13,6 +13,7 @@
"Unit_hipInit_Negative",
"Unit_hipDeviceReset_Positive_Basic",
"Unit_hipDeviceReset_Positive_Threaded",
"Unit_hipOccupancyMaxActiveBlocksPerMultiprocessor_Negative_Parameters",
"Unit_hipGraphMemcpyNodeSetParamsToSymbol_Positive_Basic",
"Unit_hipGraphMemcpyNodeSetParamsFromSymbol_Positive_Basic",
"Unit_hipKernelNameRef_Negative_Parameters",
Expand Down
1 change: 1 addition & 0 deletions catch/hipTestMain/config/config_amd_windows_common.json
Original file line number Diff line number Diff line change
Expand Up @@ -95,6 +95,7 @@
"Unit_hipStreamSynchronize_NullStreamAndStreamPerThread",
"Note: intermittent Seg fault failure ",
"Unit_hipGraphAddEventRecordNode_Functional_WithoutFlags",
"Unit_hipOccupancyMaxActiveBlocksPerMultiprocessor_Negative_Parameters",
"Unit_hipGraphMemcpyNodeSetParamsToSymbol_Positive_Basic",
"Unit_hipGraphExecMemcpyNodeSetParamsToSymbol_Positive_Basic",
"Unit_hipGraphMemcpyNodeSetParamsFromSymbol_Positive_Basic",
Expand Down
2 changes: 2 additions & 0 deletions catch/unit/occupancy/CMakeLists.txt
Original file line number Diff line number Diff line change
@@ -1,7 +1,9 @@
# Common Tests - Test independent of all platforms
set(TEST_SRC
hipOccupancyMaxActiveBlocksPerMultiprocessor.cc
hipOccupancyMaxActiveBlocksPerMultiprocessor_old.cc
hipOccupancyMaxPotentialBlockSize.cc
hipOccupancyMaxPotentialBlockSize_old.cc
hipOccupancyMaxPotentialBlockSizeVariableSMemWithFlags.cc
)

Expand Down
139 changes: 83 additions & 56 deletions catch/unit/occupancy/hipOccupancyMaxActiveBlocksPerMultiprocessor.cc
Original file line number Diff line number Diff line change
@@ -1,5 +1,5 @@
/*
Copyright (c) 2021 Advanced Micro Devices, Inc. All rights reserved.
Copyright (c) 2022 Advanced Micro Devices, Inc. All rights reserved.
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
Expand All @@ -16,76 +16,103 @@ LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
THE SOFTWARE.
*/
#include <hip_test_common.hh>

static __global__ void f1(float *a) { *a = 1.0; }
/*
Testcase Scenarios :
Unit_hipOccupancyMaxActiveBlocksPerMultiprocessor_Positive_RangeValidation - Test correct execution
of hipOccupancyMaxActiveBlocksPerMultiprocessor for diffrent parameter values
Unit_hipOccupancyMaxActiveBlocksPerMultiprocessor_Positive_TemplateInvocation - Test correct
execution of hipOccupancyMaxActiveBlocksPerMultiprocessor template for diffrent parameter values
Unit_hipOccupancyMaxActiveBlocksPerMultiprocessor_Negative_Parameters - Test unsuccessful execution
of hipOccupancyMaxActiveBlocksPerMultiprocessor api when parameters are invalid
*/
#include "occupancy_common.hh"

template <typename T>
static __global__ void f2(T *a) { *a = 1; }
static __global__ void f1(float* a) { *a = 1.0; }

/**
* Defines
*/
#define OccupancyDisableCachingOverride 0x01
template <typename T> static __global__ void f2(T* a) { *a = 1; }

TEST_CASE("Unit_hipOccupancyMaxActiveBlocksPerMultiprocessor_Negative") {
hipError_t ret;
int numBlock = 0, blockSize = 0;
int gridSize = 0, defBlkSize = 32;
TEST_CASE("Unit_hipOccupancyMaxActiveBlocksPerMultiprocessor_Negative_Parameters") {
int numBlocks = 0;
int blockSize = 0;
int gridSize = 0;

// Get potential blocksize
HIP_CHECK(hipOccupancyMaxPotentialBlockSize(&gridSize, &blockSize, f1, 0, 0));

// Validate each argument
ret = hipOccupancyMaxActiveBlocksPerMultiprocessor(NULL, f1, blockSize, 0);
REQUIRE(ret != hipSuccess);

ret = hipOccupancyMaxActiveBlocksPerMultiprocessor(&numBlock, NULL, blockSize, 0);
REQUIRE(ret != hipSuccess);

ret = hipOccupancyMaxActiveBlocksPerMultiprocessor(&numBlock, f1, 0, 0);
REQUIRE(ret != hipSuccess);

ret = hipOccupancyMaxActiveBlocksPerMultiprocessor(&numBlock, f1, 0,
std::numeric_limits<std::size_t>::max());
REQUIRE(ret != hipSuccess);

ret = hipOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(&numBlock, f1,
defBlkSize, 0, OccupancyDisableCachingOverride);
REQUIRE(ret == hipSuccess);
// Common negative tests
MaxActiveBlocksPerMultiprocessorNegative(
[](int* numBlocks, int blockSize, size_t dynSharedMemPerBlk) {
return hipOccupancyMaxActiveBlocksPerMultiprocessor(numBlocks, f1, blockSize,
dynSharedMemPerBlk);
},
blockSize);

SECTION("Kernel function is NULL") {
HIP_CHECK_ERROR(hipOccupancyMaxActiveBlocksPerMultiprocessor(&numBlocks, NULL, blockSize, 0),
hipErrorInvalidDeviceFunction);
}
}

TEST_CASE("Unit_hipOccupancyMaxActiveBlocksPerMultiprocessor_rangeValidation") {
TEST_CASE("Unit_hipOccupancyMaxActiveBlocksPerMultiprocessor_Positive_RangeValidation") {
hipDeviceProp_t devProp;
int numBlock = 0, blockSize = 0;
int blockSize = 0;
int gridSize = 0;

// Get potential blocksize
HIP_CHECK(hipOccupancyMaxPotentialBlockSize(&gridSize, &blockSize, f1, 0, 0));

HIP_CHECK(hipGetDeviceProperties(&devProp, 0));

HIP_CHECK(hipOccupancyMaxActiveBlocksPerMultiprocessor(&numBlock, f1, blockSize, 0));

// Check if numBlocks and blockSize are within limits
REQUIRE(numBlock > 0);
REQUIRE((numBlock * blockSize) <= devProp.maxThreadsPerMultiProcessor);

// Validate numBlock after passing dynSharedMemPerBlk
HIP_CHECK(hipOccupancyMaxActiveBlocksPerMultiprocessor(&numBlock, f1, blockSize,
devProp.sharedMemPerBlock));

// Check if numBlocks and blockSize are within limits
REQUIRE(numBlock > 0);
REQUIRE((numBlock * blockSize) <= devProp.maxThreadsPerMultiProcessor);
SECTION("dynSharedMemPerBlk = 0") {
// Get potential blocksize
HIP_CHECK(hipOccupancyMaxPotentialBlockSize(&gridSize, &blockSize, f1, 0, 0));

MaxActiveBlocksPerMultiprocessor(
[blockSize](int* numBlocks) {
return hipOccupancyMaxActiveBlocksPerMultiprocessor(numBlocks, f1, blockSize, 0);
},
blockSize, devProp.maxThreadsPerMultiProcessor);
}
SECTION("dynSharedMemPerBlk = sharedMemPerBlock") {
// Get potential blocksize
HIP_CHECK(
hipOccupancyMaxPotentialBlockSize(&gridSize, &blockSize, f1, devProp.sharedMemPerBlock, 0));

MaxActiveBlocksPerMultiprocessor(
[blockSize, devProp](int* numBlocks) {
return hipOccupancyMaxActiveBlocksPerMultiprocessor(numBlocks, f1, blockSize,
devProp.sharedMemPerBlock);
},
blockSize, devProp.maxThreadsPerMultiProcessor);
}
}

TEST_CASE("Unit_hipOccupancyMaxActiveBlocksPerMultiprocessor_templateInvocation") {
int blockSize = 32;
int numBlock = 0;
TEST_CASE("Unit_hipOccupancyMaxActiveBlocksPerMultiprocessor_Positive_TemplateInvocation") {
hipDeviceProp_t devProp;
int blockSize = 0;
int gridSize = 0;

HIP_CHECK(hipOccupancyMaxActiveBlocksPerMultiprocessor<void(*)(int *)>
(&numBlock, f2, blockSize, 0));
REQUIRE(numBlock > 0);
}
HIP_CHECK(hipGetDeviceProperties(&devProp, 0));

SECTION("dynSharedMemPerBlk = 0") {
// Get potential blocksize
HIP_CHECK(hipOccupancyMaxPotentialBlockSize<void (*)(int*)>(&gridSize, &blockSize, f2, 0, 0));

MaxActiveBlocksPerMultiprocessor(
[blockSize](int* numBlocks) {
return hipOccupancyMaxActiveBlocksPerMultiprocessor<void (*)(int*)>(numBlocks, f2,
blockSize, 0);
},
blockSize, devProp.maxThreadsPerMultiProcessor);
}

SECTION("dynSharedMemPerBlk = sharedMemPerBlock") {
// Get potential blocksize
HIP_CHECK(hipOccupancyMaxPotentialBlockSize<void (*)(int*)>(&gridSize, &blockSize, f2,
devProp.sharedMemPerBlock, 0));

MaxActiveBlocksPerMultiprocessor(
[blockSize, devProp](int* numBlocks) {
return hipOccupancyMaxActiveBlocksPerMultiprocessor<void (*)(int*)>(
numBlocks, f2, blockSize, devProp.sharedMemPerBlock);
},
blockSize, devProp.maxThreadsPerMultiProcessor);
}
}
Original file line number Diff line number Diff line change
@@ -0,0 +1,91 @@
/*
Copyright (c) 2021 Advanced Micro Devices, Inc. All rights reserved.
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in
all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
THE SOFTWARE.
*/
#include <hip_test_common.hh>

static __global__ void f1(float *a) { *a = 1.0; }

template <typename T>
static __global__ void f2(T *a) { *a = 1; }

/**
* Defines
*/
#define OccupancyDisableCachingOverride 0x01

TEST_CASE("Unit_hipOccupancyMaxActiveBlocksPerMultiprocessor_Negative") {
hipError_t ret;
int numBlock = 0, blockSize = 0;
int gridSize = 0, defBlkSize = 32;

// Get potential blocksize
HIP_CHECK(hipOccupancyMaxPotentialBlockSize(&gridSize, &blockSize, f1, 0, 0));

// Validate each argument
ret = hipOccupancyMaxActiveBlocksPerMultiprocessor(NULL, f1, blockSize, 0);
REQUIRE(ret != hipSuccess);

ret = hipOccupancyMaxActiveBlocksPerMultiprocessor(&numBlock, NULL, blockSize, 0);
REQUIRE(ret != hipSuccess);

ret = hipOccupancyMaxActiveBlocksPerMultiprocessor(&numBlock, f1, 0, 0);
REQUIRE(ret != hipSuccess);

ret = hipOccupancyMaxActiveBlocksPerMultiprocessor(&numBlock, f1, 0,
std::numeric_limits<std::size_t>::max());
REQUIRE(ret != hipSuccess);

ret = hipOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(&numBlock, f1,
defBlkSize, 0, OccupancyDisableCachingOverride);
REQUIRE(ret == hipSuccess);
}

TEST_CASE("Unit_hipOccupancyMaxActiveBlocksPerMultiprocessor_rangeValidation") {
hipDeviceProp_t devProp;
int numBlock = 0, blockSize = 0;
int gridSize = 0;

// Get potential blocksize
HIP_CHECK(hipOccupancyMaxPotentialBlockSize(&gridSize, &blockSize, f1, 0, 0));

HIP_CHECK(hipGetDeviceProperties(&devProp, 0));

HIP_CHECK(hipOccupancyMaxActiveBlocksPerMultiprocessor(&numBlock, f1, blockSize, 0));

// Check if numBlocks and blockSize are within limits
REQUIRE(numBlock > 0);
REQUIRE((numBlock * blockSize) <= devProp.maxThreadsPerMultiProcessor);

// Validate numBlock after passing dynSharedMemPerBlk
HIP_CHECK(hipOccupancyMaxActiveBlocksPerMultiprocessor(&numBlock, f1, blockSize,
devProp.sharedMemPerBlock));

// Check if numBlocks and blockSize are within limits
REQUIRE(numBlock > 0);
REQUIRE((numBlock * blockSize) <= devProp.maxThreadsPerMultiProcessor);
}

TEST_CASE("Unit_hipOccupancyMaxActiveBlocksPerMultiprocessor_templateInvocation") {
int blockSize = 32;
int numBlock = 0;

HIP_CHECK(hipOccupancyMaxActiveBlocksPerMultiprocessor<void(*)(int *)>
(&numBlock, f2, blockSize, 0));
REQUIRE(numBlock > 0);
}

Loading