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
381 lines
18 KiB
C++
381 lines
18 KiB
C++
/*******************************************************************************
|
|
*
|
|
* MIT License
|
|
*
|
|
* Copyright (C) 2021-2025 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.
|
|
*
|
|
*******************************************************************************/
|
|
#ifndef ROCWMMA_TEST_GEMM_PREDICATES_BASE_HPP
|
|
#define ROCWMMA_TEST_GEMM_PREDICATES_BASE_HPP
|
|
|
|
#include "gemm_test_traits.hpp"
|
|
|
|
namespace rocwmma
|
|
{
|
|
|
|
template <uint32_t BlockM,
|
|
uint32_t BlockN,
|
|
uint32_t BlockK,
|
|
typename InputT,
|
|
typename OutputT,
|
|
typename ComputeT,
|
|
uint32_t BlocksX,
|
|
uint32_t BlocksY,
|
|
uint32_t TBlockX,
|
|
uint32_t TBlockY,
|
|
uint32_t WaveSize,
|
|
uint32_t ArchId>
|
|
struct GemmPredicatesBase
|
|
{
|
|
using TestTraits = GemmTestTraits<BlockM,
|
|
BlockN,
|
|
BlockK,
|
|
InputT,
|
|
OutputT,
|
|
ComputeT,
|
|
BlocksX,
|
|
BlocksY,
|
|
WaveSize,
|
|
ArchId>;
|
|
|
|
protected:
|
|
enum struct GlobalPredicates : bool
|
|
{
|
|
// ThreadblockX must be a multiple of the wave size
|
|
TBlockXTest = (TBlockX % WaveSize == 0u),
|
|
|
|
// Ensure that we have at least 1 wave
|
|
MinTBlockTest = (TBlockX >= WaveSize && TBlockY >= 1),
|
|
|
|
// Ensure that we only build for the current compiler target, which
|
|
// will also exclude the host.
|
|
CurrentArchTest = (ArchId == Constants::AMDGCN_CURRENT_ARCH_ID),
|
|
|
|
// Ensure that we only build for the current wave size
|
|
CurrentWaveSizeTest = (WaveSize == Constants::AMDGCN_WAVE_SIZE),
|
|
|
|
// Only supported hardware allowed
|
|
ArchTest = (bool)TestTraits::Arch::IsGfx9 || (bool)TestTraits::Arch::IsGfx11
|
|
|| (bool)TestTraits::Arch::IsGfx12,
|
|
|
|
// During the build phase, we have information about current target arch.
|
|
// This means only the current arch and wave size are valid.
|
|
EnableBuild
|
|
= (TBlockXTest && MinTBlockTest && CurrentArchTest && CurrentWaveSizeTest && ArchTest),
|
|
|
|
// During run phase on the host, we don't have compile time info about current arch or wave size.
|
|
// We have to trust that the runtime params obtained through HipDevice will dispatch correctly for
|
|
// the current arch and wave size.
|
|
EnableRun = (TBlockXTest && MinTBlockTest && ArchTest),
|
|
};
|
|
|
|
#if !NDEBUG
|
|
static constexpr void debugGlobalPredicates()
|
|
{
|
|
std::cout << "Global Predicates:\n";
|
|
std::cout << "TBlockXTest: " << (bool)GlobalPredicates::TBlockXTest << std::endl;
|
|
std::cout << "MinTBlockTest: " << (bool)GlobalPredicates::MinTBlockTest << std::endl;
|
|
std::cout << "CurrentArchTest: " << (bool)GlobalPredicates::CurrentArchTest
|
|
<< std::endl;
|
|
std::cout << "CurrentWaveSizeTest: " << (bool)GlobalPredicates::CurrentWaveSizeTest
|
|
<< std::endl;
|
|
std::cout << "ArchTest: " << (bool)GlobalPredicates::ArchTest << std::endl;
|
|
std::cout << "EnableBuild: " << (bool)GlobalPredicates::EnableBuild << std::endl;
|
|
std::cout << "EnableRun: " << (bool)GlobalPredicates::EnableRun << std::endl;
|
|
}
|
|
#endif // !NDEBUG
|
|
|
|
enum struct Gfx9Predicates : bool
|
|
{
|
|
ArchTest = (bool)TestTraits::Arch::IsGfx9,
|
|
|
|
WaveSizeTest = (bool)TestTraits::Arch::IsWave64,
|
|
|
|
TBlockTest
|
|
= (TBlockX * TBlockY >= Constants::AMDGCN_WAVE_SIZE_64) && (TBlockX * TBlockY <= 256u),
|
|
|
|
InputTypesTest
|
|
= (bool)TestTraits::InputType::IsFloat8 || (bool)TestTraits::InputType::IsBFloat8
|
|
|| (bool)TestTraits::InputType::IsFloat8Fnuz
|
|
|| (bool)TestTraits::InputType::IsBFloat8Fnuz || (bool)TestTraits::InputType::IsInt8
|
|
|| (bool)TestTraits::InputType::IsFloat16 || (bool)TestTraits::InputType::IsBFloat16
|
|
|| (bool)TestTraits::InputType::IsFloat32 || (bool)TestTraits::InputType::IsXFloat32
|
|
|| (bool)TestTraits::InputType::IsFloat64,
|
|
|
|
// Gfx940/1/2 arch req'd for float8_fnuz_t, bfloat8_fnuz_t and xfloat32_t
|
|
F8FnuzXF32ArchTest = !((bool)TestTraits::InputType::IsFloat8Fnuz
|
|
|| (bool)TestTraits::InputType::IsBFloat8Fnuz
|
|
|| (bool)TestTraits::InputType::IsXFloat32)
|
|
|| (bool)TestTraits::Arch::IsGfx942,
|
|
|
|
// Gfx950 arch req'd for float8_t
|
|
F8ArchTest
|
|
= !((bool)TestTraits::InputType::IsFloat8 || (bool)TestTraits::InputType::IsBFloat8)
|
|
|| (bool)TestTraits::Arch::IsGfx950,
|
|
|
|
// All archs except gfx908 can run float64_t
|
|
F64ArchTest
|
|
= !(bool)TestTraits::InputType::IsFloat64 || !(bool)TestTraits::Arch::IsGfx908,
|
|
|
|
// General int8_t block size
|
|
// BlockM/N = 16; Block K >= 16
|
|
// BlockM/N = 32; Block K >= 8
|
|
I8BlockSizeTest = !((bool)TestTraits::InputType::IsInt8)
|
|
|| ((bool)TestTraits::BlockSizes::isBlockMN16 && (BlockK >= 16u)
|
|
&& (BlockK % 16u == 0u))
|
|
|| ((bool)TestTraits::BlockSizes::isBlockMN32 && (BlockK >= 16u)
|
|
&& (BlockK % 16u == 0u)),
|
|
|
|
// Follow-on to gfx940/1/2/950 int8_t.
|
|
// BlockM/N = 16; Block K >= 32
|
|
// BlockM/N = 32; Block K >= 16
|
|
ExtI8BlockSizeTest
|
|
= !((bool)TestTraits::InputType::IsInt8
|
|
&& ((bool)TestTraits::Arch::IsGfx942 || (bool)TestTraits::Arch::IsGfx950))
|
|
|| ((bool)TestTraits::BlockSizes::isBlockMN16 && (BlockK >= 32u)
|
|
&& (BlockK % 32u == 0u))
|
|
|| ((bool)TestTraits::BlockSizes::isBlockMN32 && (BlockK >= 16u)
|
|
&& (BlockK % 16u == 0u)),
|
|
|
|
// General float8_t / bfloat8_t block size
|
|
// BlockM/N = 16; Block K >= 32
|
|
// BlockM/N = 32; Block K >= 16
|
|
F8BlockSizeTest
|
|
= !((bool)TestTraits::InputType::IsFloat8 || (bool)TestTraits::InputType::IsBFloat8
|
|
|| (bool)TestTraits::InputType::IsFloat8Fnuz
|
|
|| (bool)TestTraits::InputType::IsBFloat8Fnuz)
|
|
|| ((bool)TestTraits::BlockSizes::isBlockMN16 && (BlockK >= 32u)
|
|
&& (BlockK % 32u == 0u))
|
|
|| ((bool)TestTraits::BlockSizes::isBlockMN32 && (BlockK >= 16u)
|
|
&& (BlockK % 16u == 0u)),
|
|
|
|
// General float16_t / hfloat16_t / bfloat16_t block size
|
|
// BlockM/N = 16; Block K >= 16
|
|
// BlockM/N = 32; Block K >= 8
|
|
F16BlockSizeTest
|
|
= !((bool)TestTraits::InputType::IsFloat16 || (bool)TestTraits::InputType::IsBFloat16)
|
|
|| ((bool)TestTraits::BlockSizes::isBlockMN16 && (BlockK >= 16u)
|
|
&& (BlockK % 16u == 0u))
|
|
|| ((bool)TestTraits::BlockSizes::isBlockMN32 && (BlockK >= 16u)
|
|
&& (BlockK % 16u == 0u)),
|
|
|
|
// Older gfx908 arch has half BlockK on bfloat16_t
|
|
// BlockM/N = 16; Block K >= 8
|
|
// BlockM/N = 32; Block K >= 4
|
|
Gfx908BF16BlockSizeTest
|
|
= !((bool)TestTraits::InputType::IsBFloat16 && (bool)TestTraits::Arch::IsGfx908)
|
|
|| ((bool)TestTraits::BlockSizes::isBlockMN16 && (BlockK >= 8u)
|
|
&& (BlockK % 8u == 0u))
|
|
|| ((bool)TestTraits::BlockSizes::isBlockMN32 && (BlockK >= 4u)
|
|
&& (BlockK % 4u == 0u)),
|
|
|
|
// General float32_t block size
|
|
// BlockM/N = 16; Block K >= 4
|
|
// BlockM/N = 32; Block K >= 2
|
|
F32BlockSizeTest = !((bool)TestTraits::InputType::IsFloat32)
|
|
|| ((bool)TestTraits::BlockSizes::isBlockMN16 && (BlockK >= 4u)
|
|
&& (BlockK % 4u == 0u))
|
|
|| ((bool)TestTraits::BlockSizes::isBlockMN32 && (BlockK >= 8u)
|
|
&& (BlockK % 8u == 0u)),
|
|
|
|
// General xfloat32_t block size
|
|
// BlockM/N = 16; Block K >= 8
|
|
// BlockM/N = 32; Block K >= 4
|
|
XF32BlockSizeTest = !((bool)TestTraits::InputType::IsXFloat32)
|
|
|| ((bool)TestTraits::BlockSizes::isBlockMN16 && (BlockK >= 8u)
|
|
&& (BlockK % 8u == 0u))
|
|
|| ((bool)TestTraits::BlockSizes::isBlockMN32 && (BlockK >= 4u)
|
|
&& (BlockK % 4u == 0u)),
|
|
|
|
// General float64_t block size
|
|
// BlockM/N = 16; Block K >= 4
|
|
F64BlockSizeTest = !((bool)TestTraits::InputType::IsFloat64)
|
|
|| ((bool)TestTraits::BlockSizes::isBlockMN16 && (BlockK >= 4u)
|
|
&& (BlockK % 4u == 0u)),
|
|
|
|
Enable = (ArchTest && WaveSizeTest && TBlockTest && InputTypesTest && F8FnuzXF32ArchTest
|
|
&& F8ArchTest && F64ArchTest && I8BlockSizeTest && ExtI8BlockSizeTest
|
|
&& F8BlockSizeTest && F16BlockSizeTest && Gfx908BF16BlockSizeTest
|
|
&& F32BlockSizeTest && XF32BlockSizeTest && F64BlockSizeTest)
|
|
};
|
|
|
|
#if !NDEBUG
|
|
static constexpr void debugGfx9Predicates()
|
|
{
|
|
std::cout << "Gfx9 Predicates:\n";
|
|
std::cout << "ArchTest: " << (bool)Gfx9Predicates::ArchTest << std::endl;
|
|
std::cout << "WaveSizeTest: " << (bool)Gfx9Predicates::WaveSizeTest << std::endl;
|
|
std::cout << "TBlockTest: " << (bool)Gfx9Predicates::TBlockTest << std::endl;
|
|
std::cout << "InputTypesTest: " << (bool)Gfx9Predicates::InputTypesTest << std::endl;
|
|
std::cout << "F8XF32ArchTest: " << (bool)Gfx9Predicates::F8XF32ArchTest << std::endl;
|
|
std::cout << "F64ArchTest: " << (bool)Gfx9Predicates::F64ArchTest << std::endl;
|
|
std::cout << "I8BlockSizeTest: " << (bool)Gfx9Predicates::I8BlockSizeTest << std::endl;
|
|
std::cout << "Gfx940I8BlockSizeTest: " << (bool)Gfx9Predicates::Gfx940I8BlockSizeTest
|
|
<< std::endl;
|
|
std::cout << "F8BlockSizeTest: " << (bool)Gfx9Predicates::F8BlockSizeTest << std::endl;
|
|
std::cout << "F16BlockSizeTest: " << (bool)Gfx9Predicates::F16BlockSizeTest
|
|
<< std::endl;
|
|
std::cout << "Gfx908BF16BlockSizeTest: "
|
|
<< (bool)Gfx9Predicates::Gfx908BF16BlockSizeTest << std::endl;
|
|
std::cout << "F32BlockSizeTest: " << (bool)Gfx9Predicates::F32BlockSizeTest
|
|
<< std::endl;
|
|
std::cout << "XF32BlockSizeTest: " << (bool)Gfx9Predicates::XF32BlockSizeTest
|
|
<< std::endl;
|
|
std::cout << "F64BlockSizeTest: " << (bool)Gfx9Predicates::F64BlockSizeTest
|
|
<< std::endl;
|
|
std::cout << "Enable: " << (bool)Gfx9Predicates::Enable << std::endl;
|
|
}
|
|
#endif // !NDEBUG
|
|
|
|
enum struct Gfx11Predicates : bool
|
|
{
|
|
// Valid for gfx11 only
|
|
ArchTest = (bool)TestTraits::Arch::IsGfx11,
|
|
|
|
// Wave size on gfx11 is 32
|
|
WaveSizeTest = (bool)TestTraits::Arch::IsWave32,
|
|
|
|
// Max recommended TBlock size is 256
|
|
TBlockTest
|
|
= (TBlockX * TBlockY >= Constants::AMDGCN_WAVE_SIZE_32) && (TBlockX * TBlockY <= 256u),
|
|
|
|
// Input types supported
|
|
InputTypesTest = (bool)TestTraits::InputType::IsInt8
|
|
|| (bool)TestTraits::InputType::IsFloat16
|
|
|| (bool)TestTraits::InputType::IsBFloat16,
|
|
|
|
// General int8_t block size
|
|
// BlockM/N = 16; Block K >= 16
|
|
I8BlockSizeTest = !((bool)TestTraits::InputType::IsInt8)
|
|
|| ((bool)TestTraits::BlockSizes::isBlockMN16 && (BlockK >= 16u)
|
|
&& (BlockK % 16u == 0u)),
|
|
|
|
// General float16_t / hfloat16_t / bfloat16_t block size
|
|
// BlockM/N = 16; Block K >= 16
|
|
F16BlockSizeTest
|
|
= !((bool)TestTraits::InputType::IsFloat16 || (bool)TestTraits::InputType::IsBFloat16)
|
|
|| ((bool)TestTraits::BlockSizes::isBlockMN16 && (BlockK >= 16u)
|
|
&& (BlockK % 16u == 0u)),
|
|
|
|
Enable = (ArchTest && WaveSizeTest && TBlockTest && InputTypesTest && I8BlockSizeTest
|
|
&& F16BlockSizeTest)
|
|
};
|
|
|
|
#if !NDEBUG
|
|
static constexpr void debugGfx11Predicates()
|
|
{
|
|
std::cout << "Gfx11 Predicates:\n";
|
|
std::cout << "ArchTest: " << (bool)Gfx11Predicates::ArchTest << std::endl;
|
|
std::cout << "WaveSizeTest: " << (bool)Gfx11Predicates::WaveSizeTest << std::endl;
|
|
std::cout << "TBlockTest: " << (bool)Gfx11Predicates::TBlockTest << std::endl;
|
|
std::cout << "InputTypesTest: " << (bool)Gfx11Predicates::InputTypesTest << std::endl;
|
|
std::cout << "I8BlockSizeTest: " << (bool)Gfx11Predicates::I8BlockSizeTest << std::endl;
|
|
std::cout << "F16BlockSizeTest: " << (bool)Gfx11Predicates::F16BlockSizeTest
|
|
<< std::endl;
|
|
std::cout << "Enable: " << (bool)Gfx11Predicates::Enable << std::endl;
|
|
}
|
|
#endif // !NDEBUG
|
|
|
|
enum struct Gfx12Predicates : bool
|
|
{
|
|
// Valid for gfx11 only
|
|
ArchTest = (bool)TestTraits::Arch::IsGfx12,
|
|
|
|
// Wave size on gfx11 is 32
|
|
WaveSizeTest = (bool)TestTraits::Arch::IsWave32,
|
|
|
|
// Max recommended TBlock size is 256
|
|
TBlockTest
|
|
= (TBlockX * TBlockY >= Constants::AMDGCN_WAVE_SIZE_32) && (TBlockX * TBlockY <= 256u),
|
|
|
|
// Input types supported
|
|
InputTypesTest
|
|
= (bool)TestTraits::InputType::IsInt8 || (bool)TestTraits::InputType::IsFloat16
|
|
|| (bool)TestTraits::InputType::IsBFloat16 || (bool)TestTraits::InputType::IsFloat8
|
|
|| (bool)TestTraits::InputType::IsBFloat8 || (bool)TestTraits::InputType::IsUInt8,
|
|
|
|
// General int8_t block size
|
|
// BlockM/N = 16; Block K >= 16
|
|
I8BlockSizeTest
|
|
= !((bool)TestTraits::InputType::IsInt8 || (bool)TestTraits::InputType::IsFloat8
|
|
|| (bool)TestTraits::InputType::IsBFloat8)
|
|
|| ((bool)TestTraits::BlockSizes::isBlockMN16 && (BlockK >= 16u)
|
|
&& (BlockK % 16u == 0u)),
|
|
|
|
// General float16_t / hfloat16_t / bfloat16_t block size
|
|
// BlockM/N = 16; Block K >= 16
|
|
F16BlockSizeTest
|
|
= !((bool)TestTraits::InputType::IsFloat16 || (bool)TestTraits::InputType::IsBFloat16)
|
|
|| ((bool)TestTraits::BlockSizes::isBlockMN16 && (BlockK >= 16u)
|
|
&& (BlockK % 16u == 0u)),
|
|
|
|
Enable = (ArchTest && WaveSizeTest && TBlockTest && InputTypesTest && I8BlockSizeTest
|
|
&& F16BlockSizeTest)
|
|
};
|
|
|
|
#if !NDEBUG
|
|
static constexpr void debugGfx12Predicates()
|
|
{
|
|
std::cout << "Gfx12 Predicates:\n";
|
|
std::cout << "ArchTest: " << (bool)Gfx12Predicates::ArchTest << std::endl;
|
|
std::cout << "WaveSizeTest: " << (bool)Gfx12Predicates::WaveSizeTest << std::endl;
|
|
std::cout << "TBlockTest: " << (bool)Gfx12Predicates::TBlockTest << std::endl;
|
|
std::cout << "InputTypesTest: " << (bool)Gfx12Predicates::InputTypesTest << std::endl;
|
|
std::cout << "I8BlockSizeTest: " << (bool)Gfx12Predicates::I8BlockSizeTest << std::endl;
|
|
std::cout << "F16BlockSizeTest: " << (bool)Gfx12Predicates::F16BlockSizeTest
|
|
<< std::endl;
|
|
std::cout << "Enable: " << (bool)Gfx12Predicates::Enable << std::endl;
|
|
}
|
|
#endif // !NDEBUG
|
|
|
|
public:
|
|
constexpr static bool enableBuild()
|
|
{
|
|
return ((bool)GlobalPredicates::EnableBuild
|
|
&& ((bool)Gfx9Predicates::Enable || (bool)Gfx11Predicates::Enable
|
|
|| (bool)Gfx12Predicates::Enable));
|
|
}
|
|
|
|
constexpr static bool enableRun()
|
|
{
|
|
return ((bool)GlobalPredicates::EnableRun
|
|
&& ((bool)Gfx9Predicates::Enable || (bool)Gfx11Predicates::Enable
|
|
|| (bool)Gfx12Predicates::Enable));
|
|
}
|
|
|
|
#if !NDEBUG
|
|
constexpr static void debugPredicates()
|
|
{
|
|
debugGlobalPredicates();
|
|
debugGfx9Predicates();
|
|
debugGfx11Predicates();
|
|
debugGfx12Predicates();
|
|
|
|
std::cout << "Overall enable build: " << enableBuild() << std::endl;
|
|
std::cout << "Overall enable run: " << enableRun() << std::endl;
|
|
}
|
|
#endif // !NDEBUG
|
|
};
|
|
|
|
} // namespace rocwmma
|
|
|
|
#endif // ROCWMMA_TEST_GEMM_PREDICATES_BASE_HPP
|