intel/compute-runtime
1
0
Fork 0
mirror of https://github.com/intel/compute-runtime.git synced 2026-01-04 15:53:45 +08:00
Code Issues Packages Projects Releases Wiki Activity

Fix: Unify fp16/fp32/fp64 flags across all platforms

Browse Source 
Unify fp16/fp32/fp64 across all platforms. The capabilities indicated by
those flags now refer to both emulated and native-supported (HW) ones:

- Global/local atomic load: HW support on all platforms (handled by native
i16 atomic or) for FP16, FP32 and FP64.
- Global/local atomic store: HW support on all platforms (handled by
native i16 atomic exchange) for FP16, FP32 and FP64.
- Global/local atomic compare/exchange: HW support on all platforms
for FP32.
- Global/local atomic min/max: Emulation support on all platforms for
FP64, HW support on all platforms for FP32, HW support on XE+ platforms
and emulation support on all others for FP16.
- Global atomic add: HW support for PVC+ platforms, emulation support on
all other platforms for FP64, HW support on XE+ platforms and emulation
support on all other platforms for FP32.
- Local atomic add: Emulation on all platforms for both FP64 and FP32.

Signed-off-by: Kacper Nowak <kacper.nowak@intel.com>
Related-To: NEO-7734
This commit is contained in:
Kacper Nowak
2023-04-18 13:53:43 +00:00
committed by Compute-Runtime-Automation
parent 1ddfc9af39
commit 83e9a148ca
14 changed files with 127 additions and 92 deletions
Download Patch File Download Diff File Expand all files Collapse all files

30
level_zero/core/test/unit_tests/gen11/test_module_gen11.cpp
View File

@@ -1,10 +1,11 @@
/*
* Copyright (C) 2021-2022 Intel Corporation
* Copyright (C) 2021-2023 Intel Corporation
*
* SPDX-License-Identifier: MIT
*
*/
#include "shared/source/kernel/kernel_properties.h"
#include "shared/test/common/test_macros/hw_test.h"
#include "level_zero/core/test/unit_tests/fixtures/device_fixture.h"
@@ -22,9 +23,30 @@ HWTEST2_F(KernelPropertyTest, givenKernelExtendedPropertiesStructureWhenKernelPr
ze_result_t res = device->getKernelProperties(&kernelProperties);
EXPECT_EQ(res, ZE_RESULT_SUCCESS);
EXPECT_EQ(0u, kernelExtendedProperties.fp16Flags);
EXPECT_EQ(0u, kernelExtendedProperties.fp32Flags);
EXPECT_EQ(0u, kernelExtendedProperties.fp64Flags);
const auto &fp16Properties = kernelExtendedProperties.fp16Flags;
EXPECT_TRUE(fp16Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_LOAD_STORE);
EXPECT_TRUE(fp16Properties & FP_ATOMIC_EXT_FLAG_LOCAL_LOAD_STORE);
EXPECT_TRUE(fp16Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_MIN_MAX);
EXPECT_TRUE(fp16Properties & FP_ATOMIC_EXT_FLAG_LOCAL_MIN_MAX);
EXPECT_FALSE(fp16Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_ADD);
EXPECT_FALSE(fp16Properties & FP_ATOMIC_EXT_FLAG_LOCAL_ADD);
const auto &fp32Properties = kernelExtendedProperties.fp32Flags;
EXPECT_TRUE(fp32Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_LOAD_STORE);
EXPECT_TRUE(fp32Properties & FP_ATOMIC_EXT_FLAG_LOCAL_LOAD_STORE);
EXPECT_TRUE(fp32Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_MIN_MAX);
EXPECT_TRUE(fp32Properties & FP_ATOMIC_EXT_FLAG_LOCAL_MIN_MAX);
EXPECT_TRUE(fp32Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_ADD);
EXPECT_TRUE(fp32Properties & FP_ATOMIC_EXT_FLAG_LOCAL_ADD);
const auto &fp64Properties = kernelExtendedProperties.fp64Flags;
EXPECT_TRUE(fp64Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_LOAD_STORE);
EXPECT_TRUE(fp64Properties & FP_ATOMIC_EXT_FLAG_LOCAL_LOAD_STORE);
EXPECT_TRUE(fp64Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_MIN_MAX);
EXPECT_TRUE(fp64Properties & FP_ATOMIC_EXT_FLAG_LOCAL_MIN_MAX);
EXPECT_TRUE(fp64Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_ADD);
EXPECT_TRUE(fp64Properties & FP_ATOMIC_EXT_FLAG_LOCAL_ADD);
}
} // namespace ult

30
level_zero/core/test/unit_tests/gen12lp/test_module_gen12lp.cpp
View File

@@ -1,10 +1,11 @@
/*
* Copyright (C) 2021-2022 Intel Corporation
* Copyright (C) 2021-2023 Intel Corporation
*
* SPDX-License-Identifier: MIT
*
*/
#include "shared/source/kernel/kernel_properties.h"
#include "shared/test/common/test_macros/hw_test.h"
#include "level_zero/core/test/unit_tests/fixtures/device_fixture.h"
@@ -22,9 +23,30 @@ HWTEST2_F(KernelPropertyTest, givenKernelExtendedPropertiesStructureWhenKernelPr
ze_result_t res = device->getKernelProperties(&kernelProperties);
EXPECT_EQ(res, ZE_RESULT_SUCCESS);
EXPECT_EQ(0u, kernelExtendedProperties.fp16Flags);
EXPECT_EQ(0u, kernelExtendedProperties.fp32Flags);
EXPECT_EQ(0u, kernelExtendedProperties.fp64Flags);
const auto &fp16Properties = kernelExtendedProperties.fp16Flags;
EXPECT_TRUE(fp16Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_LOAD_STORE);
EXPECT_TRUE(fp16Properties & FP_ATOMIC_EXT_FLAG_LOCAL_LOAD_STORE);
EXPECT_TRUE(fp16Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_MIN_MAX);
EXPECT_TRUE(fp16Properties & FP_ATOMIC_EXT_FLAG_LOCAL_MIN_MAX);
EXPECT_FALSE(fp16Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_ADD);
EXPECT_FALSE(fp16Properties & FP_ATOMIC_EXT_FLAG_LOCAL_ADD);
const auto &fp32Properties = kernelExtendedProperties.fp32Flags;
EXPECT_TRUE(fp32Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_LOAD_STORE);
EXPECT_TRUE(fp32Properties & FP_ATOMIC_EXT_FLAG_LOCAL_LOAD_STORE);
EXPECT_TRUE(fp32Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_MIN_MAX);
EXPECT_TRUE(fp32Properties & FP_ATOMIC_EXT_FLAG_LOCAL_MIN_MAX);
EXPECT_TRUE(fp32Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_ADD);
EXPECT_TRUE(fp32Properties & FP_ATOMIC_EXT_FLAG_LOCAL_ADD);
const auto &fp64Properties = kernelExtendedProperties.fp64Flags;
EXPECT_TRUE(fp64Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_LOAD_STORE);
EXPECT_TRUE(fp64Properties & FP_ATOMIC_EXT_FLAG_LOCAL_LOAD_STORE);
EXPECT_TRUE(fp64Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_MIN_MAX);
EXPECT_TRUE(fp64Properties & FP_ATOMIC_EXT_FLAG_LOCAL_MIN_MAX);
EXPECT_TRUE(fp64Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_ADD);
EXPECT_TRUE(fp64Properties & FP_ATOMIC_EXT_FLAG_LOCAL_ADD);
}
} // namespace ult

30
level_zero/core/test/unit_tests/gen9/test_module_gen9.cpp
View File

@@ -1,10 +1,11 @@
/*
* Copyright (C) 2021-2022 Intel Corporation
* Copyright (C) 2021-2023 Intel Corporation
*
* SPDX-License-Identifier: MIT
*
*/
#include "shared/source/kernel/kernel_properties.h"
#include "shared/test/common/test_macros/hw_test.h"
#include "level_zero/core/test/unit_tests/fixtures/device_fixture.h"
@@ -29,9 +30,30 @@ HWTEST2_F(KernelPropertyTest, givenKernelExtendedPropertiesStructureWhenKernelPr
ze_result_t res = device->getKernelProperties(&kernelProperties);
EXPECT_EQ(res, ZE_RESULT_SUCCESS);
EXPECT_EQ(0u, kernelExtendedProperties.fp16Flags);
EXPECT_EQ(0u, kernelExtendedProperties.fp32Flags);
EXPECT_EQ(0u, kernelExtendedProperties.fp64Flags);
const auto &fp16Properties = kernelExtendedProperties.fp16Flags;
EXPECT_TRUE(fp16Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_LOAD_STORE);
EXPECT_TRUE(fp16Properties & FP_ATOMIC_EXT_FLAG_LOCAL_LOAD_STORE);
EXPECT_TRUE(fp16Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_MIN_MAX);
EXPECT_TRUE(fp16Properties & FP_ATOMIC_EXT_FLAG_LOCAL_MIN_MAX);
EXPECT_FALSE(fp16Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_ADD);
EXPECT_FALSE(fp16Properties & FP_ATOMIC_EXT_FLAG_LOCAL_ADD);
const auto &fp32Properties = kernelExtendedProperties.fp32Flags;
EXPECT_TRUE(fp32Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_LOAD_STORE);
EXPECT_TRUE(fp32Properties & FP_ATOMIC_EXT_FLAG_LOCAL_LOAD_STORE);
EXPECT_TRUE(fp32Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_MIN_MAX);
EXPECT_TRUE(fp32Properties & FP_ATOMIC_EXT_FLAG_LOCAL_MIN_MAX);
EXPECT_TRUE(fp32Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_ADD);
EXPECT_TRUE(fp32Properties & FP_ATOMIC_EXT_FLAG_LOCAL_ADD);
const auto &fp64Properties = kernelExtendedProperties.fp64Flags;
EXPECT_TRUE(fp64Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_LOAD_STORE);
EXPECT_TRUE(fp64Properties & FP_ATOMIC_EXT_FLAG_LOCAL_LOAD_STORE);
EXPECT_TRUE(fp64Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_MIN_MAX);
EXPECT_TRUE(fp64Properties & FP_ATOMIC_EXT_FLAG_LOCAL_MIN_MAX);
EXPECT_TRUE(fp64Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_ADD);
EXPECT_TRUE(fp64Properties & FP_ATOMIC_EXT_FLAG_LOCAL_ADD);
}
} // namespace ult

42
level_zero/core/test/unit_tests/xe_hpc_core/test_module_xe_hpc_core.cpp
View File

@@ -1,5 +1,5 @@
/*
* Copyright (C) 2021-2022 Intel Corporation
* Copyright (C) 2021-2023 Intel Corporation
*
* SPDX-License-Identifier: MIT
*
@@ -23,26 +23,30 @@ HWTEST2_F(KernelPropertyTest, givenKernelExtendedPropertiesStructureWhenKernelPr
ze_result_t res = device->getKernelProperties(&kernelProperties);
EXPECT_EQ(res, ZE_RESULT_SUCCESS);
EXPECT_TRUE(kernelExtendedProperties.fp16Flags & FP_ATOMIC_EXT_FLAG_GLOBAL_LOAD_STORE);
EXPECT_TRUE(kernelExtendedProperties.fp16Flags & FP_ATOMIC_EXT_FLAG_GLOBAL_ADD);
EXPECT_TRUE(kernelExtendedProperties.fp16Flags & FP_ATOMIC_EXT_FLAG_GLOBAL_MIN_MAX);
EXPECT_FALSE(kernelExtendedProperties.fp16Flags & FP_ATOMIC_EXT_FLAG_LOCAL_LOAD_STORE);
EXPECT_FALSE(kernelExtendedProperties.fp16Flags & FP_ATOMIC_EXT_FLAG_LOCAL_ADD);
EXPECT_FALSE(kernelExtendedProperties.fp16Flags & FP_ATOMIC_EXT_FLAG_LOCAL_MIN_MAX);
const auto &fp16Properties = kernelExtendedProperties.fp16Flags;
EXPECT_TRUE(fp16Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_LOAD_STORE);
EXPECT_TRUE(fp16Properties & FP_ATOMIC_EXT_FLAG_LOCAL_LOAD_STORE);
EXPECT_TRUE(fp16Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_MIN_MAX);
EXPECT_TRUE(fp16Properties & FP_ATOMIC_EXT_FLAG_LOCAL_MIN_MAX);
EXPECT_TRUE(kernelExtendedProperties.fp32Flags & FP_ATOMIC_EXT_FLAG_GLOBAL_LOAD_STORE);
EXPECT_TRUE(kernelExtendedProperties.fp32Flags & FP_ATOMIC_EXT_FLAG_GLOBAL_ADD);
EXPECT_TRUE(kernelExtendedProperties.fp32Flags & FP_ATOMIC_EXT_FLAG_GLOBAL_MIN_MAX);
EXPECT_FALSE(kernelExtendedProperties.fp32Flags & FP_ATOMIC_EXT_FLAG_LOCAL_LOAD_STORE);
EXPECT_FALSE(kernelExtendedProperties.fp32Flags & FP_ATOMIC_EXT_FLAG_LOCAL_ADD);
EXPECT_FALSE(kernelExtendedProperties.fp32Flags & FP_ATOMIC_EXT_FLAG_LOCAL_MIN_MAX);
EXPECT_FALSE(fp16Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_ADD);
EXPECT_FALSE(fp16Properties & FP_ATOMIC_EXT_FLAG_LOCAL_ADD);
EXPECT_TRUE(kernelExtendedProperties.fp64Flags & FP_ATOMIC_EXT_FLAG_GLOBAL_LOAD_STORE);
EXPECT_TRUE(kernelExtendedProperties.fp64Flags & FP_ATOMIC_EXT_FLAG_GLOBAL_ADD);
EXPECT_TRUE(kernelExtendedProperties.fp64Flags & FP_ATOMIC_EXT_FLAG_GLOBAL_MIN_MAX);
EXPECT_FALSE(kernelExtendedProperties.fp64Flags & FP_ATOMIC_EXT_FLAG_LOCAL_LOAD_STORE);
EXPECT_FALSE(kernelExtendedProperties.fp64Flags & FP_ATOMIC_EXT_FLAG_LOCAL_ADD);
EXPECT_FALSE(kernelExtendedProperties.fp64Flags & FP_ATOMIC_EXT_FLAG_LOCAL_MIN_MAX);
const auto &fp32Properties = kernelExtendedProperties.fp32Flags;
EXPECT_TRUE(fp32Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_LOAD_STORE);
EXPECT_TRUE(fp32Properties & FP_ATOMIC_EXT_FLAG_LOCAL_LOAD_STORE);
EXPECT_TRUE(fp32Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_MIN_MAX);
EXPECT_TRUE(fp32Properties & FP_ATOMIC_EXT_FLAG_LOCAL_MIN_MAX);
EXPECT_TRUE(fp32Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_ADD);
EXPECT_TRUE(fp32Properties & FP_ATOMIC_EXT_FLAG_LOCAL_ADD);
const auto &fp64Properties = kernelExtendedProperties.fp64Flags;
EXPECT_TRUE(fp64Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_LOAD_STORE);
EXPECT_TRUE(fp64Properties & FP_ATOMIC_EXT_FLAG_LOCAL_LOAD_STORE);
EXPECT_TRUE(fp64Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_MIN_MAX);
EXPECT_TRUE(fp64Properties & FP_ATOMIC_EXT_FLAG_LOCAL_MIN_MAX);
EXPECT_TRUE(fp64Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_ADD);
EXPECT_TRUE(fp64Properties & FP_ATOMIC_EXT_FLAG_LOCAL_ADD);
}
} // namespace ult

40
level_zero/core/test/unit_tests/xe_hpg_core/test_module_xe_hpg_core.cpp
View File

@@ -24,26 +24,30 @@ HWTEST2_F(KernelPropertyTest, givenKernelExtendedPropertiesStructureWhenKernelPr
ze_result_t res = device->getKernelProperties(&kernelProperties);
EXPECT_EQ(res, ZE_RESULT_SUCCESS);
EXPECT_FALSE(kernelExtendedProperties.fp16Flags & FP_ATOMIC_EXT_FLAG_GLOBAL_LOAD_STORE);
EXPECT_FALSE(kernelExtendedProperties.fp16Flags & FP_ATOMIC_EXT_FLAG_GLOBAL_ADD);
EXPECT_FALSE(kernelExtendedProperties.fp16Flags & FP_ATOMIC_EXT_FLAG_GLOBAL_MIN_MAX);
EXPECT_FALSE(kernelExtendedProperties.fp16Flags & FP_ATOMIC_EXT_FLAG_LOCAL_LOAD_STORE);
EXPECT_FALSE(kernelExtendedProperties.fp16Flags & FP_ATOMIC_EXT_FLAG_LOCAL_ADD);
EXPECT_FALSE(kernelExtendedProperties.fp16Flags & FP_ATOMIC_EXT_FLAG_LOCAL_MIN_MAX);
const auto &fp16Properties = kernelExtendedProperties.fp16Flags;
EXPECT_TRUE(fp16Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_LOAD_STORE);
EXPECT_TRUE(fp16Properties & FP_ATOMIC_EXT_FLAG_LOCAL_LOAD_STORE);
EXPECT_TRUE(fp16Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_MIN_MAX);
EXPECT_TRUE(fp16Properties & FP_ATOMIC_EXT_FLAG_LOCAL_MIN_MAX);
EXPECT_FALSE(kernelExtendedProperties.fp32Flags & FP_ATOMIC_EXT_FLAG_GLOBAL_LOAD_STORE);
EXPECT_TRUE(kernelExtendedProperties.fp32Flags & FP_ATOMIC_EXT_FLAG_GLOBAL_ADD);
EXPECT_FALSE(kernelExtendedProperties.fp32Flags & FP_ATOMIC_EXT_FLAG_GLOBAL_MIN_MAX);
EXPECT_FALSE(kernelExtendedProperties.fp32Flags & FP_ATOMIC_EXT_FLAG_LOCAL_LOAD_STORE);
EXPECT_FALSE(kernelExtendedProperties.fp32Flags & FP_ATOMIC_EXT_FLAG_LOCAL_ADD);
EXPECT_FALSE(kernelExtendedProperties.fp32Flags & FP_ATOMIC_EXT_FLAG_LOCAL_MIN_MAX);
EXPECT_FALSE(fp16Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_ADD);
EXPECT_FALSE(fp16Properties & FP_ATOMIC_EXT_FLAG_LOCAL_ADD);
EXPECT_FALSE(kernelExtendedProperties.fp64Flags & FP_ATOMIC_EXT_FLAG_GLOBAL_LOAD_STORE);
EXPECT_FALSE(kernelExtendedProperties.fp64Flags & FP_ATOMIC_EXT_FLAG_GLOBAL_ADD);
EXPECT_FALSE(kernelExtendedProperties.fp64Flags & FP_ATOMIC_EXT_FLAG_GLOBAL_MIN_MAX);
EXPECT_FALSE(kernelExtendedProperties.fp64Flags & FP_ATOMIC_EXT_FLAG_LOCAL_LOAD_STORE);
EXPECT_FALSE(kernelExtendedProperties.fp64Flags & FP_ATOMIC_EXT_FLAG_LOCAL_ADD);
EXPECT_FALSE(kernelExtendedProperties.fp64Flags & FP_ATOMIC_EXT_FLAG_LOCAL_MIN_MAX);
const auto &fp32Properties = kernelExtendedProperties.fp32Flags;
EXPECT_TRUE(fp32Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_LOAD_STORE);
EXPECT_TRUE(fp32Properties & FP_ATOMIC_EXT_FLAG_LOCAL_LOAD_STORE);
EXPECT_TRUE(fp32Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_MIN_MAX);
EXPECT_TRUE(fp32Properties & FP_ATOMIC_EXT_FLAG_LOCAL_MIN_MAX);
EXPECT_TRUE(fp32Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_ADD);
EXPECT_TRUE(fp32Properties & FP_ATOMIC_EXT_FLAG_LOCAL_ADD);
const auto &fp64Properties = kernelExtendedProperties.fp64Flags;
EXPECT_TRUE(fp64Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_LOAD_STORE);
EXPECT_TRUE(fp64Properties & FP_ATOMIC_EXT_FLAG_LOCAL_LOAD_STORE);
EXPECT_TRUE(fp64Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_MIN_MAX);
EXPECT_TRUE(fp64Properties & FP_ATOMIC_EXT_FLAG_LOCAL_MIN_MAX);
EXPECT_TRUE(fp64Properties & FP_ATOMIC_EXT_FLAG_GLOBAL_ADD);
EXPECT_TRUE(fp64Properties & FP_ATOMIC_EXT_FLAG_LOCAL_ADD);
}
HWTEST2_F(KernelPropertyTest, givenDG2WhenGetInternalOptionsThenWriteBackBuildOptionIsSet, IsDG2) {