- For gfx >= 10, one work group processor encompasses 2 CUs & hence the CUs need to be enabled in pair
Change-Id: I359df1b221b4400b260b02201b7b0385054784f3
[ROCm/hip commit: 2b09d94ea2]
- Add hipDeviceSynchronize after hipMemset with device memory, as hipMemset is asynchronous in this case
Change-Id: Ia55d106931984fed450be29c1b745c8069151f76
[ROCm/hip commit: 6e021f1adc]
- Increase allocation size to 30MB to make sure new page table allocations happen
- Change validation criteria to check for increase in VRAM usage for visible HIP devices
Change-Id: Ie5bf1a89ceda0ff50140f426235047adcca0fdea
[ROCm/hip commit: c5d856348e]
- Port image tests to catch2.
- Disable 3D filter test for gfx9a.
- Some other improvement
- Temporarily disable linear filter tests on Windows
Change-Id: Ieeddac9f8c50aa9b6b12ca9d8fd792a51973b8ca
[ROCm/hip commit: f3c171c984]
Enable more texture supports for hip-rocclr.
Skip texture tests on device that doesn't support images.
SWDEV-322257 - Fix issues of image tests skipped on
NVidia devices.
Change-Id: Ia99d06b1e97fc945f1b740e47710f4dcd70f38ca
[ROCm/hip commit: 643f8dbd36]
hiprtc is made into a separate lib and hence tests should link with it explicitly during compilation
Also, updates the hiprtc programming guide
Change-Id: Ie1f9d7250d4a8622b77ef92938fc0decbaeb72f9
[ROCm/hip commit: ef9505d68a]
Address modes are valid for array rather than linear
memory. Comment out codes of address test for linear
memory.
Add tests of mode combination(address, linear and
normorlized coordinates) in 1D, 2D and 3D.
Change-Id: I9e7e97d9bf619743c311f68f51d5bf115fdd534d
[ROCm/hip commit: e05f98967e]
Modify texture tests to pass on devices that don't support texture.
Add hipDeviceAttributeImageSupport to check image support in runtime
Change-Id: Ia89c494e651a6198a24448b59a91e046a9ebea38
[ROCm/hip commit: 5c8af57480]
Fix random failure of cooperative_streams_half_capacity by adding HIPCHECK(hipSetDevice(dev))
Change-Id: Ic4baf10a2c69f981aa7134a999c4779642852f53
[ROCm/hip commit: 57f6cd0827]
Enable the following test directed_tests/runtimeApi/module/hipModuleTexture2dDrv --tests 0x14
Change-Id: I98eabb7412f615733a999bf5e89c2c80b91cfbb2
[ROCm/hip commit: b05a0503f1]
* SWDEV-266829 - Enable more tests on AMD and NV devices
1. Enable tests on AMD and NV devices
tests/src/runtimeApi/event/hipEventMultiThreaded.cpp
Loops and threads per core are changed smaller so that test can
finish in a shorter time.
tests/src/runtimeApi/stream/hipStreamCreateWithPriority.cpp
Fix logic error on how to get priority_normal
2. Simply enable test on AMD device
tests/src/runtimeApi/memory/hipManagedKeyword.cpp
tests/src/runtimeApi/module/hipManagedKeyword.cpp
tests/src/runtimeApi/stream/hipStreamACb_MultiThread.cpp
tests/src/runtimeApi/memory/p2p_copy_coherency.cpp
3. Simply enable test on NV device
tests/src/runtimeApi/module/hipModuleLoadDataMultThreaded.cpp
4. Fix typo
tests/src/runtimeApi/stream/hipStreamAddCallbackCatch.cpp
5. Remove useless tests
tests/src/hipC.c
tests/src/hipHcc.cpp
Change-Id: Ia4406353e64d69bd34c58ebb56185701f7ce1caa
* Remove tests/src/runtimeApi/module/hipModuleLoadDataMultThreaded.cpp for cuda test
Co-authored-by: anusha GodavarthySurya <Anusha.GodavarthySurya@amd.com>
Co-authored-by: Jenkins <jenkins-compute@amd.com>
[ROCm/hip commit: 3fd16c0b5b]
Fix the following failed tests on NV,
hipCGMultiGridGroupType
hipCGMultiGridGroupTypeViaBaseType
hipCGMultiGridGroupTypeViaPublicApi
1. Fix wrong logic in kernel for both AMD and NV.
2. Remove unnecessary hipDeviceSynchronize().
3. In hipCGMultiGridGroupTypeViaBaseType.cpp, change
multi_grid_group as thread_group which is originally expected.
4. hipFree(syncResultD) is fixed as hipHostFree(syncResultD)
5. Optimize some host codes.
Change-Id: I3fe6dac35a7b14bab12adf397b7885df83d28059
[ROCm/hip commit: c57e0f8fe5]
Enable cooperativeGrps/cooperative_streams on NV.
Add test cases of the least/half/full capacity.
Verify data in terms of AMD/NV devices.
Optimize codes
Change-Id: I3fe6dbc35b7b24abb11adf297b7885df83d28154
[ROCm/hip commit: 67b3681d26]
Add test cases for filter modes: hipFilterModePoint and hipFilterModeLinear
Change-Id: I3fe6dbc35a7b14aab12adf297b7885df83d28056
[ROCm/hip commit: 48d8040b06]
1. In kernel/hipDynamicShared
Fix shared memory size and type mismatch in host and kernel.
2. In kernel/hipDynamicShared2
Cuda kernels relying on shared memory allocations over 48 KB require
to explicitly set size using hipFuncSetAttribute().
Change-Id: I4248b6cebd3dc156f9d5d427e1897da22fb964ed
[ROCm/hip commit: 5b739b0373]
ROCm CI Service Account
Satyanvesh Dittakavi
Paulius Velesko
Sarbojit2019
ansurya
Sourabh U Betigeri
TomSang
shadidashmiz
kjayapra-amd
Julia Jiang
arjun-raj-kuppala
Satyanvesh Dittakavi