Unit_hipMalloc_AllocateMoreThanTotalRAM is an invalid test. hipMalloc is able to allocate more than RAM size if VRam is larger, unless the test is meant to only allocate on host device.
Previously this test may have passed because of an incorrect check on hipMalloc limit: we had incorrectly limited allocation to physical ram size. This is no longer correct. We could always allocate Vram limit or sometimes system ram limit if Vram is full, fixed in https://github.com/AMD-ROCm-Internal/clr/pull/122
Co-authored-by: Jimbo Xie <jiabaxie@amd.com>
The test wrongly assumed that the reserved memory should always
be equal to the allocation size. However, suballocations are possible
Change-Id: I1d1f3c3cdb231188f8b35ed64ffe4e2a3a2f16f9
* This also removes test that is invalid on GPUs
- As it requires releasing memory on the device (which it does not do)
- Copying back the kernel argument to the host
- It is also a NO-OP as the queue has no pushes to pop
Change-Id: I8050cd400d34f78d5afce71bf0e3d57d1878bef0
Unit_hipMemPoolApi_BasicAlloc expects to work on device 0, but other
tests will set not-0 devices in mgpu. This leads to hang of
Unit_hipMemPoolApi_BasicAlloc. Fix by set device 0 in head code
of Unit_hipMemPoolApi_BasicAlloc.
SWDEV-508872 - Fix Perf_hipPerfMemFill_test
When mem size is 2G, the test is so slow that it looks like stuckness.
Set top mem size to 1G can make the test pass in an acceptiable time.
Change-Id: Ie26dbf597e5ba8cb898d1aae5ed5ecf0267c3228
The test sets the hipMemCreate and hipMemSetAccess properties
to device but accesses the mapped memory directly from the host.
Change-Id: I2ae51a2658acc01aff39fc371ba97191cc9b3878
Module and stream should be created on the same device, while hipModuleGetFunction can be called from any device.
Change-Id: I4a424f9631678684292e40e05ffbaf3d887aea55
- For H2D transfer cases, we cannot enfore ErrorNotReady behavior if the
source is unpinned. In that case the driver can sync depending on if
it pins or stages the source buffer.
Change-Id: I436a4b667bff172cfbadc58631a46a215c40fcb1
Add dynamic arrays of devices to avoid array overflow.
Fixed the following tests:
Unit_hipDeviceGetByPCIBusId_WrongBusID)
Unit_hipSetDevice_ValidVisibleDeviceList
Unit_hipSetDevice_MinRvdMaxHvdDevicesList
Change-Id: I73d022321609d20ef2bd419f2cff287f6002a4d2
Fix Unit_hipExtMallocWithFlags_CoherentTst hang on device fine
grain access memory because host builtin atomic functions malfunctions
on it.
Fix SWDEV-507938 also.
Change-Id: Ie26dbf6b7e5aa8cb898d11ae5ed5ecf1267c3228
1) Cannot assume that blockIdx.x = 0 and threadIdx.x = 0 will be
run first in TestKernel. Initialize flags outside kernels.
2) use __hip_atomic_store in device code and __atomic_store_n for host.
Change-Id: If4e9274d2c16af55b53a626c3ba2fb0db7052d4b
Guan, Zichuan
Hernandez, Gerardo
Hila, Nino
Sang, Tao
Stojiljkovic, Vladana
Xie, Jiabao(Jimbo)
Mallya, Ameya Keshava
Gupta, Maneesh
Brzak, Branislav
Ioannis Assiouras
German Andryeyev
Aidan Belton-Schure
Jaydeep Patel
Julia Jiang
Aidan Belton-Schure
Marko Arandjelovic
Marko Arandjelovic
Saleel Kudchadker
Rambabu Swargam
Pengda Xie
Dragoslav Sicarov
Anusha GodavarthySurya
Rahul Manocha