[ROCm][TCT][HIP] cooperative stream test case is failing.
Make sure lockXfer() in the blit manager returns a valid value.
Port the latest PAL backend logic into the ROCr backend.
This change doesn't fix the issue, reported in the ticket.
Change-Id: I54101a824f49a2dcfbbf5414cb5b3af41745306d
- Once device assertion occurs, abort the host execution as well.
- TODO: This's the initial support. As we need to drain hostcall queue
to ensure device assertion message being flushed out, hostcall
listener needs an interface to explicitly drain its queue.
Change-Id: I8a04400aa7109bfd054ae5777c41a4abbf0db4a9
- Problem with CL_DEVICE_GLOBAL_FREE_MEMORY_AMD query.
Check if allocated memory exceeds the total size.
Change-Id: Ieed8829860663bac1acfa41d21309dff4d8772c7
1. Enable pitch workaround
2. When we use copy image, we don't need to create the custom pitch image
3. wrtBackImageBuffer_ stores device memory object, not amd image object.
Tests:
conformance kernel read / write test pass with this code change.
Change-Id: I7dca3127adde6ac83e78dd270a2256ebed55c60d
Duplicate similar blit logic from PAL path
Tests:
1D Array image read/write tests and copy image tests passed
Change-Id: I838bbde252ad0108bfeb82c0c2b669881747c0af
Don't error when querying the number of devices if there are no devices present in the system.
We should just return 0 for the number of devices in this case and let the application handle this situation.
Change-Id: I20614ade5e649f3ce9ddd970d4b38bfe296f6cdb
- Add cache free on OCL context destroy
- Remove std::mem_fun() usage, since it was removed in c++17
Change-Id: If6acd08f13a2298912ecd78fc025dcf0b32aee54
Test window will quit or show error message when run OpenCL API.
Add a workaround for the race condition with the first page
during pinning.
Change-Id: I9a27b4e173cf94c84aefcb94e255f11169453d94
The printf call in the device code is expanded by the compiler into a
series of hostcalls that together form a "message". This change
introduces the following functionality in the runtime:
1. Receive a generic message consisting of a series of hostcalls.
2. Process a printf message.
Change-Id: I9d667d6f91607a907a96e46cc5fca55734339747
~45% to 50% of Performance drop on rocBLAS_int8 test
Add support for active waits without blocking the host thread.
Change-Id: Ie7bb48dcafcb4c93d448bf74749b829b626c3578
cl_bool needed to be replaced with uint32_t instead of bool. This is due to cl_bool being a typedef of cl_uint32.
Currently clGetDeviceInfo() reports incorrect size for the return value, due to cl_bool being 4 bytes and c++ bool being 1 byte.
Change-Id: I647a4b8873627059865c84c8ca27694dbc0916de
Add MS HWS support. PAL reports just one compute engine
in that mode and runtime needs extra logic to detect RT queues.
Change-Id: I011f1f1b18dec6a7195a4f1fe939f8029bc269ae
When we're aligning rowPitch to imagePitchAlignment, rowPitch is in pixels,
but imagePitchAlignment_ is bytes, so we end up overaligning the pitch.
Convert imagePitchAlignment_ to pixels before doing any logic.
Change-Id: Ia5ab9d54bed150fe974e86b060dbadc196165b29
hip_threadfence_system passes locally with this change. This also fixes
hipHostMalloc() failures when hipHostMallocMapped flag is used.
Change-Id: Id412efe502accc7c6e7676b52c05ccb9d8fbbe67
Remove a workaround to CS_PARTIAL_FLUSH added in CL#1495187,
since PAL is no longer uses CS_PARTIAL_FLUSH.
Change-Id: I03edc7595459e19aad33b2b0901f0ebe4754d310
[hipclang-vdi-rocm][perf]~45% to 50% of Performance drop on
rocBLAS_int8 test
- Enable AMD_OPT_FLUSH optimization by default to match HCC
- Disable CPU writes to GPU memory on boards with large bar,
because it requires HDP flush tracking.
- Enable L2 cache on kernel arguments, because L2 will be
invalidated on memory reuse .
Change-Id: I124cf250bdd4d19c523ce542c163813828f8fbdc
Update a use of the deprecated amd_comgr_action_info_set_options to
instead use amd_comgr_action_info_set_option_list.
Completely remove all references to amd_comgr_action_info_set_options
and amd_comgr_action_info_get_options from the runtime.
Change-Id: I12a0803c87430722364ec22818e249caf3798c88
The last commit to replace the cl_* types with standard types
failed to correct issues introduced in the PAL and GPU backend.
Change-Id: I926997234dfbe346fc165a7bc4e1b8aabab7bac5
German Andryeyev
Michael LIAO
kjayapra-amd
Matt Arsenault
Vlad Sytchenko
Payam
Alex Xie
Todd tiantuo Li
jujiang
Jason Tang
Sameer Sahasrabuddhe
Saleel Kudchadker
Scott Linder
Laurent Morichetti