-Move staging buffer locks inside the staging buffer code.
-Remove dedicated per-device completion_signal + per-device lock -
instead allocated signal from the per-stream pool. This elimintes
the lock and allows more concurrency.
-remove switch HIP_DISABLE_BIDIR_MEMCPY
[ROCm/hip commit: 0af4d3623f]
- refactor staging buffer to operate on hsa* data structures not
hc::accelerator.
- use hsa_memory_allocate to allocate staging buffers rather than
am_alloc.
- Refactor device reset with single member function. Don't reallocate
staging buffers on reset.
- Properly track dependencies based on command type. Add new deps for
H2D and D2D rather than overloading H2D.
[ROCm/hip commit: 7d500599fa]
Still #include staging_buffer.cpp into hip_hcc.cpp.
Directed tests compile hip_hcc to static library and use the library.
[ROCm/hip commit: 28ee7aff71]
- Control with HIP_DB=mask (env var). See src/hip_hcc.cpp for mask
values:
#define DB_API 0 /* 0x01 - shortcut to enable HIP_TRACE_API on single switch */
#define DB_SYNC 1 /* 0x02 - trace synchronization pieces */
#define DB_MEM 2 /* 0x04 - trace memory allocation / deallocation */
#define DB_COPY1 3 /* 0x08 - trace memory copy commands. . */
#define DB_SIGNAL 4 /* 0x10 - trace signal pool commands */
- Combine with HIP_TRACE to see debug with API trace.
- Use colors to distinguish different flows of debug.
- Add define COMPILE_DB_TRACE to allow removing all debug at compile-time
[ROCm/hip commit: aa03e1264c]
In this change, the cpu memcpy will wait until all the commands in the current stream are done.
Note that, it only waits on current stream. But not on other streams.
[ROCm/hip commit: 180bc32db0]
Add dependency for host-to-host copy.
Add debug mode for HIP_DISABLE_HW_COPY_DEP and
HIP_DISABLE_HW_KERNEL_DEP - setting these to -1 now ignores
all dependencies.
[ROCm/hip commit: ba9ad6be80]
hipEnvVar is the base test case, to be called by hipEnvVarDriver
at the run time.
Test case includes tests for normal use case of the environment
variable, invalid value/sequence and use CUDA_VISIBLE_DEVICES as a
alternative.
[ROCm/hip commit: 39b44cb484]
- Add chicken bits to use host-side dependency management.
- Add optional PinInPlace path for unpinned copies
- Synchronize before pinned memcpy path.
- Add mutex to protect two threads launching to same stream.
[ROCm/hip commit: 8b64c0dc62]
On HIP path property obtaining done through hsa_iterate_agents and counting the devices of HSA_DEVICE_TYPE_GPU type.
P.S.
On multi-boards systems it might be problems with detection what board a GPU plugged into (not tested).
[ROCm/hip commit: 57e212606d]
- Add device-side signal waits when transitioning between command classes
(Kernel, H2D copy, D2H copy).
- Support waiting in staged memory copies as well.
- Add several chicken bits to control implementation:
- HIP_DISABLE_ENQ_BARRIER
- HIP_DISABLE_BIDIR_MEMCPY
- HIP_ONESHOT_COPY_DEP
- Refactor signal pool to support efficient deallocation based on
signsequnm.
- Deallocate copy signals on eventSynchronize.
- Improve copy tests, add pingpong.
[ROCm/hip commit: 549b18ce77]
Ben Sander
Aditya Atluri
pensun
Evgeny Mankov