Граф коммитов

227 Коммитов

Автор SHA1 Сообщение Дата
Ben Sander dfca4faf27 disable mt streams tests (for now) 2016-03-19 03:10:31 -05:00
Ben Sander 2dc7e9ccfa Describe HIP env vars 2016-03-19 03:09:57 -05:00
Ben Sander 6472c8b915 Fix copy and sync bugs. Remove extra sync in default stream.
- NULL stream was waiting for itself to be empty before each command.
- Force "blocking" streams to wait for NULL to empty.  This was missing
  before.
- async copy was disabling itself via trueAsync=false for common cases.

Refactor:
- rename _null_stream to _default_stream.
- move some null sync function to defaultSync, move to dev member func.
2016-03-19 02:44:26 -05:00
Ben Sander 7ff5b16d2a Add beastperiteration and onesize for testing.
onesize allows running tests at one specific size.
2016-03-19 02:43:04 -05:00
Ben Sander 85fce5f21e Improve formatting - line up cols 2016-03-18 23:43:04 -05:00
Ben Sander c2102847a4 Print Pinned or Unpinned in result summary 2016-03-18 21:28:29 -05:00
Ben Sander 618556eaf9 Supported --aliged mode. Add results check for H2D and D2H. 2016-03-18 03:09:52 -05:00
Ben Sander 0af4d3623f Refactor copy code.
-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
2016-03-18 03:02:00 -05:00
Ben Sander 7d500599fa Refactor staging buffer and sync copies.
- 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.
2016-03-17 20:09:10 -05:00
Ben Sander e7586adb33 Refactor to isolate staging buffer code. 2016-03-17 00:20:56 -05:00
Ben Sander 28ee7aff71 Start separaration of staging_buffer.cpp code.
Still #include staging_buffer.cpp into hip_hcc.cpp.
Directed tests compile hip_hcc to static library and use the library.
2016-03-16 22:26:49 -05:00
Ben Sander c02bccc153 Add aligned alloc 2016-03-16 21:55:57 -05:00
Ben Sander ec50cbb3aa Checkpoint code cleanup.
- Refactor ihipStream in prep for thread-safe implementation.
- Do some work on PinInPlace implementation.
2016-03-16 21:16:29 -05:00
Ben Sander 70c5f5e3f5 print device config info 2016-03-14 23:02:49 -05:00
Ben Sander e1617b9604 Merge branch 'privatestaging' of https://github.com/AMDComputeLibraries/HIP-privatestaging into privatestaging
Conflicts:
	src/hip_hcc.cpp
	tests/src/CMakeLists.txt
2016-03-14 15:01:26 -05:00
Ben Sander 1a27e5134e enable DB, comments 2016-03-14 14:40:41 -05:00
Ben Sander 5606bee076 Add Bidir copy test and help. 2016-03-14 14:39:23 -05:00
Ben Sander ac6ed35ba0 refactor, add support for speccing xfers in bytes 2016-03-13 09:41:06 -05:00
Ben Sander b0dca6bd0c Remove ROCR_V2, assume=1 2016-03-12 04:08:13 -06:00
Ben Sander a5a2c7f33c Remove USE_AM_TRACKER (assume 1) 2016-03-12 04:04:57 -06:00
Ben Sander 250739666d Improve error reporting.
use throw with error class.
fix bug when memcpyDefault resolved to D2D copy.
2016-03-12 04:02:04 -06:00
Aditya Atluri 102f173396 Added hipHostRegister for hip with tests and added copyright 2016-03-08 12:57:22 -06:00
Aditya Atluri cbb42c6b6a Added a random test for memcpyAsync 2016-03-08 03:40:56 -06:00
Aditya Atluri 18c0e69f51 v2 added support for unpinned async memcpy 2016-03-08 03:19:50 -06:00
Aditya Atluri 45965aaa2d Revert "Fixed passing unpinned memory to async memcpy"
This reverts commit 216af71480.
2016-03-08 03:17:27 -06:00
Aditya Atluri 81824926be Merge branch 'privatestaging' of https://github.com/AMDComputeLibraries/HIP-privatestaging into privatestaging 2016-03-08 03:15:33 -06:00
Aditya Atluri 216af71480 Fixed passing unpinned memory to async memcpy 2016-03-08 03:14:49 -06:00
Ben Sander cc7517b57e Refactor memccpy tests.
- Add "hipMemcpy_simple" - some simple smoke tests.
- Run memcpy 3 times w/ different switches for kinds, sizes,
  multi-thread.
2016-03-07 17:15:48 -06:00
Aditya Atluri d9429dd4ec Added hipHostRegister flags 2016-03-07 10:52:40 -06:00
Aditya Atluri 4ed0b1cb1a Added hipHostRegister feature for CUDA backend and its tests 2016-03-07 03:42:50 -06:00
Aditya Atluri de85c80eb0 added copyrights for newly added tests 2016-03-07 02:01:25 -06:00
Aditya Atluri 3799fd55e2 Merge branch 'privatestaging' of https://github.com/AMDComputeLibraries/HIP-privatestaging into privatestaging 2016-03-07 01:41:10 -06:00
Aditya Atluri faaee7e9bf Increased size of arrays for single stream tests 2016-03-07 01:40:31 -06:00
Ben Sander aa03e1264c Enhance HIP trace debug functions.
- 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
2016-03-06 23:50:52 -06:00
Maneesh Gupta 39d5a2c079 Fix typo in nvcc_detail/hip_runtime_api.h 2016-03-07 09:40:15 +05:30
Aditya Atluri 75952029d6 added feature for hipHostGetFlags for CUDA and HIP 2016-03-06 12:17:30 -06:00
Aditya Atluri 1f3695b224 added flags support to be used later 2016-03-06 08:40:33 -06:00
Aditya Atluri d3ba2b9782 corrected hipDeviceGetProperties to hipGetDeviceProperties - not docs 2016-03-06 08:31:04 -06:00
Aditya Atluri 3aa764d5eb Added hipHostAlloc with hipHostAllocMapped flag 2016-03-05 15:57:56 -06:00
Aditya Atluri f479531be5 Added hipHostAlloc feature for CUDA 2016-03-05 13:58:56 -06:00
Aditya Atluri a5408ed7b6 v2 Added canHostMapMemory 2016-03-05 13:15:07 -06:00
Aditya Atluri 2ebbdd6ec5 Revert "Added canMapHostMemory feature"
This reverts commit af4edd277f.
2016-03-05 13:08:57 -06:00
Aditya Atluri af4edd277f Added canMapHostMemory feature 2016-03-05 13:06:37 -06:00
Aditya Atluri 4b271ec013 Added canMapHostMemory to hipDeviceProp 2016-03-05 19:30:29 -06:00
Ben Sander 6d4c830613 Merge branch 'privatestaging' of https://github.com/AMDComputeLibraries/HIP-privatestaging into privatestaging 2016-03-04 08:22:56 -06:00
Ben Sander ed71cded1c Add FORCE_SAMEDIR_DEP for copies. 2016-03-04 08:21:15 -06:00
Aditya Atluri bd23279271 v2: Fixed implementation of hipMemcpyDefault flag 2016-03-04 06:00:04 -06:00
Ben Sander 5e51ff177a Add define to disable STREAM thread-safety 2016-03-03 10:49:25 -06:00
Aditya Atluri c154e1f4e4 Added support for hipMemcpyDefault 2016-03-03 10:30:06 -06:00
Aditya Avinash Atluri 40eefc1cde Update hip_hcc.cpp 2016-03-03 13:59:43 -06:00