Revīziju grafs

26 Revīzijas

Autors SHA1 Ziņojums Datums
Ben Sander c441d5ec29 Step1 in staging buffer copy.
- use StagingBuffer class for copies.
- refactor g_device to use array rather than vector.
   (keeps pointers from moving).


[ROCm/hip commit: 24c1fdb864]
2016-02-12 18:24:08 -06:00
Ben Sander b9dc0e9497 Query tracked memory sizes.
Support more accurate hipMemGetInfo.  Add test to hipPointerAttrib.


[ROCm/hip commit: d7396b5af3]
2016-02-12 18:24:08 -06:00
Ben Sander 680b600b4a Tracker improvements
- add API to add / remove user-pointers from the tracker.
- test for thread-safety with MultiThreadtest_2 - rapid
  insertions/removal.
- add mutex to provide thread-safety.
- rename tracker interface to "memtracker_..." for consistency.
- add am_memtracker_reset, connect to hipDeviceReset.
-


[ROCm/hip commit: de45e2291e]
2016-02-12 18:24:08 -06:00
Ben Sander d4a90f8afd Create address tracker for am_alloc.
Tracks device where memory is allocated, pinned-host or device, and
more.

Uses memory-range-based lookups - so pointers that exist anywhere in

the range of hostPtr + size will find the associated AmPointerInfo.

The insertions and lookups use a self-balancing binary tree and
should support O(logN) lookup speed.


[ROCm/hip commit: 4ee2a5229b]
2016-02-12 18:24:08 -06:00
Evgeny Mankov fcd154097f Fix typo: maxThreadsPerMultiProcessor -> MaxSharedMemoryPerMultiprocessor
Device property MaxSharedMemoryPerMultiprocessor set equal to totalGlobalMem (HIP path).
Reason: MaxSharedMemoryPerMultiprocessor should be as the same as group memory size. Group memory will not be paged out, so, the physical memory size = total shared memory size = group region size. NVCC path remains untouched: CUDA's device property MaxSharedMemoryPerMultiprocessor is reported.

hipify is updated as well.


[ROCm/hip commit: ea8f99702d]
2016-02-12 01:29:20 +03:00
Evgeny Mankov 4eade0ce83 BDFID (BusID/DeviceID/FunctionID) support.
Except FunctionID (or DomainID in CUDA) support, because cudaDeviceProp::pciDomainID is not reported by CUDA.


[ROCm/hip commit: 33f60c300d]
2016-02-11 22:26:01 +03:00
Evgeny Mankov 3a032ff317 Formatting, no functional changes
[ROCm/hip commit: 254da4ec53]
2016-02-10 17:21:18 +03:00
gargrahul 1ab2294657 Removed atomicInc and atomicDec support from HIP
[ROCm/hip commit: 8c40a4ace4]
2016-02-10 04:29:55 +05:30
Evgeny Mankov c38a69ef33 Device property concurrentKernels is added to hipDeviceProp_t struct.
For HCC path concurrentKernels is set to true since all ROCR hardware supports this feature.
For NVCC path concurrentKernels is obtained from CUDA's device property cudaDeviceProp::concurrentKernels.


[ROCm/hip commit: 950c3baacd]
2016-02-09 17:10:35 +03:00
Maneesh Gupta 77f61d1a46 Move HIP_DEVICE_COMPILE defines to hip_common.h
[ROCm/hip commit: 3291e0ec96]
2016-02-09 10:57:20 +05:30
Ben Sander a2dac9e12c minor doc touchup
[ROCm/hip commit: 9e2c3c8df3]
2016-02-08 22:11:11 -06:00
Ben Sander 9cb14a455c Fix getdeviceattr compilation for NVCC
[ROCm/hip commit: 76ebe6dcfd]
2016-02-04 16:26:33 -06:00
Sam Kolton 2306293526 Implementation of hipDeviceGetAttribute()
[ROCm/hip commit: 0a27507208]
2016-02-04 17:39:27 +03:00
Peng Sun 503ec9ad24 Fix all TODO-doc
[ROCm/hip commit: c73996d041]
2016-02-02 21:29:09 -06:00
Peng Sun 03630ee0a4 Finish all TODO for error code
[ROCm/hip commit: 8b74333204]
2016-02-02 17:39:46 -06:00
scchan 39fb16bc5f add inline attribute to shfl functions
[ROCm/hip commit: 265c42500f]
2016-02-02 12:53:17 -06:00
streamhsa af8cc35552 Adjusted the value of __any as per CUDA -sandeep
[ROCm/hip commit: 974d491902]
2016-02-02 15:25:42 +05:30
streamhsa e4635c36a0 ADDED Support for __ffs() and __ffsll() having signed input -sandeep
[ROCm/hip commit: 23904df99b]
2016-02-02 15:05:46 +05:30
scchan ca142c6d9c adding shfl, shfl_up, shfl_down, shfl_xor intrinsics
[ROCm/hip commit: 04f3e3e598]
2016-02-01 23:55:31 -06:00
Maneesh Gupta 9ed3ef50fe Add double and integer intrinsics to test
[ROCm/hip commit: 861cba6f75]
2016-02-01 16:00:45 +05:30
Maneesh Gupta 1b4ad3eedf Add few more single precision intrinsics to hcc_detail/hip_runtime.h
[ROCm/hip commit: d2c6125a7c]
2016-02-01 14:29:50 +05:30
Maneesh Gupta 01c51ce734 Restrict using namespace hc::precise_math to device only
[ROCm/hip commit: 3b19fd578d]
2016-02-01 14:26:50 +05:30
Maneesh Gupta 3e13c7dae7 Remove redundant #define __HCC__ in hcc_detail/hip_runtime.h
[ROCm/hip commit: e55f3778e0]
2016-02-01 14:24:41 +05:30
sunway513 1b93c2f456 Fix some typos and incorrect namings in comments
[ROCm/hip commit: 02fa107967]
2016-01-28 13:17:44 -06:00
sunway513 11edc27b87 Fix @file and @brief tag on header files
[ROCm/hip commit: 71a841d764]
2016-01-28 10:59:21 -06:00
Ben Sander 97cf8b8ac9 Initial commit for GPUOpen Launch
[ROCm/hip commit: f38e63ff18]
2016-01-26 20:14:33 -06:00