Commit Graph

1191 Commits

Author SHA1 Message Date
Aditya Atluri abf6872b2b fixed multi-dim module kernel launch
Change-Id: Id1d81f2375d058979ab526433f905cf0ea3d23d6
2016-11-11 12:25:23 -06:00
Ben Sander 1e5515ee9f Add option to deny peer access.
Also fix test.

Change-Id: I1b247f6c4271442b008e560669bca4daf8eb94c7
2016-11-10 23:12:48 -06:00
Ben Sander 65584e48de Use forceUnpinnedCopy to resolve P2p corner cases.
Change-Id: I2aebb419881246cebb696bec87798635bc71acc2
2016-11-10 23:12:48 -06:00
Ben Sander d3d6feb4de Enable async copy again.
Also add HIP_FORCE_SYNC_COPY chicken bit.

Change-Id: I76a385410494b99bf27305d3c08f55dd81987565
2016-11-10 23:12:48 -06:00
Ben Sander 8724273f28 Doc change only - add comments to test.
Change-Id: Ie42087cf3c78e49337b18bb71f3f0e1e7950ee1b
2016-11-10 23:12:48 -06:00
Ben Sander ced9d72d94 Refactor copy and P2P logic.
Prefer use of source-engine for DMA copies, even if user submits copy
in a stream attached to a different device.
The stream is now used only for synchronization, and HIP
makes the most optimal decision for which engine to perform the
copy - typically the source copy engine.

HIP now makes decision on which engine should perform the copy
and passes this to HCC using new apis.
HIP has additional information about peer
visibility and will make a decision which agent should perform
the copy .

Change-Id: I0cf4cfebeae256e6ca795f08a7ed7130f4857d1f
2016-11-10 23:12:48 -06:00
Ben Sander 2dea3a0b1a Improve memory debug
Change-Id: I0f033139aa4e4b47039eb016e404009127bd0a44
2016-11-10 23:12:48 -06:00
pensun 1ec5761a11 Update depreciated information for threadfence_system()
Change-Id: Id13d2f81edb51eb42b896a5c06913d59ec907c55
2016-11-10 11:55:12 -06:00
Maneesh Gupta a12d5a8989 CMakeLists.txt: Cascade CMAKE_BUILD_TYPE to tests
Change-Id: I53a3ea951c1fd57e43a02381a457c1dedc1a34f7
2016-11-10 21:26:34 +05:30
Rahul Garg fcb94863f7 hipDeviceGetByPCIBusId support for HIP/NVCC
Change-Id: I8f82890e88d2a15f592bff192179e7d5c5362722
2016-11-10 11:40:59 +05:30
Maneesh Gupta 669d734624 hipcc: Default to HIP_LIB_TYPE=1
Change-Id: I83b05accd76f7bc94bd724c66ae060fa0095bc8d
2016-11-10 11:34:00 +05:30
Maneesh Gupta 36024deb3a hcc_dialects/Makefile: use clamp-config
Change-Id: I86df82f75b75125825e22d0545209a19386d9936
2016-11-10 11:31:50 +05:30
pensun 4d7ac1e091 resolve conflicts for git pull
Change-Id: Ie353b831e2241bc28042069b6cc7405257e871e1
2016-11-09 21:38:43 -06:00
pensun 24c621db5b Add documentation on threadfence_system workaround guidelines.
Change-Id: I9636a3808798f3dabe992285ce5652187cee6eb8
2016-11-09 21:36:30 -06:00
pensun f7e9f12bf1 Add option to alloc fingrained system memory
Change-Id: Ia13c8e058cb988b5857e75a590a4d67411362ae1
2016-11-09 21:36:30 -06:00
Maneesh Gupta 72c722c3d6 Merge branch 'rocm-rel-1.3' into amd-develop
Conflicts:
	include/hip/nvcc_detail/hip_runtime_api.h

Change-Id: I990a7d008da9e8dcc68250cebbc8ee6e723c7e01
2016-11-10 08:56:38 +05:30
pensun 57cd3c8244 fix hipProfiler* apis on NV path
Change-Id: I6adca6151fef3a9b35348163eb6bd13f5c414172
2016-11-09 15:44:01 -06:00
pensun 76c3c20da6 fix for hipcallback function on NV path
Change-Id: If80c0cfe60b1f3b1a71627b5f3f79503cba4d491
2016-11-09 11:33:23 -06:00
Maneesh Gupta 7c68768a0d Update release notes for 1.0 release
Change-Id: I74fa2b41afc334a76c309b125c27aa141cd59554
2016-11-08 16:31:56 +05:30
Ben Sander 7106dc6796 Fix tests to read warpSize from device props.
Change-Id: I9583577793afad49f9eb1ee9069bd4c6963a6023
2016-11-06 04:26:28 -06:00
Ben Sander c59fe09a81 Update gitignore for some common output files
Change-Id: I9cd60f042af4dba07fe0fdbd2ee442936ff8c7bd
2016-11-06 04:26:15 -06:00
Ben Sander d728819d17 Improve Peer support and testing.
Change-Id: Icadc65988aaf145a265587ab0357c5bf4d26f3eb
2016-11-06 03:22:36 -06:00
Ben Sander 092b3dacda Set forceHostCopyEngine for other copy dirs. Support HIP_FORCE_P2P_HOST
Also: more debug for copy and P2p.

Change-Id: I87030c525410e041b2a00baaf6c68e6c0977ff42
2016-11-04 19:53:23 -05:00
Ben Sander ff3298f40e Expand hipP2PSimple testing.
Cover cases where P2P is used for H2D copies, where host is pinned
but not accessible to the copy agent.

Change-Id: I9464b787228b40f93473708c3fde9726e1986365
2016-11-04 16:13:32 -05:00
Ben Sander 5d79384832 Refactor resolve-mem step1
Change-Id: I7b8b2bbb56d7b31a97b48ebd42002641cd07a460
2016-11-04 09:37:56 -05:00
Ben Sander 3f0a2b8dc1 Add debug for Peer APIs. Enable PeerMemcpy APIs by default.
Change-Id: I46e39a9e7b07686a78484c1f3b5495b08e052fbb
2016-11-04 08:51:16 -05:00
Ben Sander 5832349e5e Print non-peers too
Change-Id: I2a6905edcdf144aa732ae3120c17780477f232ac
2016-11-04 06:34:07 -05:00
Ben Sander f0bb817c11 Pre-pend HIP_PATH/lib to linker, so we find developer object code
Previously might pick up libs from /opt/rocm/lib.

Change-Id: Ia7adb345defe433d5952aa61706fe03fd7cbcd35
2016-11-04 06:06:04 -05:00
pensun 774de273d0 Update document for workaround suggestion on threadfence_system()
Change-Id: Icccab8270604a0e578a8614b9afb3f95372f4966
2016-11-02 16:08:27 -05:00
pensun 4817131cdc Update hipStreamNonBlocking to use cuda define on NV path
Change-Id: I74ea09db99d602ba1c5f192b36ff7f2781176e6a
2016-11-01 20:30:56 -05:00
Aditya Atluri f48c53534e added inter thread data movement intrinsics
Change-Id: I2a8a8ed49429cb7f96439bd28c4b83b5142737df
2016-11-01 16:37:33 -05:00
Rahul Garg 2d15d0741c Added hipDeviceGetByPCIBusId in hip/hcc path
Change-Id: I3cca0dc533d0281689d8a407c7da16ca1ba6a3a8
2016-11-01 10:57:48 +05:30
Evgeny Mankov 1b64c167f6 [HIPIFY] wrap kernel name with HIP_KERNEL_NAME macros...
only in case of commas in it.
2016-10-28 20:05:51 +03:00
Evgeny Mankov b0f37d0d2e * [HIPIFY] Initial Profiler support.
CUDA Driver API porting to HIP:

+ cuProfilerStart, cuProfilerStop.
- cuProfilerInitialize & cudaProfilerInitialize - unsupported yet by HIP.
2016-10-28 18:32:13 +03:00
Ben Sander 87a2e8f12b add hip_profile.h
Change-Id: Id43a4336db53567020584cb7842baf5c1649fd8e
2016-10-28 07:08:46 -05:00
Maneesh Gupta c39ddad215 hipdemangleatp: Try handling HC kernels as well
Change-Id: Ie438ddd28e5bc6067fcd682df849d3183046b40a
2016-10-28 15:46:59 +05:30
Maneesh Gupta c5a2ad4964 CMakeLists.txt: Update include paths needed for Fedora support
Change-Id: Ib84f9dba30d2c64f344d6f8e85ddbe15f30af1a0
2016-10-28 14:12:53 +05:30
Maneesh Gupta 6fa9bc3b61 hipcc: Update flags for Fedora support
Change-Id: I90be7768410e491b4f11c3b0f08470246d781d80
2016-10-28 14:12:13 +05:30
Ben Sander 024d9ab090 Print short hipLaunchKernel correctly.
Change-Id: I6ca03d7c707cd03d6982199830213953d5855f17
2016-10-27 23:09:32 -05:00
Ben Sander bb58f4f6fc Add initial hipProfileStart/Stop
And modify sample to show how to use.
Still needs some work to understand interaction with CXL.

Change-Id: I2579824d2dd7863ea23874d34f0dabb3cb305d3e
2016-10-27 23:09:32 -05:00
Ben Sander 32b086400e show how to use variety of HIP_PROFILE features
Change-Id: I6edd66ac4c068b64e1dc3787d7f1f69ab3238469
2016-10-27 23:09:32 -05:00
Ben Sander ef8eac9b66 Add two levels of HIP_PROFILE_API (1=short,2=long)
Change-Id: I7ef98589f8731fb879db109fd573c62b489f2b61
2016-10-27 23:09:31 -05:00
Ben Sander e5ef8a2fd7 HIPCC adds paths to profiler include and libs by default.
Users who desire otherwise can set HIP_ATP_MARKER=0.
Also remove old unused hipcc_explicit_lib option.

Change-Id: I2bf07ba880329e7a3b1365dd33a3b2be6794370f
2016-10-27 23:09:31 -05:00
Ben Sander f95482c7c5 Add new hipdemangleatp and snapshot sample update for new functionality
Change-Id: Ie19c683b2b0bdfeb0c3fcf89444c2e21b7f606e7
2016-10-27 23:09:31 -05:00
Ben Sander ab1836544a Fix scoped marker so begin/end ATP timestamps correct
Change-Id: Ic944d3fc00d7bc31b756c0e6c327b99eb489537e
2016-10-27 23:09:31 -05:00
Ben Sander e9056798f6 Rename HIP_ATP_MARKER and profiling vars
HIP_PROFILE_API
HIP_DB_START_API
HIP_DB_STOP_API

Change-Id: I6c4da67212ff8217e6356a2622d4c6278a188c34
2016-10-27 23:09:31 -05:00
Ben Sander f5e8090f2f Allow HIP_DB to be number or string flags (ie HIP_DB=api+mem+sync)
Add callbacks for processing env vars.

Change-Id: I4ddf50e2da56b1dae43f50657bc693b07b23c03d
2016-10-27 23:09:31 -05:00
Ben Sander 710be682ca Add HIP_PROFILE_START_API, HIP_PROFILE_STOP_API
Refactor HIP_INIT_API to call recordApiTrace.

Change-Id: Ieff4b5018236f59e49e1b9841474440a34f821df
2016-10-27 23:09:31 -05:00
Ben Sander 739bc37503 Add per-thread API seqnum to debug
Change-Id: Ib13733a3e84cd56bae13a32bae40f936c20b7543
2016-10-27 23:09:31 -05:00
Ben Sander 354091f357 Don't call allow-access if allocating device's only peer is self.
Change-Id: Iac58e6c3e460675833f10b1e8b2e393de223654d
2016-10-27 23:09:31 -05:00