Commit Graph

13222 Commits

Author SHA1 Message Date
Ben Sander 3f810a203b Enabled USE_DISPATCH_HSA_KERNEL, with serialization in hipModuleUnload. 2016-12-06 10:14:01 -06:00
Rahul Garg afc81c955e Simple test case for hipDeviceGetPCIBusId
Change-Id: I0fb6d1bef4739a5b6280928f7b349d95c1656431
2016-12-06 17:31:54 +05:30
Rahul Garg a019f1ce39 Build Error correction in hipDeviceGetPCIBusId
Change-Id: I50ff4d95b7a732924c7a991cba60400b1c93c0de
2016-12-06 17:09:21 +05:30
Rahul Garg 0017419521 Changed hipDeviceGetPCIBusId to return Bus ID as string
Change-Id: I6d5aa7362084109d34bc015d948f8723b2a38ee9
2016-12-06 16:55:17 +05:30
foreman 22369c6db7 P4 to Git Change 1349829 by johtaylo@johtaylo-jtincrementor-increment on 2016/12/06 03:00:05
SWDEV-2 - Change OpenCL version number from 2296 to 2297.

Affected files ...

... //depot/stg/opencl/drivers/opencl/runtime/utils/versions.hpp#2043 edit
2016-12-06 03:21:38 -05:00
Maneesh Gupta b8413a69c8 Remove redundant variable g_malloc_heap_size
Change-Id: Idaf47be70488f0deb3eab05a86d9c5a413d3fff7
2016-12-06 10:19:03 +05:30
pensun 4cb1579d4a HIP resource leaks fix from Jack
Change-Id: I93f3ad7cb94ff1cba1577bd8acc90e826693d12e
2016-12-05 20:21:33 -06:00
foreman f42576b37c P4 to Git Change 1349651 by todli@todli-win-opencl-kv1 on 2016/12/05 18:05:57
SWDEV-107271 - [OpenCL][GFXIP9 Bring up] force Raven(gfx901) to use PAL backend

Affected files ...

... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpudevice.cpp#561 edit
2016-12-05 18:14:43 -05:00
Maneesh Gupta ac93376c26 Revert "Enable USE_DISPATCH_HSA_KERNEL."
This reverts commit f8bcbe8680.
2016-12-05 16:55:26 +05:30
foreman 897be96aed P4 to Git Change 1349269 by johtaylo@johtaylo-jtincrementor-increment on 2016/12/05 03:00:05
SWDEV-2 - Change OpenCL version number from 2295 to 2296.

Affected files ...

... //depot/stg/opencl/drivers/opencl/runtime/utils/versions.hpp#2042 edit
2016-12-05 03:09:52 -05:00
Maneesh Gupta ac368cc60f Don't share g_malloc_heap_size between libraries
Change-Id: Ic70bf83d4f865bc5c453941fdbc1814c77f0ad9d
2016-12-05 11:03:45 +05:30
foreman bcd923b89e P4 to Git Change 1349133 by johtaylo@johtaylo-jtincrementor-increment on 2016/12/04 03:00:08
SWDEV-2 - Change OpenCL version number from 2294 to 2295.

Affected files ...

... //depot/stg/opencl/drivers/opencl/runtime/utils/versions.hpp#2041 edit
2016-12-04 03:17:24 -05:00
Ben Sander f8bcbe8680 Enable USE_DISPATCH_HSA_KERNEL.
Optimize hipLaunchModule dispatch latency.
2016-12-04 00:13:19 -06:00
foreman ac1461ccc1 P4 to Git Change 1349078 by johtaylo@johtaylo-jtincrementor-increment on 2016/12/03 03:00:06
SWDEV-2 - Change OpenCL version number from 2293 to 2294.

Affected files ...

... //depot/stg/opencl/drivers/opencl/runtime/utils/versions.hpp#2040 edit
2016-12-03 03:13:14 -05:00
Ben Sander 783ac156ce Add additional controls for forcing serialization and blocking.
Move HIP_COHERENT_HOST_ALLOC so it is read once at init time.
Add HIP_LAUNCH_BLOCKING_KERNELS, HIP_API_BLOCKING.
Update docs on debug and chicken bits.

Conflicts:
	src/hip_hcc.cpp
2016-12-02 18:03:59 -06:00
foreman b705f74166 P4 to Git Change 1349031 by lmoriche@lmoriche_opencl_dev on 2016/12/02 18:33:20
SWDEV-107568 - [ROCm CQE][OCL][CZ] Basic 2.0 conformance test giving Segmentation fault (core dumped) at "progvar_prog_scope_uninit"
	- Detect if writable program scope variables are present in the program, and if so, insert barrier each dispatch of a kernel from this program.

Affected files ...

... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rockernel.hpp#11 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocprogram.cpp#48 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocprogram.hpp#16 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocvirtual.cpp#21 edit
2016-12-02 18:40:16 -05:00
foreman c7d2b188de P4 to Git Change 1348779 by nhaustov@spb-hlc-fiji1_opencl_rocm on 2016/12/02 06:23:08
SWDEV-102966 - Dump code object disassembly in OpenCL rocm device.

	Invoke DumpExecutableAsText from driver library.
	Update build to depend on some more LLVM libraries.

	LLVM changes are included, but will come through amd-common.
	Driver changes will come through ROCm-OpenCL-driver.

	Testing: Run some SDK samples/test_basic with AMD_OCL_BUILD_OPTIONS_APPEND=-save-temps

	Reviewed by: Laurent Morichetti, German Andryeyev.

Affected files ...

... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palprogram.cpp#27 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocprogram.cpp#47 edit
2016-12-02 06:41:36 -05:00
foreman 0fff20f59d P4 to Git Change 1348697 by johtaylo@johtaylo-jtincrementor-increment on 2016/12/02 03:00:05
SWDEV-2 - Change OpenCL version number from 2292 to 2293.

Affected files ...

... //depot/stg/opencl/drivers/opencl/runtime/utils/versions.hpp#2039 edit
2016-12-02 03:07:40 -05:00
foreman d4cbb3a801 P4 to Git Change 1348476 by lmoriche@lmoriche_opencl_dev on 2016/12/01 21:48:08
SWDEV-79278 - [OpenCL][GFXIP9 Bring up] Add the Vega10 DeviceInfo entry and the gfx90X control functions.

Affected files ...

... //depot/stg/opencl/drivers/opencl/library/build/Makefile.library#59 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocdefs.hpp#8 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocdevice.cpp#29 edit
2016-12-01 21:55:50 -05:00
foreman c1c625fc49 P4 to Git Change 1348453 by skudchad@skudchad_test_win_opencl2 on 2016/12/01 19:09:47
SWDEV-107855 - [DVR][CF] Slave GPU exit BACO when DVR is turned on
	- Enable partial initiazation when querying devices for Win7.

	ReviewBoardURL = http://ocltc.amd.com/reviews/r/11946/diff/

Affected files ...

... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gslbe/src/rt/GSLDevice.cpp#177 edit
2016-12-01 19:16:48 -05:00
foreman 33de5b1e09 P4 to Git Change 1348415 by gandryey@gera-w8 on 2016/12/01 17:11:27
SWDEV-98415 - [CQE OCL][OpenCL on PAL] Failures/Crash is observed with multiple modules of ocltst
	- Fix OCLImage2DFromBuffer test. Recalculate the original pitch if a view was created on the top of another ImageBuffer.

Affected files ...

... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palresource.cpp#19 edit
2016-12-01 17:17:58 -05:00
foreman 0b136f4765 P4 to Git Change 1348378 by gandryey@gera-ocl-lc on 2016/12/01 16:16:14
SWDEV-98415 - [CQE OCL][OpenCL on PAL] Failures/Crash is observed with multiple modules of ocltst
	- Fix program compilation on LC path. Add program constructor for the real device.

Affected files ...

... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palprogram.hpp#14 edit
2016-12-01 16:27:14 -05:00
foreman bf9217468e P4 to Git Change 1348313 by gandryey@gera-w8 on 2016/12/01 14:38:56
SWDEV-98415 - [CQE OCL][OpenCL on PAL] Failures/Crash is observed with multiple modules of ocltst
	- Fix a regression with oflline compilation. Mark null program.

Affected files ...

... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palprogram.cpp#26 edit
2016-12-01 14:54:04 -05:00
Maneesh Gupta fd8b1c7d41 Improve shared and static library support
hipcc accepts new parameter -use-staticlib and -use-sharedlib to
control linking behavior. Default is still static library.

Change-Id: I28fb9a939f8177c75abefd8b77d8118a6666d1f4
2016-12-01 15:33:12 +05:30
foreman e86db9a564 P4 to Git Change 1348037 by johtaylo@johtaylo-jtincrementor-increment on 2016/12/01 03:00:03
SWDEV-2 - Change OpenCL version number from 2291 to 2292.

Affected files ...

... //depot/stg/opencl/drivers/opencl/runtime/utils/versions.hpp#2038 edit
2016-12-01 03:25:34 -05:00
Maneesh Gupta f14198956f hipcc: HCC workaround no longer needs env var
Change-Id: I81f4eafddbda5e9e2f1082932dd502ab451cfc24
2016-12-01 12:51:58 +05:30
Maneesh Gupta 9cab8d4260 Fix some broken directed tests
Change-Id: I40f2661a74951f8d91824df8fd9ef0cc2312d183
2016-12-01 12:47:37 +05:30
foreman bc904d8335 P4 to Git Change 1347718 by kzhuravl@kzhuravl-fiji-oclhsa on 2016/11/30 13:40:59
SWDEV-76911 - Bring loader in sync with git (integration of CL 1347664) + necessary updates for runtimes

	ReviewBoardURL: http://ocltc.amd.com/reviews/r/11425/
	Testing: Dev/SCDevUtil build, smoke, some opencl sc stg builds/tests, precheckin (http://ocltc:8111/viewModification.html?modId=79218&personal=true&init=1&tab=vcsModificationBuilds)

Affected files ...

... //depot/stg/opencl/drivers/opencl/compiler/sc/HSAIL/ext/amdhsacod/amdhsacod.cpp#17 integrate
... //depot/stg/opencl/drivers/opencl/compiler/sc/HSAIL/ext/loader/executable.cpp#33 integrate
... //depot/stg/opencl/drivers/opencl/compiler/sc/HSAIL/ext/loader/executable.hpp#18 integrate
... //depot/stg/opencl/drivers/opencl/compiler/sc/HSAIL/ext/loader/loaders.hpp#12 integrate
... //depot/stg/opencl/drivers/opencl/compiler/sc/HSAIL/include/amd_elf_image.hpp#9 integrate
... //depot/stg/opencl/drivers/opencl/compiler/sc/HSAIL/include/amd_hsa_loader.hpp#15 integrate
... //depot/stg/opencl/drivers/opencl/compiler/sc/HSAIL/include/hsa.h#5 integrate
... //depot/stg/opencl/drivers/opencl/compiler/sc/HSAIL/include/hsa_ext_image.h#5 integrate
... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpuprogram.cpp#229 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palprogram.cpp#25 edit
2016-11-30 13:47:21 -05:00
foreman 71f5fb7e8f P4 to Git Change 1347598 by atamazov@ata_h1 on 2016/11/30 06:48:06
SWDEV-79309 - [ROCm][OCLonLC][AMDGPU asm] OclElf: Correctly skip symbols which have special section indexes.

	[Issues]
	Kernels which contain absolute symbols can't be run. Error message looks like:
	"OclElf::getSymbolInfo() failed in gelf_getshdr() - Invalid argument."

	[Reason]
	Symbols which have special section indexes (e.g. absolute symbols)
	are incorrectly handled by the compiler library.

	[Solution]
	Correctly skip symbols which have special section indexes in the loader.
	Promote the fix into compiler/legacy-lib at once.

	[Remarks/Misc]
	Absolute symbols do not normally produced by OpenCL compiler,
	but may come from kernels written in assembly language.
	AMDGPU asm exports all symbols, even ABSOLUTE LOCAL HIDDEN INTERNAL etc.
	Impossibility to disable export of useless symbols is known issue of llvm-mc
	core, see https://llvm.org/bugs/show_bug.cgi?id=27201.
	Note that presence of useless symbols does not make ELF invalid.

	[Review] http://ocltc.amd.com/reviews/r/11921/
	Evgeny Mankov (OK)

	[Tests]
	Tested that issue with asm kernels has gone.
	PASS TeamCity - Pre-checkin for CL1347161 timestamped:2016-11-29 21:54:40 Rev:0063

Affected files ...

... //depot/stg/opencl/drivers/opencl/compiler/legacy-lib/loaders/elf/elf.cpp#3 edit
... //depot/stg/opencl/drivers/opencl/compiler/lib/loaders/elf/elf.cpp#36 edit
2016-11-30 06:53:56 -05:00
foreman 61a64a192b P4 to Git Change 1347509 by johtaylo@johtaylo-jtincrementor-increment on 2016/11/30 03:00:03
SWDEV-2 - Change OpenCL version number from 2290 to 2291.

Affected files ...

... //depot/stg/opencl/drivers/opencl/runtime/utils/versions.hpp#2037 edit
2016-11-30 02:12:59 -06:00
Aditya Atluri e6df8cb28c added half add and fma intrinsic
Change-Id: Ifa60c1a7065f524f069291bb00d987b11c836cc4
2016-11-29 19:46:01 -06:00
pensun 504fcaf786 Change to use produce device name by default
Change-Id: Ie2cee2a2e94a08b5874a2f5abee5d1ab6c9fdf47
2016-11-29 11:34:06 -06:00
Rahul Garg bda0704213 Added support for hipMemGetAddressRange
Change-Id: I99a796a4eb765152cf15a12d6a86b58684d34f50
2016-11-29 22:04:09 +05:30
foreman e46fadfb28 P4 to Git Change 1346875 by johtaylo@johtaylo-jtincrementor-increment on 2016/11/29 03:00:05
SWDEV-2 - Change OpenCL version number from 2289 to 2290.

Affected files ...

... //depot/stg/opencl/drivers/opencl/runtime/utils/versions.hpp#2036 edit
2016-11-29 03:12:57 -05:00
pensun 8e2980c7ef Change the parameter type of hipDeviceGetPCIBusID to char*
Change-Id: Ia72f403126e95f65da53208fc246f45d1417381f
2016-11-28 10:47:18 -06:00
foreman 68e3bfc72d P4 to Git Change 1346404 by johtaylo@johtaylo-jtincrementor-increment on 2016/11/28 03:00:04
SWDEV-2 - Change OpenCL version number from 2288 to 2289.

Affected files ...

... //depot/stg/opencl/drivers/opencl/runtime/utils/versions.hpp#2035 edit
2016-11-28 03:09:32 -05:00
Maneesh Gupta 8a464ac850 Merge branch 'amd-develop' into amd-master
Change-Id: Ie5aa8e70607936f63cf4c763298140e83a375a68
2016-11-28 09:18:26 +05:30
foreman 8195a9043b P4 to Git Change 1346311 by johtaylo@johtaylo-jtincrementor-increment on 2016/11/27 03:00:04
SWDEV-2 - Change OpenCL version number from 2287 to 2288.

Affected files ...

... //depot/stg/opencl/drivers/opencl/runtime/utils/versions.hpp#2034 edit
2016-11-27 03:10:33 -05:00
Ben Sander a504df955e Add more debug info 2016-11-26 08:56:02 -06:00
foreman 96bed3cbf4 P4 to Git Change 1346279 by johtaylo@johtaylo-jtincrementor-increment on 2016/11/26 03:00:04
SWDEV-2 - Change OpenCL version number from 2286 to 2287.

Affected files ...

... //depot/stg/opencl/drivers/opencl/runtime/utils/versions.hpp#2033 edit
2016-11-26 03:06:23 -05:00
foreman 8c56088f43 P4 to Git Change 1346190 by gandryey@gera-w8 on 2016/11/25 15:12:02
SWDEV-86035 - Add PAL backend to OpenCL
	- Correct a fix from CL#1346120. Just remove the asserts, since size is used in the abstraction  layer setup and represents cl_mem object.

Affected files ...

... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palkernel.cpp#27 edit
2016-11-25 15:21:59 -05:00
Ben Sander a990806b6c Tweak profiler doc
Change-Id: I7be7c44467510e38ae850e1e0a14209a3b4380f1
2016-11-25 14:06:59 -06:00
foreman 41eceed8f1 P4 to Git Change 1346120 by gandryey@gera-w8 on 2016/11/25 13:00:01
SWDEV-86035 - Add PAL backend to OpenCL
	- Fix a regression after LC path integration. sampler, image and queue pointers have uint64_t size always.

Affected files ...

... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palkernel.cpp#26 edit
2016-11-25 13:04:36 -05:00
foreman 38be7dfff4 P4 to Git Change 1345996 by johtaylo@johtaylo-jtincrementor-increment on 2016/11/25 03:00:04
SWDEV-2 - Change OpenCL version number from 2285 to 2286.

Affected files ...

... //depot/stg/opencl/drivers/opencl/runtime/utils/versions.hpp#2032 edit
2016-11-25 03:20:16 -05:00
foreman 5036b5da85 P4 to Git Change 1345883 by gandryey@gera-w8 on 2016/11/24 17:31:06
SWDEV-102801 - OpenCL on PAL - mGPU SVM support
	- Update SVM path to use existent cache coherency logic for MGPU

Affected files ...

... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_memobj.cpp#80 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpudevice.cpp#560 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpumemory.cpp#130 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpuvirtual.cpp#409 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/pal/paldevice.cpp#38 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palkernel.cpp#25 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palmemory.cpp#8 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palvirtual.cpp#39 edit
... //depot/stg/opencl/drivers/opencl/runtime/platform/memory.cpp#123 edit
2016-11-24 17:43:45 -05:00
foreman afeda63f68 P4 to Git Change 1345883 by gandryey@gera-w8 on 2016/11/24 17:31:06
SWDEV-102801 - OpenCL on PAL - mGPU SVM support
	- Update SVM path to use existent cache coherency logic for MGPU

Affected files ...

... //depot/stg/opencl/drivers/opencl/api/opencl/amdocl/cl_memobj.cpp#80 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpudevice.cpp#560 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpumemory.cpp#130 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpuvirtual.cpp#409 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/pal/paldevice.cpp#38 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palkernel.cpp#25 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palmemory.cpp#8 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palvirtual.cpp#39 edit
... //depot/stg/opencl/drivers/opencl/runtime/platform/memory.cpp#123 edit
2016-11-24 17:43:45 -05:00
foreman e4dc501e91 P4 to Git Change 1345575 by johtaylo@johtaylo-jtincrementor-increment on 2016/11/24 03:00:04
SWDEV-2 - Change OpenCL version number from 2284 to 2285.

Affected files ...

... //depot/stg/opencl/drivers/opencl/runtime/utils/versions.hpp#2031 edit
2016-11-24 03:06:41 -05:00
Aditya Atluri de89b25d52 added support for rcp for float and double
Change-Id: Ibeba3a9f64494fc0a176bcb4a854fb2f56567b55
2016-11-23 20:01:18 -06:00
Aditya Atluri cc1f8a1011 added fma for double and float
1. Added fma intrinsic support for double and float
2. Added test for fma

Change-Id: I909fdbec34a3d12c03ba6eff3a39376a7128ee43
2016-11-23 18:22:05 -06:00
foreman 45bf0e3ba6 P4 to Git Change 1345399 by cpaquot@hog-ocl on 2016/11/23 17:13:30
SWDEV-107558 - Wrong path for map/unmap images
	If mapMemory is allocated, we shouldn't use the read/writeImage path since they lock/unlock the host pointer which in this case, is mapped to GPU memory already.

	ReviewBoardURL = http://ocltc.amd.com/reviews/r/11891/diff/

Affected files ...

... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocvirtual.cpp#20 edit
2016-11-23 17:20:51 -05:00