Commit Graph

13359 Commits

Author SHA1 Message Date
Jatin Chaudhary 1fdbf35d14 SWDEV-508617 - There is no NaN for E4M3 and FNUZ
Change-Id: I330b041019990231c098073f94d9d40a3c13ba76
2025-01-30 11:48:34 -05:00
Saleel Kudchadker 2d450e8b06 SWDEV-504494 - Resolve signal dependencies
- Resolve signal dependencies for barrier value packet if there are > 1
  depenent signals. Barrier Value packet accounts for only 1 dep signal
- Better log

Change-Id: Ia506ad5d80b91d598f92e7b539f41756e9b4b64b
2025-01-29 19:49:02 +00:00
Jatin Chaudhary e6fb89190a SWDEV-507817 - fix the return type of one of the atomicMin variants
Change-Id: I9915eb174d5677e21adbabae5819c9e306338ab3
2025-01-29 11:52:19 -05:00
Jimbo Xie 4abedf2a0e SWDEV-510869 - add gfx1153 id
Change-Id: I36d39a1db2392990ad9b01d70676c3c986435707
2025-01-28 18:15:46 -05:00
Saleel Kudchadker 08af3eb484 SWDEV-508225 - Improve fat binary handling
Change-Id: I78a9951f2f4c4c743c1205b1e40aac215054e27d
2025-01-28 14:38:21 -05:00
German Andryeyev ea0b092af8 SWDEV-459826 - Add a crash dump for a failed queue
The logic can analyze the AQL queue state and
find a failed AQL packet with the kernel's name

Change-Id: I1a478fa2c25462cd07a194784958bdf22454b897
2025-01-28 14:27:46 -05:00
Tao Sang f2ff56af9c SWDEV-458943 - Add fast path in wait()
wait() is redesigned with two pathes:
fast path: Use spinlock to wait for notify signal. If the
 signal hasn't been received for some loops, go to slow path.
slow path: Use condition_variable's wait().

Improve monitor wrapper for better performance.

Fix some bugs left from name removing patch.

Change-Id: I893a8353121a25d11e37c8e631caf31cc1fc1f24
2025-01-28 12:19:55 -05:00
Saleel Kudchadker d208e8052f SWDEV-504494 - Set active engine for SDMA
Change-Id: I4cec84e71903c5813a7063e8b9ff1ea4473f4720
2025-01-27 17:54:36 -05:00
Gerardo Hernandez b073063612 SWDEV-510589 - Use libgcc1 package (on Debian 10 only)
Change-Id: Ibe945e366468a84fd717e0e425cfaf7dab5a99c4
2025-01-27 11:02:30 -05:00
Marko Arandjelovic 269ec54252 SWDEV-489619 - Added checks for memcpy capture path
Change-Id: I0e156099282f0b6393bcbcee2e9b96c31034a851
2025-01-27 03:51:34 -05:00
Jacob Lambert 1fc7c6bb9a SWDEV-360440 - Prepare CLR CMake for Comgr V3 transition
Change-Id: Ia279928fd3549a45bae561d0d2d8fcf110d8c245
2025-01-27 01:09:23 -05:00
Ioannis Assiouras 21c223f8df SWDEV-510319 - Fixed random segfaults in graph tests
This change fixes random segfaults in graph tests that
are seen after the change make internal callbacks non-blocking.
The callback thread that decreases the GraphExec ref count
may now run after the runtime shutdown. This can cause a segfault
because the hip::device that is accessed in GraphExec destructor
is already destroyed during runtime shutdown. This patch ensures
that the hip::device object  stays alive until after the
callback thread completes.

Change-Id: I75a6ac01f27a0b2250bbd10ed389ebfb322927af
2025-01-25 09:54:15 -05:00
Sourabh Betigeri c460b0541b SWDEV-502219 - Adds validity checks for negative parameters passed
Change-Id: Ib8a531533306a27143d74b81c074de81051eb896
2025-01-24 16:32:29 -05:00
Saleel Kudchadker 9b7e0ad48a SWDEV-510186 - Improve logging of kernel names
- Demangle kernel names in logs

Change-Id: I9aa58e8c109becb45ef7fc747d991bd657c4190a
2025-01-24 11:43:02 -05:00
zichguan-amd 272ef9a7bf SWDEV-509518 - Allow LLVM_ROOT and Clang_ROOT to be used with find_program
Fixes #123. find_program doesn't follow CMP0074 and thus ignores LLVM_ROOT and Clang_ROOT. This change adds LLVM_ROOT and Clang_ROOT to the search path of find_program for llvm-mc and clang in hiprtc to mimics previous add_package behaviour.
Caveat: cmake-specific variables like CMAKE_PREFIX_PATH will take precedence over paths specified with HINTS for find_program, there's no way to change the ordering unless we skip cmake-specific variables all together using NO_CMAKE_PATH and NO_CMAKE_ENVIRONMENT_PATH.

Change-Id: I1fedb60cda09744416e19b3c6e3e0c5c9045f8e7
2025-01-23 11:50:36 -05:00
taosang2 799e54aa0d SWDEV-507969 - Fix wrong VGPRs for some devices
Change-Id: Ia8fc19564272e2c7171d991376bf896a99085a97
2025-01-22 10:11:47 -05:00
Jaydeep Patel 57df1b348f SWDEV-508982 - [6.4 Preview] - Handle hipMemPoolCreate, hipMemPoolDestory & hipDeviceSetMemPool during stream capture.
Change-Id: Ia195442041803896df814798c3d2053c0ba7770c
2025-01-22 05:28:47 -05:00
Jatin Chaudhary bd7d40a4d8 SWDEV-491248 - Fix build_mask
thread_rank() gives thread index in a block. Limit the range to the
current warp size.

Change-Id: Ib5c9831236096485cf99ba7ab0b911a3b10de31c
2025-01-22 04:46:01 -05:00
Jaydeep Patel b4df9fb6ec SWDEV-457316 - Use phy memory obj stored in user data instead of querying from memObjs.
Change-Id: Id837eb00195d88b50904441f01cf8153fa752ecd
2025-01-21 22:05:14 -05:00
Pengda Xie f76733e5b8 SWDEV-508590 - Fix segfault issue with hipModuleLoad
- Ensuring devProgram pointer isn't nullptr

Change-Id: Ia5786d0a2441f3a512d79b4998eb314beb98b35e
2025-01-21 11:38:37 -05:00
Sourabh Betigeri e4ba0b6262 SWDEV-507104 - Removes alignment requirement for Semaphore class to resolve runtime misaligned memory issues
Change-Id: I1be3eb6e9fdcf12e995c8fe8ee30592c94f7f97a
2025-01-20 11:27:47 -05:00
Konstantin Zhuravlyov dd4378611a SWDEV-341212 - HIP header changes for supporting SPIR-V
This removes almost all uses of the deprecated
__AMDGCN_WAVEFRONT_SIZE macro, which is unavailable
when targeting SPIR-V, and adds a SPIR-V compatible
formulation of warpSize (which should end up as the
sole definition of warpSize once we remove support
for treating it as a compile time constant). It
is incomplete in that the cooperative_groups
implementation will need additional surgery.

Squashed commit of the following:

commit 6840826c3fec8516857dc4f2092d84358550f588
Author: Alex Voicu <alexandru.voicu@amd.com>
Date:   Fri Dec 6 23:36:32 2024 +0000

    Add deprecation warning for constexpr uses of `warpSize`.

commit a72307a7353034c2de53fd164e016967945fd0d1
Author: Alex Voicu <alexandru.voicu@amd.com>
Date:   Fri Dec 6 23:12:14 2024 +0000

    Prepare HIP RT for SPIR-V.

commit 5e40dd746ac4f8c93b521ef048ff9d494905ba95
Author: Alex Voicu <alexandru.voicu@amd.com>
Date:   Fri Dec 6 22:46:05 2024 +0000

    Revert stale change.

commit 231fe91c53dba4cabd832fc84eaa6ddb402271a0
Merge: a48905ec9 78f62d323
Author: Alex Voicu <alexandru.voicu@amd.com>
Date:   Fri Dec 6 22:37:24 2024 +0000

    Merge branch 'amd-staging' of https://github.com/ROCm/clr into amd-staging

commit a48905ec9cfe0e017cc64943195be82b530117d7
Author: Alex Voicu <alexandru.voicu@amd.com>
Date:   Tue Sep 17 03:14:56 2024 +0100

    Add scaffolding for SPIR-V support.

Change-Id: I2e84bbe90df58a5f9a8709b619905f04fa5b96dc
2025-01-20 08:42:24 -05:00
Jatin Chaudhary 6a5d19059d SWDEV-341217 - Initial work to use SPIRV in HIP
Change-Id: If5c09b5e86b498e7ac5eb05adf28cb7a1fac8101
2025-01-20 03:54:23 -05:00
Jaydeep Patel 1aa8383b09 SWDEV-509664 - Specify type explicitly.
Change-Id: Ia0c0478682fa15eae7a31a2360310f08151716d4
2025-01-17 13:09:48 -05:00
Julia Jiang fe5c68d8a3 SWDEV-509295 - Update changelog with newly added HIP APIs for 6.4
Change-Id: I6052e3bf4f17d1fec23e6cc835aa2526ee0bc48c
2025-01-17 10:13:11 -05:00
Julia Jiang f497b79111 SWDEV-509295 - Merging changelog from 6.3.2 into amd-staging for 6.4
Change-Id: I0cd56b44402e82499616e961212abd2b3569c164
2025-01-17 10:12:44 -05:00
Branislav Brzak 3fd46a3783 SWDEV-508743 - [6.4 Preview] Add ROCm 7.0 breaking change fields
Change-Id: I07bff42731e74a4c409505cf8981342e22ce26be
2025-01-17 06:25:27 -05:00
Vladana Stojiljkovic 30cb2d0e67 SWDEV-498061 - Add capture support for hipModuleLaunchCooperativeKernel
Change-Id: I5ed188e046c680c2785b3952391f59ed1d0c21b8
2025-01-16 10:54:30 -05:00
Marko Arandjelovic db8527f655 SWDEV-489619 - Fix memcpy tests with capture stream enabled
- Added missing validation as graph node should not be created
 if parameters are invalid
 - Fix conversion of input params to graphNode params

Change-Id: I37ab04942b5fb2eb07386850cb7dbbf26f9ca967
2025-01-16 10:31:04 -05:00
Marko Arandjelovic c46f843b99 SWDEV-504084 - Make hipModuleGetFunction use the device the module is loaded on
If a module is loaded on one device, hipModuleGetFunction and other similar APIs should be able to run successfully from another device.

Change-Id: I96084cbd6c6dcf2a81019779a6ab1842ef2f35d1
2025-01-16 10:16:42 -05:00
Ioannis Assiouras 5e3a29078d SWDEV-505503 - Use internal device synchronize function in __hipUnregisterFatBinary
This is to avoid calling the HIP_INIT macro during the shutdown process.

Change-Id: I2e65f6e10491918a17445ee1e8ddd08286070358
2025-01-15 18:57:34 -05:00
Sourabh Betigeri 9d8d35ae40 SWDEV-507960 - Return with error code if stream of type hipStreamLegacy is being attempted to destroy
Change-Id: Iee7ada6a5a905b44360a7e4049fc8b1a45c80db0
2025-01-13 18:17:37 -05:00
Ioannis Assiouras 44b6b6813d SWDEV-503760 - Only consider allocations that are less than X% larger in a mempool request
Change-Id: I94acbca606fd4c575e2e1a9e34959ce650571867
2025-01-13 16:57:26 -05:00
Rahul Manocha 2b32d9aada SWDEV-497288 - Add validation checks to hipGraphExecNodeSetParams
Change-Id: If8cde47bc8e62414333768e01064298d8a3d80ee
2025-01-13 16:49:45 -05:00
Rahul Manocha 93cff75928 SWDEV-492165 - Add support for 16 bit atomicAdd and atomicCAS
Change-Id: I7f1e4876fe2960dff1775d27cb6443a89e146c86
2025-01-13 13:35:40 -05:00
Jacob Lambert 805462a77b SWDEV-477039 - Fix spelling typos
Change-Id: Ia46e07ec0a001b7e19bec999597028acd4ee7077
2025-01-13 12:17:46 -05:00
Aidan Belton-Schure 451b0ce768 SWDEV-482851 - Do not release last suballocator chunk
Change-Id: Ib28dc9df68e454ee0c0c699c1ff17588fd55f802
2025-01-13 10:14:40 -05:00
Daniel Livingston 541c449ce2 SWDEV-489003 - [Ubertrace] OCL/HIP profiles are missing event instrumentation
Adds UberTrace support for pre-dispatch markers and barrier begin/end markers.

Moves shared definitions out of palgpuopen.hpp into shared header
palcapturemgr.hpp.

Change-Id: I9f464c689e7ff12c54eca043fc1ad65e1836a64f
2025-01-10 11:28:52 -05:00
Julia Jiang c6e25b2be7 SWDEV-507699 - Update CLR license date
Change-Id: I51b641c58b1e9b8c84637af2d22f905bcdab8f56
2025-01-10 11:17:19 -05:00
Julia Jiang bdf48bdbf9 SWDEV-497634 - Update change log for hipMalloc allocation fix on Windows
Change-Id: If4351cb4f75141661538e1d26c96e600df3d0b39
2025-01-10 10:12:39 -05:00
Anusha GodavarthySurya 683a942364 SWDEV-480209 - Make internal callbacks non-blocking
Change-Id: Ic918d08f341abfd9a7c167d09f9c723cdc43157f
2025-01-10 02:16:11 -05:00
Saleel Kudchadker c9dd95bf6c SWDEV-504494 - Use system scope for D2H
- When using shader copy, make sure to use release scope for the AQL
  packet. This is a potential bug but is hidden as hipMemcpyAsync always
needs synchronization(which inserts a barrier with release scope). For
hipMemcpy we use a barrier packet to make sure its blocking. Eitherways
a barrier gets always used and hides in some ways a potential bug.

Change-Id: I57fb7f769c3179e76d712471c0905104c801d7ba
2025-01-10 00:34:08 -05:00
Saleel Kudchadker a22c45d635 SWDEV-508004 - Improve hipEventRecord
- Resolve stream once for event record. We should avoid calling
  getStream again in addMarker

Change-Id: I78448c4f151ae10a5c8e8c248b2f4078b84191cb
2025-01-09 16:47:46 -05:00
German Andryeyev 45a12208b6 SWDEV-507019 - Change the function lock to the module lock
Multiple functions can be located in the same module.

Change-Id: Ia4ca3db64fe5b0822584059d3770c91103665c63
2025-01-09 12:35:29 -05:00
Evgenii Averin b62995ce1a SWDEV-505769 - Fix typo
Change-Id: I2d3f65ed68157718c4439a9da7d2dcdfcbb9f93d
2025-01-08 21:31:17 -05:00
Sourabh Betigeri f5b2516f5d SWDEV-505971 - Fix size mismatch of count type to uint32_t
Change-Id: Ie526f828f816e6681ef1735d5edb2db895dace57
2025-01-08 12:47:36 -05:00
Saleel Kudchadker 39801b5750 SWDEV-506251 - Disable blit copy thresold for OpenCL
Change-Id: Id0ca43b13d5792791a42da263f6aa4496382cea6
2025-01-08 02:46:01 +00:00
Rahul Manocha 05baf9ff22 SWDEV-504215 - fix rocalution perf drop by disabling cpu wait
Change-Id: I878f3420073b05cc6241f524ac428e47c0ce823d
2025-01-07 17:02:24 -05:00
Saleel Kudchadker b63005d550 SWDEV-504494 - Flush to systemscope when copying non-coherent mem
- When we use blit(compute) copies, two subsequent copies may read for
  the same source buffer, the buffer may get modified by the host in
between and if the src buffer was allocated with non-coherent flag, the
device may simply use stale value from previous cacheline fetch. This is
a corner case.

Change-Id: I2ce261c6f6fa4e5bb608f116548e5cc711ae6f3c
2025-01-07 12:49:22 -05:00
Jatin Jaikishan Chaudhary 446ed661a0 Revert "SWDEV-505971 - change setArgument arg from uint32_t to uint64_t"
This reverts commit a6f2a2c2af.

Reason for revert: There needs to be memcpy size change

Change-Id: If4f51769731e54743ac705b19b4f81b2d5925d5a
2025-01-06 18:03:23 -05:00