AI/rocm-systems
2
0
Forka 0
Codice Problemi Pull Request Actions Pacchetti Progetti Rilasci Wiki Attività
4.687 Commit 3 Rami (Branch) 0 Tag
b5799c4dbe60247593226ea3c19a893abc8595f2
Grafico dei commit

4687 Commit

This Branch
This Branch
Tutti i branch
Autore SHA1 Messaggio Data
Satyanvesh Dittakavi b5799c4dbe SWDEV-363536 - HIP_VISIBLE_DEVICES= should disable the device 
Setting HIP_VISIBLE_DEVICES= should be treated as invalid device
which makes all the devices invisible to the app. This matches
the CUDA behavior

Change-Id: I937ac4c0b7dacff776cdbe692d4576c81b86ee2d
 2022-11-10 01:25:46 -05:00
Maneesh Gupta a52f5bda8f SWDEV-299940 - Query global memory available on device with HSA attribute HSA_AMD_AGENT_INFO_MEMORY_AVAIL 
This reverts commit 4afca0647e.

Reason for revert: ROCr query should now be usable in upcoming release.

Change-Id: I2207761ca6af5d585d090bae1af09eb9a8e9bad6
 2022-11-09 10:48:47 -05:00
Chauncey Hui 0b4465efdb SWDEV-2 - Change OpenCL version number from 3525 to 3526 2022-11-09 03:00:07 -05:00
Juan Manuel MARTINEZ CAAMAÑO 40f75306d5 SWDEV-286150 - [NFC] Refactor repeated option parsing code into function 
Change-Id: I606dc1cd48d880974142e523d16f5d9ac6f3aff1
 2022-11-08 10:29:13 -05:00
Juan Manuel MARTINEZ CAAMAÑO f5e982cc07 SWDEV-286150 - [NFC] Avoid copying string when calling HashString 
Change-Id: Ic2f4d469ca319f7d31814b3955430dbbfdad5bb2
 2022-11-08 07:07:06 -05:00
Chauncey Hui 773d85e3d9 SWDEV-2 - Change OpenCL version number from 3524 to 3525 2022-11-08 03:00:07 -05:00
Julia Jiang 9d616ad0f4 SWDEV-360748 - Update return value for hipExtGetLinkTypeAndHopCount_Positive_Basic test on Windows 
Change-Id: I0eb984da6e96c2409e1b1bbda61ae694c8bebe94
 2022-11-07 10:00:32 -05:00
German c1b22840fa SWDEV-286150 - Switch PAL to 775 version 
Change-Id: I8e64cbe0fea273d4bd6f7d3ebd66feb628fac6b9
 2022-11-07 09:54:22 -05:00
Chauncey Hui d0a57e4905 SWDEV-2 - Change OpenCL version number from 3523 to 3524 2022-11-05 03:00:05 -04:00
Saleel Kudchadker feca11d5e3 SWDEV-301667 - Improve logging 
Change-Id: Ifa6da876b85cb503967cf09aac6d477b10db8e63
 2022-11-04 18:23:18 -04:00
Saleel Kudchadker 3e465c5ff8 SWDEV-364052 - Print free memory on queue callback 
Change-Id: I7d15b6f6277326d5f3e9b784a3443f18ec6ba58a
 2022-11-04 13:19:33 -04:00
Todd tiantuo Li 246429bd68 SWDEV-302996 - add gfx1103 Phoenix support - OCL runtime 
Change-Id: Ib6b7efa085b734bda6b5b50c720fe9d93e6dbf95
 2022-11-04 12:07:22 -04:00
Chauncey Hui cfd9ced308 SWDEV-2 - Change OpenCL version number from 3522 to 3523 2022-11-04 03:00:06 -04:00
kjayapra-amd ea0ce45ad3 SWDEV-329423 - Enable NV3x target ID related changes on mainline. 
Change-Id: I54fafb371976aaa16a452ec65b0c340c5682c264
 2022-11-03 21:11:24 +00:00
German 473621c008 SWDEV-365121 - Use CP DMA for tiny transfers 
Sync between compute and SDMA engines can be very expensive under Windows.
Use CP DMA for tiny transfers (< 1KiB) to avoid syncs and improve performance.

Change-Id: I9db39a2199f7b9e337ed08fd36d9cbc150502f1f
 2022-11-03 10:04:53 -04:00
Tao Sang 06573ac92f SWDEV-363140 - Support performance counter on gfx11 
Change-Id: I2216f71f4d4db6dd3766333b0c821cb3d35d7851
 2022-11-03 07:59:24 -04:00
Chauncey Hui 4b3ff92abd SWDEV-2 - Change OpenCL version number from 3521 to 3522 2022-11-03 03:00:09 -04:00
Laurent Morichetti b043b4f5a2 SWDEV-351980 - Store the correlation ID in the dispatch packet 
The profiler can retrieve this correlation ID to attribute waves to
specific dispatch locations.

Change-Id: I700e8a91219d612f6a2028c0dda0c92753f3526a
 2022-11-02 23:40:10 -04:00
Chauncey Hui b070ba9cc3 SWDEV-2 - Change OpenCL version number from 3520 to 3521 2022-11-01 03:00:05 -04:00
jatang 166ef6b622 SWDEV-293816 - Only dispatch Counter Aql packet when it is created successfully. 
Change-Id: Ic8a7374d19012754c6de5d483ec5d07a56661d1b
 2022-10-31 15:11:41 -04:00
Saleel Kudchadker cf1fd98074 SWDEV-301667 - Use queue base address for AQL log 
Change-Id: Ie525991b341deeb93ab3cb2d77550e8e1d87ee20
 2022-10-31 13:54:49 -04:00
German ff6b4db70b SWDEV-363074 - Clean-up sync between SDMA and compute 
HIP can't rely on the resource tracking, used in OCL and requires different explicit sync.
Make sure ROCCLR syncs compute only when SDMA is used and vise versa.
The new logic will allow to enable CPDMA without unnecessary waits.

Change-Id: Ib9d1788cfd5afa5ea2fec4c96a37d8b9c4d0059d
 2022-10-31 10:02:01 -04:00
Chauncey Hui 3bce4df27d SWDEV-2 - Change OpenCL version number from 3519 to 3520 2022-10-29 03:00:09 -04:00
German dc4ad8c99c SWDEV-363074 - Enable resource cache for SVM 
Blender creates and destroys big allocations during the benchmark.
That causes big delays, because vidmm has to page-in/page-out memory.

Change-Id: I2baf4545807127406e3d2870a7581ff9ae7bcdb5
 2022-10-27 16:39:29 -04:00
Chauncey Hui cff74ae551 SWDEV-2 - Change OpenCL version number from 3518 to 3519 2022-10-25 03:00:08 -04:00
German e223b0f678 SWDEV-352487 - Don't add notifications as the last command 
Change-Id: Ifed34485839ef2c9491e8e8f6bb3569932160b1c
 2022-10-24 09:39:03 -04:00
Chauncey Hui a65ed69c1b SWDEV-2 - Change OpenCL version number from 3517 to 3518 2022-10-22 03:00:06 -04:00
Ajay 261620a2ef SWDEV-357207 - Linux: enable StreamWrite/Wait APIs in deviceAttribute 
Change-Id: I593df2dbb7a0233dd42f8d873510d85bbb27feff
 2022-10-21 12:44:35 -04:00
German c1c5127875 SWDEV-360183 - Finish the trace if queue was destroyed 
- Make sure SQQT trace is captured for RGP server if the queue is destroyed before normal capture is done.
- Remove prepare queue from the logic. It's not really used for any HW capture and can cause RGP server abort if destroyed before capture is even started(delayed capture)

Change-Id: I6eb19963190a5769c6477a5496c1b831a6d59b89
 2022-10-21 10:44:10 -04:00
Alex (Bin) Xie 06867f0c04 SWDEV-361591 - clean up header files 
Change-Id: I25ca37a94524da0521f5b856a9fa9bab73f103c1
 2022-10-21 10:23:55 -04:00
Chauncey Hui 904f3e3d75 SWDEV-2 - Change OpenCL version number from 3516 to 3517 2022-10-21 03:00:52 -04:00
Tao Sang 3d281114fb SWDEV-360706 - Fix WriteAqlArgAt() crash in X86 
In X86 on Windows, sizeof(size_t)=4, but size=8, for
amd::KernelParameterDescriptor::HiddenGlobalOffsetX/Y/Z items.
Loose the condition to prevent crash.

Change-Id: I2216f71f4d4fd6dd3766023b1c821cb3d35d7848
 2022-10-20 18:04:49 -04:00
Sourabh Betigeri b2665ccc4b SWDEV-361730 - Fix to match the kernel arg allocation sizes between gpu visible mem and host/runtime mem 
Change-Id: I6495aa57e97b46373c1d90cc230435780fdcf3ef
 2022-10-20 14:20:37 -04:00
Ajay a5b550e828 SWDEV-360944 - palBlit setArgument desc.size_ --> argSize 
Fixes AMF tests that were impacted on Windows

Change-Id: I0326e794ec5a34fdec7a5ff4ed79d5bd055997d2
 2022-10-20 12:44:15 -04:00
Laurent Morichetti 9a82118c85 SWDEV-362046 - Report HIP_OPS activities using the ROCr driver_node_id instead of the device's index 
The ROCclr assigns zero-based IDs to GPUs in the order they are
discovered. That zero-based ID is what is used to identify the GPU
on which the HIP_OPS activity took place.

When multiple ranks are used, each rank's first logical device always
has GPU ID 0, regardless of which physical device is selected with
CUDA_VISIBLE_DEVICES. Because of this, when merging trace files from
multiple ranks, GPU IDs from different processes may overlap.

The long term solution is to use the KFD's gpu_id which is stable
across APIs and processes. Unfortunately the gpu_id is not yet exposed
by the ROCr, so for now use the driver's node id.

Change-Id: Ib78854527d600d175bb76e2df0747c33f898c615
 2022-10-20 12:31:30 -04:00
Julia Jiang dacd55f3d7 SWDEV-357122 - fix failure in vdi so as to handle unreasonable input size in MallocManagedNegativeTest 
Change-Id: I1ed1916b652afc67327b0935c3c60fc2a404df30
 2022-10-20 12:19:22 -04:00
Chauncey Hui 05b2bd7995 SWDEV-2 - Change OpenCL version number from 3515 to 3516 2022-10-20 03:00:06 -04:00
Christophe Paquot 470c4d7335 SWDEV-322620 - Virtual Memory Management 
Hooked up the start VA hint address to PAL.

Change-Id: I4301d3eaaabcc85faf63fb25291291dcc3e8ebd1
 2022-10-19 10:50:07 -04:00
Chauncey Hui 5cad630425 SWDEV-2 - Change OpenCL version number from 3514 to 3515 2022-10-19 03:00:13 -04:00
kjayapra-amd 951a5b9e7d SWDEV-307411 - Require comment/message string in guarantee(). 
Change-Id: I55a699c5366f6a7d167ede4c9be2ec5b15cea9d0
 2022-10-13 16:21:21 -04:00
Chauncey Hui 4e0e0666d6 SWDEV-2 - Change OpenCL version number from 3513 to 3514 2022-10-13 03:00:09 -04:00
Saleel Kudchadker 6405b6cdba SWDEV-359548 - Detect fence optimization 
- Use a dirty flag to determine fence optimization
- If fence is dirty submit a marker at top level to sync.

Change-Id: I53fb19b5bb05b7c7b37c41637a6c7aaf870b639a
 2022-10-10 14:15:16 -04:00
Chauncey Hui 621faa03ec SWDEV-2 - Change OpenCL version number from 3512 to 3513 2022-10-08 03:00:09 -04:00
Saleel Kudchadker eef2f99513 SWDEV-356482 - Improve logging 
Change-Id: I0c6286fcdb9d94398ed66262e95e8837898ff9fe
 2022-10-06 11:13:24 -04:00
Chauncey Hui 3e3fa68657 SWDEV-2 - Change OpenCL version number from 3511 to 3512 2022-10-06 03:00:06 -04:00
Sourabh Betigeri 84fbb30b7c SWDEV-357246 - Adds a missing return statement 
Change-Id: I2216f71f4d4fb6dd3766023b0c821cb3d35d7849
 2022-10-05 16:29:32 -04:00
Laurent Morichetti e00965df50 SWDEV-351980 - Add FillBuffer byte count to the record 
Change-Id: I90c791f5810b8a3f6b1d6a9e81c165b1a7515c92
 2022-09-30 21:20:14 -07:00
Chauncey Hui 800fe4b239 SWDEV-2 - Change OpenCL version number from 3510 to 3511 2022-09-30 03:00:10 -04:00
Jeremy Newton 65fd4f5f07 SWDEV-1 - Fix typo in compiler utils 
Found by the Debian maintainers:
https://github.com/ROCm-Developer-Tools/ROCclr/issues/28

Signed-off-by: Jeremy Newton <Jeremy.Newton@amd.com>
Change-Id: I8c09913f7b7f4ad02f12a0630f15e66bdd5360a7
 2022-09-29 18:14:29 -04:00
Chauncey Hui f93d921c4f SWDEV-2 - Change OpenCL version number from 3509 to 3510 2022-09-23 03:00:11 -04:00