AI/rocm-systems
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
12,156 Commits 3 Branches 0 Tags
3da30ed60741c9abefdcd540ccb2de136b0cd670
Commit Graph

204 Commits

Author SHA1 Message Date
Xiaomeng Hou 35a0855e4e SWDEV-373400 - Lower the threshold of single alloc percent for APUs w/ small VRAM 
Change-Id: Ie8e3504a0c468bc44633991cdfde66bcf0b74f06
 2023-03-06 10:12:01 -05:00
Saleel Kudchadker d429314b21 SWDEV-380035 - Do not create Arena Memobj for pinned memory 
Change-Id: I4da281f0b7139efb4bb46b47f812358a0212445a
 2023-02-24 12:31:25 -08:00
Anusha GodavarthySurya f98dcf9d7f SWDEV-366636 - Fix performance drop in TF-RCCL models 
Change-Id: Idc845bb0dab858b94b9d2720cae8308cac2e7328
 2023-02-17 00:48:13 -05:00
Maneesh Gupta 5dc104b3ea SWDEV-368235 - Revert "Remove obsolete env variables" 
This reverts commit 7b50c935f8.

Reason for revert: Deferred to a future release.

Change-Id: Ia66c37f0ab9734dee73c930d10d7469d5fd57254
 2023-02-15 07:25:00 +00:00
Anusha GodavarthySurya fa35239b4b SWDEV-380035 - Check for agent and ptr match for hsa LOCKED ptr 
Change-Id: I2503aa05512aebc3535963e188ca8d1dbfab08f0
 2023-02-15 00:12:42 -05:00
David Yat Sin fa1b7925cf SWDEV-365908 - Do not align mem size for IPC 
Remove alignment to granularity for IPC handles as ROCr has a patch that
will internally validate pointer sizes against requested size during
allocation instead of size aligned to page size. This patch is needed
together with this patch from ROCr:
f8a42a3a:Use user requested size for memory fragments

Change-Id: I28b25558ea03c836b44fafdb34b7330cf6887424
 2023-02-13 17:32:02 -05:00
German Andryeyev 67af16bdf3 SWDEV-381633 - Use blit manager on app's queue 
Heap initialization used device queue, but it shoudl be used for
cooperative launches only. Heap initialization must use the same queue
as the current dispatch.

Change-Id: I856621bf82bbdeb1c2d0fbc4970e90d09af805cb
 2023-02-09 09:43:13 -05:00
Saleel Kudchadker 7619b53e9b SWDEV-381633 - Better log 
Change-Id: Ie151d73aa4b1dae7d0d9acd0457e7dbdf35f21ee
 2023-02-08 17:54:11 -05:00
jatang b798c85272 SWDEV-380792 - Fix floating point exception when maxEngineClockFrequency_ is 0 
Change-Id: Ic443ceae586c4c84995ed2abef9bd7f32f8b60f9
 2023-02-07 11:43:10 -05:00
German 7b50c935f8 SWDEV-368235 - Remove obsolete env variables 
Change-Id: I7e14d53297e79e2f68b3a6cc40251ad7db9eb5ab
 2023-02-03 13:44:24 -05:00
David Yat Sin 08d72c4f98 SWDEV-380258 - Fix invalid iterator after erase 
Fix rare segfault due to invalid iterator after erase

Change-Id: Id5b54d6cf10075deff0d613fec12af249c6c55a3
 2023-01-31 17:46:15 +00:00
Jiabao Xie 3e88081961 SWDEV-366886 - Revert "SWDEV-366886 - force svm alloc for rocm" 
This reverts commit 8278e86918.

Reason for revert: performance drop in Quicksilver app

Change-Id: I2bdf42ad0a235a74e2bb4d38f86471e9affbd7a6
 2023-01-25 15:54:53 -05:00
German 6dad2fc306 SWDEV-377991 - Remove liquidflash support 
Remove amdgpu-pro interface for persistent memory,
used in Liquidflash

Change-Id: I7d1720ad0875a62ebb2d7f96cba39601d560a5df
 2023-01-23 10:40:57 -05:00
jiabaxie 8278e86918 SWDEV-366886 - force svm alloc for rocm 
Change-Id: If85a6aaeb5861ff1550a0bc3f0a25073ff8145e7
 2023-01-18 16:26:34 -05:00
German c8927cd84e SWDEV-377991 - Remove Liquidflash extension 
Initial check-in to untie dependencies with HIP and OCL repos

Change-Id: I363b63954c3f118f40a6ed893545d6a4ac44144c
 2023-01-18 13:16:20 -05:00
Anusha GodavarthySurya 2b50636809 SWDEV-373173 - Correct max VGPRs and VGPR Granularity in Occupancy calculation for different asics 
Change-Id: I6f595428ed31733cdc04e65e7c3e383f454ab72b
 2023-01-18 00:22:02 -05:00
Anusha GodavarthySurya 274f2de391 SWDEV-364576 - initialize device malloc heap state using blit kernel 
Change-Id: I5d0172aff7d2c04b322a4d828b8a2b438158b80f
 2023-01-07 06:53:53 +00:00
jatang e8cd1e93e8 SWDEV-1 - Not to truncate size_t in print when it's 64bit 
Change-Id: Iab8b7eb128c767b6808b19cbeac5169b7a97f1ad
 2022-12-13 09:13:33 -05:00
kjayapra-amd e56a611b92 SWDEV-371904 - Adding pseudo fine grain flag to hsa memory allocation for device fine grained memory. 
Change-Id: I8cada90f0e3880dfbc5bf5a3fac4554e7a0cb08e
 2022-12-11 08:15:17 -05:00
pghafari 402504f548 SWDEV-366279 - updating p2p error msg 
Change-Id: I294e7770efd11f511dd5817eb28bd3c97d9d4926
 2022-12-09 00:05:04 -05:00
Ioannis Assiouras b445717f72 SWDEV-369547 - Ensure that ipc_mem_detach is not called for non-ipcShared MemObj. 
Change-Id: Ia73b60b9f08d593301879e0f72c395edbc215112
 2022-11-30 02:16:13 -05:00
Anusha GodavarthySurya 446a3fc688 SWDEV-354074 - cache hsa queue and delete when device is destroyed 
Change-Id: I26365521d785f0bc612e32bdcdb6caacb3af9a11
 2022-11-23 07:17:45 +00:00
German b40823ccc1 SWDEV-328670 - Enable ROCr interop 
Change-Id: I932afa2a0e75b1f1771c4be1e984b32966636afb
 2022-11-17 09:40:34 -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
Saleel Kudchadker 3e465c5ff8 SWDEV-364052 - Print free memory on queue callback 
Change-Id: I7d15b6f6277326d5f3e9b784a3443f18ec6ba58a
 2022-11-04 13:19:33 -04:00
Ajay 261620a2ef SWDEV-357207 - Linux: enable StreamWrite/Wait APIs in deviceAttribute 
Change-Id: I593df2dbb7a0233dd42f8d873510d85bbb27feff
 2022-10-21 12:44:35 -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
kjayapra-amd 951a5b9e7d SWDEV-307411 - Require comment/message string in guarantee(). 
Change-Id: I55a699c5366f6a7d167ede4c9be2ec5b15cea9d0
 2022-10-13 16:21:21 -04:00
Saleel Kudchadker 9b5cbd37a2 SWDEV-352001 - Store last scopes for dispatch 
- Store last fence scopes and use the last value to determine if we need a cache flush again. This helps cases where hipExtLaunchKernel API is
used.
- Purge code for ROC_EVENT_NO_FLUSH

Change-Id: I531cf9c9c60d5e2b3a9e265d0f52f79ed2fa8a8c
 2022-09-22 11:34:10 -04:00
Maneesh Gupta 4afca0647e SWDEV-352502 - SWDEV-355630 - Revert "SWDEV-299940 - Query global memory available on device with HSA attribute HSA_AMD_AGENT_INFO_MEMORY_AVAIL" 
This reverts commit 73021582d3.

Change-Id: I86175ccf2b543c3ff3dffec6ccae410976972f22
 2022-09-12 07:34:32 -04:00
Sourabh Betigeri b15b3173f4 SWDEV-326798 - Avoids waiting for the signal with a timeout and instead queries the current status 
Change-Id: I8b4052d5439ca930ceb924da30cb0335f0826a95
 2022-09-06 12:49:46 -04:00
sdashmiz 50e0ddb055 SWDEV-350569 - add proper indexing for attribute 
- rocr attribute needs to be updated after each iteration

Signed-off-by: sdashmiz <shadi.dashmiz@amd.com>
Change-Id: I3afb2d7954ef3de37f5f5f9d3cc7757fdacffcec
 2022-08-18 09:59:42 -04:00
Sarbojit Sarkar a28b22d9b4 SWDEV-344210 - Fixed page fault when mempool accessed from remote device 
Change-Id: Ie41b0c0499f7733d4541ccd50b3d0d976c6431c9
 2022-08-09 23:26:27 -04:00
Anusha Godavarthy Surya 73021582d3 SWDEV-299940 - Query global memory available on device with HSA attribute HSA_AMD_AGENT_INFO_MEMORY_AVAIL 
Change-Id: Ifcfb95f59d110b3b037a7131be21d5348601d2e2
 2022-08-05 06:55:12 -04:00
sdashmiz e176e27bf7 SWDEV-334233 - add support for p2p in windows 
Signed-off-by: sdashmiz <shadi.dashmiz@amd.com>
Change-Id: I9109120b5444c400e65cfff869cb36e876ffd1fc
 2022-07-14 15:07:33 -04:00
Tao Sang 1e26165cd0 SWDEV-286739 - Support hipDeviceAttributeWallClockRate 
Part 1: Query constant frequence of wall clock from RocR

Change-Id: I52cbba6d67d11cde6d019c5ab530059f426a9bf2
 2022-07-12 17:53:11 -04:00
Saleel Kudchadker faaa41aab8 SWDEV-335626 - Use ROCr copy for IPC 
Detect IPC buffer and use ROCr copy api instead of blit

Change-Id: Ie6bdd6fc45dbd7457611011d81570b53d5fd5276
 2022-07-08 13:32:19 -04:00
Saleel Kudchadker 5df34a2f7a SWDEV-335780 - Indicate if handler is queued 
Maintain status of handler callback. For event records we no longer
submit callbacks to reduce the load on the async handler thread. However
without a callback we leak command memory/decrement refcounts. Indicate
status of the handler which we can use to queue a callback when
finish is called.

Change-Id: I89fd02f3d047a0e8162664ee17581a14795f1928
 2022-06-14 20:55:06 -04:00
German Andryeyev 4975f69337 SWDEV-339296 - Delay hidden heap allocation till the usage 
Move hidden heap creation to the kernel launch to make sure it's
allocated on the actual first usage.

Change-Id: I1b65a82fc06d9129ed45a69765bf14ea3d945b04
 2022-06-14 12:18:34 -04:00
German Andryeyev 830898753d SWDEV-336024 - Clear device heap to 0 in ROCr path 
Change-Id: Id100ca6d6d5bd7fb16ca8c98ff0b12c9df1d69ab
 2022-05-20 11:51:08 -04:00
German Andryeyev 07c1b9a998 SWDEV-336024 - Clear device heap to 0 
This reverts commit 04bfd93569.

Reason for revert: Fix regressions

Change-Id: I7d883e1c3cbd27bb64b581ec800243ad7dfe24fd
 2022-05-19 09:10:08 -04:00
German Andryeyev 04bfd93569 SWDEV-336024 - Clear device heap to 0 
The heap must be cleared once per device, but ROCclr doesn't
create a queue per device in HIP. Hence, the clear operation will
be performed during the first queue creation.

Change-Id: I52ceb06d67d11cde6d019c5ab510059f426a9bfb
 2022-05-11 11:03:56 -04:00
Christophe Paquot b4645c7d4e SWDEV-322620 - Virtual Memory Management 
Add a virtualMemoryManagement_ flag to device.info.

Change-Id: Iabd039010d83fc51b4bcef600c609f5c65e7b1ae
 2022-05-09 22:54:42 -07:00
Julia Jiang b7c7917256 SWDEV-334574 - Rename _bkendDevice in VDI 
Change-Id: I1c04dad226e08f02bca11fa0d1981fafa7ea2d2a
 2022-04-27 11:21:24 -04:00
Sarbojit Sarkar 6b15e0a1cc SWDEV-333438 - Fix for hipEnablePeerAccess segfault 
Change-Id: I60720d1d9b9c522d15fe17dcfbc609571a4fd266
 2022-04-26 05:21:52 -04:00
Christophe Paquot 67657d6099 SWDEV-322620 - Virtual Memory Management 
Implement map/unmap for PAL backend
Create commands since PAL uses the IQueue to map/unmap

Change-Id: I97e26a7d28ae5e10774c9ca65307153100945621
 2022-04-22 18:09:26 -04:00
sdashmiz 15f1632dfa SWDEV-204804 - Detecing pcie atomic support 
- check pcie atomci support for printf functionality
- if not enabled printf wont work

Signed-off-by: sdashmiz <shadi.dashmiz@amd.com>
Change-Id: Ib366e8e71772b02210c4a830bca4bd8cc7a11664
 2022-04-22 08:53:16 -04:00
Saleel Kudchadker 8864e53265 SWDEV-333237 - Disable cache status 
Enable Cache status only for ROC_EVENT_NO_FLUSH

Change-Id: I0de4c5af2226bccd66fd704be23c2db33050f2e2
 2022-04-20 18:12:33 -04:00
Christophe Paquot b5f555f9ec SWDEV-322620 - Virtual Memory Management 
Adding virtual memory management APIs to rocclr.
The HIP layer will handle virtual allocs on devices.

Change-Id: Ia978f105c2c3fed3959c77580ba228e845105754
 2022-04-15 00:10:02 -04:00