AI/rocm-systems
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
3,616 Commits 3 Branches 0 Tags
ceed43fa4e387d9ede5670ea992f8734a86fe6cb
Commit Graph

118 Commits

Author SHA1 Message Date
Tao Sang e3c1bb153b Deduce hops from numa distance 
Rocr won't return real hops so that we have to deduce hops from
numa distance as a workaround. This will be subject to change
as driver team will provide a long term solution in rocm3.7

Change-Id: Ifb939ed848db190c3d544bb7f30a5821161921e6


[ROCm/clr commit: b14d9e0a36]
 2020-06-13 17:04:17 -04:00
Jason Tang e762c29a30 Get gpu asicRevision_ 
Change-Id: I8afb53e027d912843655823eebaec29901c33d28


[ROCm/clr commit: 30e36a4925]
 2020-06-12 14:51:52 -04:00
German Andryeyev 0a6056ac82 Initial HMM support 
- Expose ROCclr interfaces for HIP usage
- ROCr interfaces aren't available in staging, thus control the
build with AMD_HMM_SUPPORT define

Change-Id: Iadc2bcc230e78d3b0dc22b235189c8cc80843446


[ROCm/clr commit: c5afd5d412]
 2020-06-12 09:06:07 -04:00
Rahul Garg d9aa6e2375 Fix IPC create memory failure 
ROCr expects granular size in hsa_amd_ipc_memory_create

Change-Id: I7c266aa2eef8304f105298a7a668f4c4b3fb5f5a


[ROCm/clr commit: 617538074c]
 2020-06-11 13:02:54 -04:00
Saleel Kudchadker fca301772c Add logging support for AQL packet 
Use AMD_LOG_LEVEL=4 and AMD_LOG_MASK=8 to print AQL log
explicitly
Change-Id: I4209d91b460e64be44261d3ab773580067e47c29


[ROCm/clr commit: 2b771d2f5f]
 2020-06-10 14:04:47 -07:00
Aryan Salmanpour a760a90d35 reuse available queue with the same priority if queue creation fails 
Change-Id: I1522935ce1a73c4724135668d039e88a9097710b


[ROCm/clr commit: 5203c5271b]
 2020-06-09 17:56:18 -04:00
Aryan Salmanpour 6059c30804 fix a build error on TeamCity due to order of initialization in rocm Device class 
Change-Id: I2e45e71e53272b7a7813422d3b949e3835077ab0


[ROCm/clr commit: 1d938697a1]
 2020-06-08 13:56:39 -04:00
kjayapra-amd 01737d2472 SWDEV-239327 - Dont call hsa_amd_memory_pool_free from ~amd::Memory() for obj created by passing devptr. 
Change-Id: I315fd502df6bb1bf66dd24eac9325f1439e432ad


[ROCm/clr commit: 1b0882e061]
 2020-06-05 18:42:03 -04:00
German Andryeyev d4a985dcd3 Make sure runtime sync before CPU access 
Change-Id: I52882788ded187b3f735257a188b1b08d8502147


[ROCm/clr commit: 3828a61413]
 2020-06-05 12:00:08 -04:00
German Andryeyev 566684b6f4 Add the sync of the current queue 
Make sure runtime waits for the current queue before
synching with device queue

Change-Id: I753b6fc0bb15a3a3d4bf03fef1152842550850c0


[ROCm/clr commit: 86e0f337fc]
 2020-06-05 11:57:59 -04:00
kjayapra-amd d6a2b2247a FEAT-30761 - Adding defaults to all Switch cases in PalDevice gfx detection. 
Change-Id: Ic998dba44d3287484d6b0da4f07f320de8e6d92b


[ROCm/clr commit: 1982c16e46]
 2020-06-04 20:49:58 -04:00
Aryan Salmanpour 55c58ebfaa Add support for setting queue priority for ROCm backend 
Change-Id: I67ed5a6868af79538f7f4522d8d11c043cdf3c1e


[ROCm/clr commit: b5552aa97f]
 2020-06-04 20:16:32 -04:00
Saleel Kudchadker 2a42d5838d Set hidden mem witholding to 0 by default 
TF doesn't reserve all available memory now. If any
client wants to reserve they can explicitly set
HIP_HIDDEN_FREE_MEM env var
Change-Id: Ied3a948b79f49aa7327f6a820e9789e39cec143b


[ROCm/clr commit: d8ca3c632c]
 2020-06-04 14:37:40 -04:00
Saleel Kudchadker 0bdb497c1a Fix a potential mem leak 
If a offset of the pointer is passed to free it may release
the mem object but may not release from MemObjMap. Erase the map
by getting the parent pointer.
Change-Id: I06b92548de2d49b4029efe6b511329225007cc55


[ROCm/clr commit: 2b6fea4348]
 2020-06-04 14:05:11 -04:00
Vlad Sytchenko 20c5916b32 Handle the option USE_COMGR_LIBRARY correctly 
This is a follow up to http://gerrit-git.amd.com/c/compute/ec/vdi/+/359563. The setting is now either ON or OFF, never "yes".

Change-Id: I031d013a8d239dc72ef610da81bd31b8b78a3ba8


[ROCm/clr commit: 3a84fcd13e]
 2020-06-03 17:25:47 -04:00
Tao Sang a0506189a1 Fix TC linux build issue due to previous Numa patch 
Change-Id: I6068edaf38cac6fad187c8429707afdb727e8d41


[ROCm/clr commit: fabfc42b68]
 2020-06-03 16:42:53 -04:00
Tao Sang 6041b06b61 Support Numa-aware cpu selection 
Select cpu in terms of the smallest Numa distance for a GPU device.
This will improve performance of hipMemcpy in the mode of
hipMemcpyHostToDevice or hipMemcpyDeviceToHost for small buffer.
`

Change-Id: I2860f1f83b79be0dff7bf5e64cf68ab4448db0a1


[ROCm/clr commit: aedb9590be]
 2020-06-01 21:01:24 -04:00
German Andryeyev 3abb1347cf Revert "Revert "Reenable cooperative groups"" 
This reverts commit c6c208099b.

Reason for revert: <INSERT REASONING HERE>

Change-Id: I93c45fae27e0a08b199542d44fb0d65fc74ea13c


[ROCm/clr commit: fb401bfe6d]
 2020-05-25 14:11:58 -04:00
kjayapra-amd c7b79be23d SWDEV-237467 - Adding return type to comply with -Wall compiler flag. 
Change-Id: I3c9935105ef262cdbf1c8ee293930b018be0197a


[ROCm/clr commit: 3603272d24]
 2020-05-23 14:22:54 -04:00
kjayapra-amd e44ec607ef FEAT-30761 - Fixing fall through in gfxip major/minor detection. 
Change-Id: Ib97b3dbe993e01df3360cbeda6bd0d9d366535b6


[ROCm/clr commit: ea0137fb22]
 2020-05-23 12:42:44 -04:00
kjayapra-amd c1e072980d SWDEV-237467 - Return proper hip error codes incase of ROCclr IPC API failures. 
Change-Id: I1d018918ed71f6d80846b3017f7a15f4ab496554


[ROCm/clr commit: 53a890b499]
 2020-05-22 22:10:15 -04:00
kjayapra-amd 0cbaf84fa6 SWDEV-236110 - Fixing uninitialized variable. 
Change-Id: I26a2a6826da643b57da9746e3ce888a46c4e78f4


[ROCm/clr commit: 618d66b5fe]
 2020-05-22 20:40:24 -04:00
kjayapra-amd 95d6ff1279 SWDEV-229840 - Remove false error messages. 
Change-Id: I0346768a2a52913d5330bc2007a7706e2a439c47


[ROCm/clr commit: 32043017ed]
 2020-05-22 18:18:41 -04:00
Aryan Salmanpour 009eed8763 check for valid queue before accessing cuMask() 
Change-Id: I8d4b0dbcd097c2ec5c31dea5a3d0060f0864a7e8


[ROCm/clr commit: fec4adfd19]
 2020-05-20 16:23:09 -04:00
German Andryeyev e6b1dc9771 Add missing memory allocation in printf 
Change-Id: I452b676612b54f70106e7ef1bcb5ce2baf7b3ffc


[ROCm/clr commit: f56a052243]
 2020-05-20 14:49:59 -04:00
Alex Xie 48068bdb89 Fix compile error in certain version of GCC 
Change-Id: I27f021db908bf114a685427a47cd9f0d6b2e5693


[ROCm/clr commit: 966448c53b]
 2020-05-20 13:13:55 -04:00
German Andryeyev e20f40119c Fix async mem clear 
Optimization for the fence release removed a sync for mem fill.
Add simple const buffer management forr the filled pattern to avoid
pattern overwriting with the async fills.

Change-Id: I63773ac09ceec31d5396d24570e4647ff096326b


[ROCm/clr commit: 2ce6bbebc4]
 2020-05-20 11:13:41 -04:00
Chauncey Hui d400543947 Modified IpcDetach to return status instead of void. 
Change-Id: I68ed94b93f0383babe25eb046b4047d249a0fdc1


[ROCm/clr commit: 0af9c06968]
 2020-05-20 03:38:21 -04:00
Matt Arsenault d2cb056d06 Remove include/ from #includes 
These are unnecessary and an obstacle to producing a relocatable
package.

Change-Id: I0059bf7a2d11fcece0cd7ab47d7545d0df4d7099


[ROCm/clr commit: 1d267c9c08]
 2020-05-19 19:35:09 -04:00
Aakash Sudhanwa c6c208099b Revert "Reenable cooperative groups" 
This reverts commit 7b00a525ba.

Reason for revert: <INSERT REASONING HERE>

Change-Id: I8954b37c354382804a139d80e2551c381fd9b2ed


[ROCm/clr commit: abc115bda8]
 2020-05-19 18:21:48 -04:00
Jason Tang f94e958680 Add major/minor/stepping to device layer 
Change-Id: If82ea55a46b166b243a98089a6e9c40ccfdb479f


[ROCm/clr commit: cd2a713d63]
 2020-05-17 12:57:34 -04:00
Jason Tang 5101aa9867 Correct the way to get subtarget 
Change-Id: I47805424c0bd69547cff0ab71c369552016052b5


[ROCm/clr commit: 9b5e1fec6c]
 2020-05-15 17:18:39 -04:00
Aryan Salmanpour dee687d2d7 Add support for setting CU mask on ROCclr for ROCm backend 
Change-Id: I0dbe2eeb33467fc0f24b26929119c10e9b455da7


[ROCm/clr commit: fed94b8604]
 2020-05-15 14:23:43 -04:00
German Andryeyev 7b00a525ba Reenable cooperative groups 
Change-Id: Ia43049ef550bffa6d21704dbd306ddb9c1d56af0


[ROCm/clr commit: 82dc1a6343]
 2020-05-15 12:41:12 -04:00
Christophe Paquot 23e520003c Use system scope for packet following sdma copies 
SWDEV-234947
SWDEV-236298
Instead of forcing a barrier packet, just inject system scope on the next packet.

Change-Id: If9bcee23e08dfe5db731235e2fcb30582cbd4c1c


[ROCm/clr commit: 6a5af4056e]
 2020-05-15 12:20:06 -04:00
Matt Arsenault 24cc417ec1 Fix missing target includes for GL/EGL headers 
Change-Id: I9a31eae40cb7187dd0264ad5b9577fab96464b41


[ROCm/clr commit: 3624b8df16]
 2020-05-14 16:56:34 -04:00
Matt Arsenault 26597f888a Improve usage of target_include_directories 
Eliminates most of the global include_directories. The install header
paths are different from the build directory, so we have to separate
those for the exported target include paths.

Change-Id: I13e4c56c1218cb31c29a316422dc5fd1d09d8b1b


[ROCm/clr commit: 3a7f2e3682]
 2020-05-13 17:25:58 -04:00
Vlad Sytchenko 6313310fa6 Load versioned comgr library 
Change-Id: I4cc81f33e6889ac81a82747159bc210256f33c21


[ROCm/clr commit: 614aaa8409]
 2020-05-13 16:46:35 -04:00
German Andryeyev 6e8e12b17c Set CPU access flag for SVM 
Make sure all GPUs have CPU access flag for the fine grain buffer.

Change-Id: Ifc843c2807e70a271b269192ae7859205ff458f3


[ROCm/clr commit: 8904848abc]
 2020-05-13 16:05:46 -04:00
German Andryeyev 59b9ef5df5 Disable cooperative groups support 
Change-Id: I1b526f2228d083ecad7907a6eaf37c1dd4428277


[ROCm/clr commit: d2b9a57c4f]
 2020-05-12 14:31:10 -04:00
Saleel Kudchadker c38c8e5c3d Add env var to toggle large bar support in runtime 
Use ROC_ENABLE_LARGE_BAR (0/1) to toggle. The support is
enabled by default.

Change-Id: I6cb93a46594cb6f5e90bf6057738330225efb553


[ROCm/clr commit: d10d691e76]
 2020-05-12 13:20:06 -04:00
Jason Tang e49bddae07 device/rocm: split gfxVersion to major/minor/stepping 
Change-Id: I1e437eaee30794147713d9516229211670f01d90


[ROCm/clr commit: b4f1239f34]
 2020-05-12 12:17:13 -04:00
German Andryeyev b5f464e2af Make sure the list of HSA agents is valid 
If HIP_VISIBLE_DEVICES is active, then make sure the list of HSA
agents contains the valid agents

Change-Id: I584aad999a230ab7f88a0cfe20dcd0abe79c43a5


[ROCm/clr commit: ae4aceb55e]
 2020-05-11 15:49:30 -04:00
Christophe Paquot f03339a1e2 Fix cooperative flag for hsa_queue creation in case they're not available 
SWDEV-233766

Change-Id: If410ecfed61f2b3bb50b847cf2ededc573139494


[ROCm/clr commit: 3ed185307e]
 2020-05-11 13:40:50 -04:00
Christophe Paquot b6ecb9ce82 Add gpu().hasPendingDispatch() in the SDMA path 
SWDEV-234947

Change-Id: I8aa501f8755d136708b0d12ee3c30229c238660d


[ROCm/clr commit: 2a02026696]
 2020-05-08 18:19:51 -04:00
Michael LIAO 0e2c6ebac3 Fix build failure. 
- Also fix `-Wreorder` warning. NFC.

Change-Id: I766fdc622c9107f901a55498bdc8fef3d821d1b7


[ROCm/clr commit: 12fcfee41d]
 2020-05-07 10:39:10 -04:00
Michael LIAO b785d25506 Clear executable permission. 
Change-Id: Ia0d363b1ba89d7947e5b5a55cb67edba86f0515e


[ROCm/clr commit: 503ef06555]
 2020-05-07 10:38:58 -04:00
German Andryeyev 6620689d2b Switch PAL version to 592 
Change-Id: I7e90b8fd55c57d8d49e4ec1273ab671f96197bae


[ROCm/clr commit: 3446d4e638]
 2020-05-06 14:51:32 -04:00
Payam f66b02a3a2 name change vdi to rocclr 
Change-Id: I856d6ac1a9a83d89715d6e33dec4aa17abc2f2f2


[ROCm/clr commit: d6100a9547]
 2020-05-06 00:54:45 -04:00
Alex Xie 1e75874295 SWDEV-234684 - hipmemcpy optimization does not work in tests 
Change-Id: I899d172c5b2af88c796fe9a36f97d15ac45caf94


[ROCm/clr commit: bfbc8cd09b]
 2020-05-05 15:58:03 -04:00