198b7b412d
SWDEV-417317 - Bypass agent init for HIP.
...
Change-Id: I3396cf7c1feced03d29f5e5ff3c77474f4f7e1ec
2023-09-06 00:40:54 -04:00
96ac99b4a0
SWDEV-419664 - Set reserved0 field of hsa_kernel_dispatch_packet to zero
...
Change-Id: I2e9c503f1c393548aacad47d40da90a2126e522c
2023-09-05 19:10:33 -04:00
7dc78d234d
SWDEV-418855 - Limits the 'no GWS' approach to gfx940, gfx11and gfx12
...
Change-Id: Iab2d34d3142902517124cec7ef3461cf7aa4b98c
2023-08-30 23:48:02 -04:00
ad56c20cd0
SWDEV-418038 - Remove opengl32.lib from linking in runtime
...
Change-Id: If9416742cd8dc6a35adaccce802d65888e808ef2
2023-08-30 14:07:28 -04:00
94e70bee26
SWDEV-419034 - Use MADV_HUGEPAGE for large host allocations
...
Change-Id: I80bb1839cdd47eb64a97467c8b01fcdf37195ad5
2023-08-30 14:03:26 -04:00
f3dc04a50d
SWDEV-410182: Link device-libs when compiling source to bitcode
...
This is related to SWDEV-410182, but it's not enough to fix it.
Functions from device-libs are precompiled into llvm-ir in a "target agnostic" way
(in reality, it's not 100% target agnostic, which brings us many headaches).
When linking builtins (like device-libs) from the command line, we use the flag
-mlink-builtin-bitcode. The difference between regular linking of bitcode and
this flag is that the later propagates target-specific attributes. If this
attributes are not propagated, we can end up with incosistent target attributes.
Comgr provides the action AMD_COMGR_ACTION_COMPILE_SOURCE_WITH_DEVICE_LIBS_TO_BC
for this exact reason. The old action is currently deprecated and this one should
be used.
Change-Id: I518415214debdf4fedf0b1d81456d6e9fb8a3d19
2023-08-30 04:24:17 -04:00
6a0f80a03d
SWDEV-381625 - Parse compiler and linker options from environment variable.
...
Change-Id: Id5a012b678e5973c4b64dff84444a909aefae006
2023-08-29 20:24:27 -04:00
bd00826446
SWDEV-3 - Move PAL to version 818
...
Restore PAL platform destruction.
Update CmdAllocatorCreateInfo::AllocInfo for the new interface.
Change-Id: Iea418eed7ee26166039a4a9cc1999438856e9097
2023-08-29 12:46:28 -04:00
1ec0ba3537
SWDEV-301667 - Use large signal pool
...
Use large signal pool if profiler is connected or profiling forced
enabled. This is needed to mitigate signal creation overhead when
profiling as signals are attached to every packet and deeper batch may
show overhead of signal allocation.
Change-Id: I8034b8a20b55328b87d593bf044f59672f9653e8
2023-08-24 19:17:05 -04:00
077311153a
SWDEV-407533 - [ABI Break]Purge unused env vars
...
Change-Id: I627950e8ebb6299affc602754a20d442dbe42b14
2023-08-24 14:11:40 -04:00
d433df4761
SWDEV-417727 - Fix hipSignalExternalSemaphoresAsync()
...
This reverts commit 44a3935cda
2023-08-23 22:31:27 -04:00
44a3935cda
SWDEV-417727 - Fix hipSignalExternalSemaphoresAsync()
...
Let ExternalSemaphores be signalled only after prior works on the
stream have been finished.
Change-Id: I856917db905f68f55fdf484f5267f7fe8ea3117f
2023-08-23 14:58:37 -04:00
d97cc0abbd
SWDEV-404889 - Inital change for debugger support
...
- Program unique AQL index for debugger. The logic manages AQL array of packets per HW queue.
- Provide debug state to PAL
Change-Id: I38fa1f5435fa711fd1d44dc391f2e61eb2a25efa
2023-08-23 13:21:58 -04:00
6f5277c701
SWDEV-408473 - Add wait time of 10 us if the waiting signal copy was < 24K.
...
Change-Id: I438ec9eb07e5034042a4a9a5e6e51d74daba2c83
2023-08-23 10:46:33 -04:00
06dcaae619
Revert "SWDEV-410182: Do not force wave64 for geekbench"
...
This patch did not consider the dicussions in SWDEV-270908
> "we found that in GeekBench5, forcing Wave64 instead of the default
> Wave32 compute policy yields big gains in every subtest except one"
This reverts commit d6dc82b220
2023-08-21 03:29:32 -04:00
7a10423a44
SWDEV-417691 - Error handling for MemoryMap OS calls.
...
Change-Id: I0a823289b274e118c8440f2d372a5216bf4ba6aa
2023-08-21 00:01:06 -04:00
9bd592e2ff
SWDEV-415115 - Fix memory commit in SVM on large bar systems
...
Skip memory commit for linux only.
Change-Id: I6304607f7c5516c76e59afa6da57de5b088f6a8c
2023-08-18 11:25:28 -04:00
b4933f1d00
SWDEV-403758 - support new ASIC Revisions for Hawk Point1 and Hawk Point2
...
Change-Id: Ib8383b16e85abaa0bbf481c1aac71a11cd5b9218
2023-08-18 10:32:09 -04:00
b14bd96d3c
SWDEV-274425 - fix bad assignment of imageOffset.y and imageOffset.z to PAL
...
Change-Id: I6adce05928f6ff61aad845f289687d9bc0d17476
2023-08-18 00:17:31 -04:00
cb426df1bd
SWDEV-416580 - Add condition when memory has direct access, only use host fill if image is small
...
Change-Id: I3509c4aa21f6413adad3b46273ec650f5c577ddd
2023-08-17 17:23:49 -04:00
0d3a5c4c81
SWDEV-414828 - set AMD_DK_ROOT variable to env DK_ROOT
...
this ensures python to be picked from DK
Change-Id: Idd006b7a2bb819c6aa747136aa55cfc28737a5a9
2023-08-17 12:21:16 -04:00
289535e805
SWDEV-412393 - Force alloc memory to avoid another hsa image creation.
...
Change-Id: Ia3cd99eb736231e6dfe013ebae6c41fd4cc657bc
2023-08-17 05:18:43 +00:00
d6dc82b220
SWDEV-410182: Do not force wave64 for geekbench
...
Change-Id: I505af6b0ca70e54a963abe9b097b91eea94d3268
2023-08-16 10:22:52 -04:00
dc401c520b
SWDEV-416457 - clr to handle metadata ActualAccQual
...
Change-Id: I4e4c52093ed0f65830dbba19266872d3405c4916
2023-08-15 13:47:34 -04:00
4fb24c48b0
SWDEV-274425 - fix bad initialization of array slice index for 1d image array in partialMemCopyTo()
...
Change-Id: Id9a0c55d4296de8cd89f1870b2c8028ce6cf601b
2023-08-15 12:55:10 -04:00
7e3d8243b7
SWDEV-385303 - remove bad row pitch assignment
...
Change-Id: If959be2fa51035c54c3b08e74711077ad52ac4c8
2023-08-14 15:21:56 -04:00
8a538b5639
SWDEV-384557 - Reset mask at in the constructor
...
Change-Id: I15dafd281213c03590bcae4bf4544e33c4a7e277
2023-08-11 17:22:27 -04:00
618cc4d9ee
SWDEV-416527 - Disable PAL platform destruction until a proper fix in PAL
...
Change-Id: I465aa15959187eb27354caa9b8308f1266be46ed
2023-08-11 10:51:37 -04:00
63af50b6e0
SWDEV-415160 - Added fix in kernel argument logging
...
Change-Id: Ic54d51264602c7bf0c91b5273f9519da3fbb1062
2023-08-03 21:39:41 +01:00
aa6eb555e2
SWDEV-384557 - Enable SDMA query
...
Change-Id: Ibb0a8d131f799985a4d4adbf753261e58c04157f
2023-08-01 18:41:23 -04:00
ff1a999f66
SWDEV-408283 - Sync scratchRegs_, privateMemSize_ and workitemPrivateSegmentSize.
...
Change-Id: I623a7140810ff9867f8816bf4c8621a1fe921744
2023-07-27 00:31:54 -04:00
0711320cb4
SWDEV-392732 - CP processes until doorbell ring
...
Change-Id: I3a0974435d3f95a40cea6ed326f744bcab5b4c0f
2023-07-25 13:10:31 -04:00
b49e8e78e1
SWDEV-409815 - Fix image view cache issue
...
Add a view bit to avoid original resource destruction when parent
dependency doesn't exist with the image view cache
Change-Id: I8277afd575af8f29951c5d1a9f7d94d784251657
2023-07-25 12:30:04 -04:00
c674d94a12
SWDEV-396372 - Add support for gfx1150 and gfx1151
...
Change-Id: If97710f18fbb61bead5cfe92ce11e11b4895b029
Signed-off-by: Lang Yu <Lang.Yu@amd.com >
2023-07-24 20:11:03 -04:00
cb7776ef07
SWDEV-412496 - treat USE_PROF_API to always be true
...
Change-Id: I77cfac409215a33d11228a0f2da9ffba63c3d73a
2023-07-21 04:42:57 -04:00
5447cf8872
SWDEV-301667 - Disable HostBlit copy for HIP
...
Change-Id: I46333ff42e8c1d402ece97e3ead7b539a27c3f82
2023-07-17 17:49:11 -04:00
04b9ab49eb
SWDEV-333557 - add PAL_HIP_IPC_FLAG for PAL HIP device allocations
...
Change-Id: I9017f4e3b03d4817bf233c788e30775fb2297589
2023-07-17 08:10:25 -04:00
b0e6f99ad7
SWDEV-392732 - Initial commit for graph doorbell optimization(AQL Buffering)
...
Change-Id: I451725006c54c249dc530c55d2af2a31594bf49b
2023-07-16 07:56:00 -04:00
8c7f59663f
SWDEV-408891 - Only register the host pointer to the current device.
...
Change-Id: If5397f148638e3b7d542b9e6fb8ac749a602b35e
2023-07-12 21:44:52 -04:00
8434feed16
SWDEV-333557 - 1.Add support for IPC memory to enable hipIpcGetMemHandle() and hipIpcOpenMemHandle() in PAL path. Set interprocess flag for device allocations in HIP PAL. 2.Fix PAL IPC path with ResourceCache and SubAllocation 3.Minor fixes for IpcBuffer constructor and IpcAttach 4.Remove redundant checks that may go wrong
...
Change-Id: Ie9d99847c7c2c7b3b2aaefaaf60d23bf71b68635
2023-07-12 17:49:32 -04:00
6b89980fb2
SWDEV-409815 - Fix image view cache issue
...
Make sure parent_ field is cleared for the internal image views.
The internal image views don't require dependencies tracking.
The issue appeard only when Navi10 pitch workaround was enabled.
Change-Id: I376d212750085a9391f8c32fc2979dcb5d93c89c
2023-07-11 11:22:29 -04:00
7912f3af89
SWDEV-409299 - Vega clinfo is not working
...
Change-Id: Ia48bc6f130bd102dff210b105de6f9c02ebbe012
2023-07-10 09:53:50 -04:00
ad2c1e899a
SWDEV-408650 - Report CL_DEVICE_HALF_FP_CONFIG query
...
If cl_khr_fp16 extension is enabled, then OCL runtime should report CL_DEVICE_HALF_FP_CONFIG.
Change-Id: I7c4ac48387f80bc704a475c57e5b52a462090d1b
2023-07-06 11:58:16 -04:00
96fb1ac289
SWDEV-1 - Bump min required ROCr version
...
ROCr version needs to be > 1.11 due to aca7d9e14a
2023-07-06 05:17:52 +00:00
aca7d9e14a
FEAT-47686 - Changes to support new uncached memory segment in ROCr.
...
Change-Id: I0ba8769d6737cdf1cc8a8644b2e82109f584a430
2023-07-05 20:47:51 -04:00
90f53925a3
Revert "SWDEV-403995 - uniform_work_group_size is interger type is 1 for true and 0 for false. Lighting kernel's init sets uniform work group flag and hence not need to set it from options."
...
This reverts commit 2c9c58e7bb
2023-07-05 04:41:40 +00:00
8c1232124e
SWDEV-405329 - Fix cuMask issue for WGP mode
...
- Enable CUs adjacent pairwise for WGP mode
- In HostQueue::terminate() do not segfault if virtual device hasn't been created
Change-Id: I94402ff333308af5824878086cc238b3993d534d
2023-06-30 01:09:01 -04:00
770b2a4711
SWDEV-384557 - Rename env var
...
- Rename HIP_USE_SDMA_QUERY to DEBUG_CLR_USE_SDMA_QUERY as this is
supposed to be a temporary env var for debug purposes only.
Change-Id: If6ebd52ab87624375a3df24ceccdcc05c60a65af
2023-06-29 13:54:55 -04:00
a034995f81
SWDEV-408213 - Move the init_heap blit kernel inside linear blit kernel grouping.
...
Change-Id: Id540018d4cceab5f01f614f7d49ef42505e69039
2023-06-29 10:36:53 -04:00
d29755452b
SWDEV-396088 - Add image view cache
...
Blit manager requires an image view to reduce the amount
of copy kernels. Creation/destruction of a view in ROCr is
an expensive operation. Thus, runtime can cache views for fast access.
Change-Id: Ia67d775b481cc8326d91215ca22d4a73c1dddb59
2023-06-28 09:44:05 -04:00
Jaydeep Patel
Ioannis Assiouras
Sourabh Betigeri
German
Satyanvesh Dittakavi
Juan Manuel Martinez
kjayapra-amd
Saleel Kudchadker
Tao Sang
Todd tiantuo Li
victzhan
Ajay
Alex Xie
Anusha GodavarthySurya
Lang Yu
Maneesh Gupta
Rakesh Roy