2dccb30f6f
SWDEV-485904 - propagate hsa_amd_vmem_address_free error to hip API
...
Unit_hipMemSetAccess_GrowVMM test fails with
HSA_STATUS_ERROR_RESOURCE_FREE silently
Change-Id: I7a78410e432de4a2e877062782abf8761645f392
[ROCm/clr commit: b498103f9b
2024-10-21 10:12:32 -04:00
d65afe707c
SWDEV-482751 - Use ocl-icd-devel package for SLES.
...
Change-Id: I30e6243d697dc984a42051c20e336551d50d8e94
[ROCm/clr commit: 1f55a707b4
2024-10-20 23:55:02 -04:00
4a2687a450
SWDEV-486602 - Fix Windows 32 bit build
...
Windows alings fields to 8 bytes even with 32bit builds.
Add BUG_CLR_SYSMEM_POOL to cotnrol sysmempool.
Change-Id: I8622aabc9f7391ed7dd8583b252ce9eb41d62293
[ROCm/clr commit: 6bb7d1afdc
2024-10-18 11:35:54 -04:00
830c72e286
SWDEV-490474 - Allow hipMallocManaged capturing only in relaxed mode
...
Change-Id: I02dccc6c45e39082ef925509a28bbe3c2a0fb7c6
[ROCm/clr commit: 6deecf1bfe
2024-10-18 04:52:01 -04:00
5c03a593e7
SWDEV-491375 - Optimize multithreaded dispatches
...
- Fix typo
Change-Id: If4c68455dcfa03fee18cb4720e8b5b438642703c
[ROCm/clr commit: 0f2342bc13
2024-10-17 17:02:23 -04:00
640ba99b75
SWDEV-486602 - Change SysmemPool implementation
...
- Remove the list of all chunks and use embedded chunk
information in each allocation. That simplifies Free() logic,
avoiding expensive loop if for some reason the number of
outstanding allocations significantly grew.
Change-Id: I9ea84d314320ce356ed24dd3180f262e2116c59b
[ROCm/clr commit: ad18146d8f
2024-10-17 12:39:39 -04:00
b8a5d8396d
SWDEV-468039,SWDEV-482579 - Enable FP8 SW Conversions on pre gfx940 archs
...
1) SW Conversions for ocp and fnuz are enabled on pre mi300 archs
2) for mi300 only fnuz is enabled
3) for gfx1200 only ocp is enabled
Change-Id: I90373752a2d15eff20d5deec874ed396ba4e1788
[ROCm/clr commit: e729f08704
2024-10-17 11:49:22 -04:00
0a03665a3f
SWDEV-491375 - Limit the SW batch size
...
Applications may submit commands withoout waits
for GPU. That causes a growth of SW unreleased commands.
Make sure runtime flushes SW queue, if it grows over some
threshold, controlled by DEBUG_CLR_MAX_BATCH_SIZE.
Change-Id: Ia4d85c24210ef91c394f638ab6b53b14323a0396
[ROCm/clr commit: 8657a77029
2024-10-17 10:53:57 -04:00
2a6792ec25
SWDEV-482751 - Depends on distro opencl icd loader
...
Since we don't distribute icd loader, we need to install distro icd loader.
Change-Id: I1ea86bcf7c642a034c53f71130b15de1fa27e31e
[ROCm/clr commit: df9ae754a4
2024-10-16 16:21:58 -04:00
b747d0986f
SWDEV-482751 - add distro path to find package AMD_ICD
...
Change-Id: I0d21f6ba6ade3ed932b134da503f639fd5d0d552
[ROCm/clr commit: ff306ce9d8
2024-10-14 15:27:34 -07:00
faea40cbb3
SWDEV-486602 - Optimize HSA callback performance
...
- Don't generate callbacks for HIP events
- Don't process profiling info in the callback for HIP events
- Wait for CPU status update of the submitted commands
every 50 calls. That will allow to drain the commands and
destroy HSA signals.
Change-Id: Ib601a350e7e7c2b6c6209a172385389baccf73a9
[ROCm/clr commit: 364dfb0ed1
2024-10-11 14:50:25 -04:00
043271a3e6
SWDEV-490323 - Fix validateMemAccess in hipMemset
...
Changed the validation to occur on the sub-object rather than the parent.
Change-Id: I87bf5ef3526d0db9304099ef9ac1a5494e9a01a9
[ROCm/clr commit: 5da72f9d52
2024-10-10 18:08:28 -04:00
170e45b879
SWDEV-472357 - support Rect copy with staging buffer for 2D & 3D memcpy in PAL
...
Change-Id: Ie32f3e5a6fa077f6b2db20fc1ab1e2e0da8344cb
[ROCm/clr commit: 41dc4545fc
2024-10-10 18:00:19 -04:00
55945b16c0
SWDEV-486510 - Delete hip::Function object, in case compiler passes duplicate hostFunction ptr.
...
Change-Id: Ic8714eb9022a0f2150b2ea5dc008cecd7a9fae27
[ROCm/clr commit: e7c0e06b5e
2024-10-10 12:45:58 -04:00
af3e9cb9e2
SWDEV-489823 - Fix hipStreamEndCapture leak when capture is invalidated
...
Change-Id: If8f5163d70e04d34a75fd0a7ba6c0a15ea59bb8b
[ROCm/clr commit: 6f2bad3998
2024-10-10 04:38:06 -04:00
7983801b0c
SWDEV-485866 - Return OOM if stream creation fails due to insufficient memory.
...
Change-Id: I4e57ecc81921bde274bb6a4e0890f0fc6a17955a
[ROCm/clr commit: 5ccc140e1b
2024-10-10 00:44:54 -04:00
da59165313
SWDEV-486137 - match behavior of int variants of hadd/uhadd/rhadd/urhadd
...
Match cases and handle cases where it can overflow.
Change-Id: I3d6f802686af230a622ef9891a844135ad3d1ae5
[ROCm/clr commit: b977101893
2024-10-09 13:47:33 -04:00
3a2d835272
SWDEV-486573 - Check the return type of commit memory.
...
Change-Id: Id158cd7a0dff37b382b858cf7113aa4cf326300a
[ROCm/clr commit: 74ebbe17e9
2024-10-09 05:10:03 -04:00
5d7b767788
SWDEV-479940 - Correct changelog in staging for 6.2.1
...
Change-Id: I3f35a85b9834841d27fa35abc52b9838d6f1c9e7
[ROCm/clr commit: d6bcabdc2c
2024-10-08 17:04:43 -04:00
d452c4ad28
SWDEV-483134 - Deprecate hipHostMalloc and hipHostFree APIs
...
Change-Id: I230ab2de2e4bdfdd9bfb0a3e59c6130a25b8b0cd
[ROCm/clr commit: 80043d38f4
2024-10-08 15:58:25 -04:00
815b9ffc36
SWDEV-489280 - Add missing hipGraphNodeSetParams API in dispatch table
...
Change-Id: I41dfd045fa4e29b49e605b8d583ec9f51dd6a6cc
[ROCm/clr commit: 15ecf834a1
2024-10-08 13:56:02 -04:00
566984676e
SWDEV-487988 - Reserve event flag in hip::Event.
...
Don't create new hip:Function if it is already registered.
Change-Id: I3ecd5d61146659be6ba434717b0f21d3fc04cfc9
[ROCm/clr commit: a6c5c6a95a
2024-10-08 05:29:32 -04:00
b31bf885a3
SWDEV-482692, SWDEV-485802, SWDEV-485489 - Handle refcounts owned by graph for user objects.
...
Change-Id: Ic739ab1ec5d3dc3143e3ae70f9591922bc0e3d9f
[ROCm/clr commit: e74ac6f580
2024-10-08 03:44:44 -04:00
3130b4639f
SWDEV-487905 - device_ptr_ is being removed and its amd:Memory obj is being deleted during ihipFree in hip::StatCO::removeFatBinary.
...
Change-Id: I89d9fdeb53dc4ce0699f1f445a28486917a36e72
[ROCm/clr commit: 164cbcc531
2024-10-08 03:38:15 -04:00
491d3828dd
SWDEV-482130 - Fix release of virtual mem obj
...
Change-Id: I893a8353aa1a25d00e36c8e601caf31cc0fc1f22
[ROCm/clr commit: 43fcac1739
2024-10-08 01:37:39 -04:00
57c5264937
SWDEV-483241 - Add a compile option to avoid including default hiprtc header
...
Change-Id: Ic23b41395588e6183abac36cb7543da02b0aba29
[ROCm/clr commit: 522ae8ead4
2024-10-07 07:56:29 -04:00
b9497ea70e
SWDEV-301667 - Enable ROCr logging
...
- Use AMD_LOG_LEVEL=5 to dump AQL packets in ROCr
Change-Id: I2c044a5304c4eaf3d3af20e62d1f54c98d4fbaa4
[ROCm/clr commit: e36666e536
2024-10-04 19:22:12 -04:00
375ed9d848
SWDEV-478065 - Revert "SWDEV-478065 - Embed host thread in shared_ptr"
...
This reverts commit 274fd2628fd3d0ca5fc6
2024-10-04 19:19:28 -04:00
169423798f
SWDEV-476542 - Unable to link to hipGraphExecGetFlags
...
Change-Id: I572baaeee31c6a73e533f9ef956bf111e9d2e688
[ROCm/clr commit: d29ebea7ac
2024-10-04 13:39:06 -04:00
5296c77138
SWDEV-301667 - Logging upgrades
...
- Use AMD_LOG_LEVEL_SIZE in MBs to set log file size truncation, by default its 2048 MB
Change-Id: Ia2f87e8c6b94148e30edfb602b279f93630817c3
[ROCm/clr commit: 35e03ea0d0
2024-10-04 13:26:25 -04:00
91b343c758
SWDEV-471422 - Free memory being double deducted on APUs due to system_total_alloced var holds local memory.
...
Change-Id: I3fbbc8f8aaa156881ff95cad6a4f82fd3df651d1
[ROCm/clr commit: 292842ad28
2024-10-04 04:49:20 -04:00
38ad03660a
SWDEV-467263 - Allow hipMalloc to use sys memory
...
PAL supports allocating from system memory once device memory is used up
or allocation is larger than the device memory.
Change-Id: Iccd3377e95a6cc6d23e45d4738a17af8b9ee32d7
[ROCm/clr commit: b07178618c
2024-10-03 11:14:08 -04:00
00cb623a67
SWDEV-488851 - Correctly remove the queue from the active set on windows
...
Change-Id: I4d21743ecf7a44636121f85566f898e62ff61e97
[ROCm/clr commit: 07bcc283f9
2024-10-02 12:06:59 +01:00
b758610a55
SWDEV-478708 - Remove forced wait of 10us in hipEventQuery
...
Change-Id: I868aae14311c3cdfc09aa03252ac324c4b79b864
[ROCm/clr commit: ade1954015
2024-10-01 06:27:42 -04:00
9f10e89d97
SWDEV-487903 - Fix for Empty Kernel Segfault in PAL
...
Change-Id: Ia1c19cf4ea24188cdb2d374b07f975f794e02dba
[ROCm/clr commit: 9da90fe848
2024-09-30 13:00:15 -04:00
bc3f3b09df
SWDEV-487905 - Managed vars are registered in __hipRegisterManagedVar however not freed.
...
Change-Id: Ic5a72ac4d64a9f7f5a3a7a88e1ed813e6dcc1f57
[ROCm/clr commit: 614b00c20b
2024-09-30 11:54:31 -04:00
493e9c7884
SWDEV-412099 - Fix CTS clFillImage sub-tests failures
...
Change-Id: I082476837c539e6ccf93cba6b1e97aae2509e65c
[ROCm/clr commit: 17c8b9f855
2024-09-30 11:13:52 -04:00
6a9b977def
SWDEV-478034 - Unable to link to hipGraphExecNodeSetParams
...
Change-Id: I0b6b8d1a4281ecda3c1789d8829ade9771aed741
[ROCm/clr commit: 939c788779
2024-09-30 02:13:43 -04:00
c0ceb1cf12
SWDEV-477324 - Capture Memcpy1D pinned H2D D2H
...
Change-Id: I1f4744f20a9caeed005ec68da44e5fde737e09f7
[ROCm/clr commit: 742b0210d3
2024-09-30 01:01:30 -04:00
887b11894b
SWDEV-482086 - Fix hipGraphInstantiate leak
...
* In a scenario where kernel is launched with hipExtLaunchKernelGGL and stop event is used, hipGraphInstantiate leaks. Since stop event is used, profiling is enabled and Timestamp (ReferencedCountedObject) is created, but it doesn't get released.
* The idea behind this solution is that profiling should be disabled when command is captured, hence the timestamp should not be created. Because information about capturing isn't available when kernel command is created, packet capturing state is used to determine whether to create a timestamp or not.
Change-Id: Ia23adac4592ded4fb5e236acf99e12e729f63692
[ROCm/clr commit: da5f1a6146
2024-09-29 11:36:53 -04:00
98539759f1
SWDEV-483436 - User spt stream as def with -fgpu-default-stream=per-thread for hipMemsetAsync.
...
Change-Id: Ia85c2b4c40fc9250754d3b64fb9fd1c615362572
[ROCm/clr commit: d6193a2f23
2024-09-29 01:42:33 -04:00
da48c3b185
[SWDEV-467733] - Add Param checking for SetCacheConfig APIs
...
Change-Id: I9e777fa0fae6791ebab539e49346e6956a6ff196
[ROCm/clr commit: 0d20383ef9
2024-09-27 11:32:58 -04:00
43a1d720d5
Fix HIP API trace versioning
...
Change-Id: I33f2be4668c96e2225d4ca9a253e61ec2dc65102
[ROCm/clr commit: 07c9c7fe56
2024-09-25 10:32:14 -04:00
98a1b3b42a
SWDEV-486816 - RenderOpDispatch usage in pal client
...
Change-Id: I11cae3e625b287b998c9500c547efdacf1034a2b
[ROCm/clr commit: 7a288ea8bf
2024-09-24 14:28:16 -04:00
1157776b30
SWDEV-479260,SWDEV-483599 - Check griddim Y,Z <= 65536
...
Gfx12 has 16 bits for grid dim Y/Z. Detect gfxIp and return error if dim y/z > 16 bits
Change-Id: I43dd14affc9e4073d0b1232e7523967f0180fa31
[ROCm/clr commit: 0a918c8f96
2024-09-23 11:36:13 -04:00
0a850d8725
SWDEV-466747 - call device sync once while unregistering
...
Basically embed hipDeviceSync in std::call_once.
Change-Id: I29ca926d61ed80e21acba5c388a8256d913487e4
[ROCm/clr commit: f8beeede22
2024-09-23 08:00:10 -04:00
f8fc11c2d8
SWDEV-483586 - Unblock staging H2D transfers
...
Although unpinned copies require synchronizations
in HIP, runtime can avoid syncs for H2D copies with
a staging buffer
Change-Id: If2203c6bc0cbd89742823688dc8e89e9acd873b2
[ROCm/clr commit: 29cc678d8d
2024-09-21 10:25:27 -04:00
e87bf2096d
SWDEV-485179 - Revert "SWDEV-459254 - Overwrite cacheline size to 256 for gfx12, as it is used for kernarg alignment."
...
This reverts commit 1f63650bf96e01e48f879aa58b80e2130dd4a567.
Reason for revert: <INSERT REASONING HERE>
Change-Id: I6d7ed87c09d9b77116548dce1f30ac4711c2c09d
[ROCm/clr commit: 2d1c6ee23e
2024-09-20 11:33:34 -04:00
f278ac9ea0
SWDEV-485904 - Fix virtual,physical mem obj leaks
...
Change-Id: Ie0456b5dcfec206ae54a6aabfc2a15a620cac693
[ROCm/clr commit: 870842201d
2024-09-19 23:04:20 -04:00
d5c396e820
SWDEV-301667 - Improve logging
...
Change-Id: I3fa06791b7ac73d84b8a9586e6b3435fa8858d25
[ROCm/clr commit: 8c84a20b01
2024-09-19 15:09:03 -04:00
Anusha GodavarthySurya
Jaydeep Patel
German Andryeyev
Vladana Stojiljkovic
Saleel Kudchadker
Rahul Manocha
Alex Xie
Ajay
Ioannis Assiouras
Todd tiantuo Li
kjayapra-amd
Jatin Chaudhary
Julia Jiang
Satyanvesh Dittakavi
Branislav Brzak
pghafari
Jonathan R. Madsen
Maneesh Gupta