intel/compute-runtime
1
0
Fork 0
mirror of https://github.com/intel/compute-runtime.git synced 2025-12-30 01:35:20 +08:00
Code Issues Packages Projects Releases Wiki Activity
20,216 Commits 42 Branches 655 Tags
c2d64cae6a4637d6d9cc9d786580154aee4e2643
Commit Graph

3473 Commits

Author SHA1 Message Date
Dominik Dabek
77470acf7a fix: l0, tag update on mem copy ext host ptr 
Require tag update on mem copy with external host ptr.
Without this, temporary allocation might not be cleaned before next copy
operation.
If a second copy operation is passed same ptr that has been reallocated,
there will be a pagefault.

Related-To: NEO-15663

Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
 2025-08-08 12:42:53 +02:00
Michal Mrozek
52e413f69e performance: reuse allocation data if available 
Signed-off-by: Michal Mrozek <michal.mrozek@intel.com>
 2025-08-08 09:36:35 +02:00
Michal Mrozek
469aeb870c refactor: remove unused code 
Signed-off-by: Michal Mrozek <michal.mrozek@intel.com>
 2025-08-08 08:07:49 +02:00
Michal Mrozek
f85e4935e2 performance: reuse allocData if available 
Signed-off-by: Michal Mrozek <michal.mrozek@intel.com>
 2025-08-07 18:26:36 +02:00
Zbigniew Zdanowicz
da2fc7a586 feature: add noop support to patch preamble 
Related-To: NEO-15376

Signed-off-by: Zbigniew Zdanowicz <zbigniew.zdanowicz@intel.com>
 2025-08-07 16:23:43 +02:00
Kamil Kopryk
1ccd72bfc9 refactor: simplify logic 
Signed-off-by: Kamil Kopryk <kamil.kopryk@intel.com>
 2025-08-07 14:01:14 +02:00
Szymon Morek
7b8e986ebe fix: flush caches after post sync write 
Related-To: NEO-14360

Flush L2 cache if post sync write from pipe control is being cached.
Otherwise, host will wait for data which is cached on GPU.

Signed-off-by: Szymon Morek <szymon.morek@intel.com>
 2025-08-07 07:33:19 +02:00
Michel Schanen
e999c352ca fix: c->c++ inttypes header 
Signed-off-by: Michel Schanen <michel.schanen@gmail.com>
 2025-08-06 19:23:57 +02:00
Mateusz Jablonski
b5ea7274b8 fix: remove const from arguments in append with arguments 
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
 2025-08-06 16:30:34 +02:00
Bartosz Dunajski
b95ba7cf05 fix: add poll for aub completion on queue synchronize 
Signed-off-by: Bartosz Dunajski <bartosz.dunajski@intel.com>
 2025-08-06 14:03:30 +02:00
Maciej Bielski
8ec774d3f0 refactor: remove surplus if() statement 
The `appendSynchronizedDispatchInitializationSection()` checks for the
same condition internally.

Related-To: NEO-15374
Signed-off-by: Maciej Bielski <maciej.bielski@intel.com>
 2025-08-06 13:02:29 +02:00
Chandio, Bibrak Qamar
47caeda487 feature: Support for pStart 
Related-To: NEO-15156, GSD-9939

Support for start address hint in zeVirtualMemReserve.

If it fails to find pStart then it defaults to the base line
allocateWithCustomAlignment(...)

Signed-off-by: Chandio, Bibrak Qamar <bibrak.qamar.chandio@intel.com>
 2025-08-06 12:40:29 +02:00
Lukasz Jobczyk
4daf9dcb83 fix: Wait for CB TS node only when required 
In case of optimized CB event, synchronize TS completion only when just
TS node is available, otherwise, if both conuter and TS are available,
synchronize by polling for in order counter. Such situation occurs when
appending non walker operation on platform eligible for optimized Cb
events without dc flush.

Resolves: HSD-18042863956

Signed-off-by: Lukasz Jobczyk <lukasz.jobczyk@intel.com>
 2025-08-06 11:45:13 +02:00
Zbigniew Zdanowicz
14b9198949 feature: sum total required noop space of command list patchlist 
Related-To: NEO-15376

Signed-off-by: Zbigniew Zdanowicz <zbigniew.zdanowicz@intel.com>
 2025-08-06 11:42:55 +02:00
Compute-Runtime-Validation
c9456cda14 Revert "feature: Support for pStart" 
This reverts commit a50c0dbecf.

Signed-off-by: Compute-Runtime-Validation <compute-runtime-validation@intel.com>
 2025-08-06 02:26:13 +02:00
Zbigniew Zdanowicz
b0e9267e62 feature: add support encodng front end command to patch preamble 
Related-To: NEO-15376

Signed-off-by: Zbigniew Zdanowicz <zbigniew.zdanowicz@intel.com>
 2025-08-05 12:58:29 +02:00
Bartosz Dunajski
cc01c35501 feature: use dedicated cmd lists for bcs split 
Related-To: NEO-14557

Signed-off-by: Bartosz Dunajski <bartosz.dunajski@intel.com>
 2025-08-05 10:49:11 +02:00
Chandio, Bibrak Qamar
a50c0dbecf feature: Support for pStart 
Related-To: NEO-15156, GSD-9939

Support for start address hint in zeVirtualMemReserve.

If it fails to find pStart then it defaults to the base line
allocateWithCustomAlignment(...)

Signed-off-by: Chandio, Bibrak Qamar <bibrak.qamar.chandio@intel.com>
 2025-08-05 10:14:01 +02:00
Zbigniew Zdanowicz
841267ecbd feature: save command buffer gpu address for front end command in command list 
Related-To: NEO-15376

Signed-off-by: Zbigniew Zdanowicz <zbigniew.zdanowicz@intel.com>
 2025-08-05 09:59:08 +02:00
Mateusz Jablonski
ee5cd678ea fix: correct naming of zer functions 
zerDriverGetDefaultContext -> zerGetDefaultContext
zerDriverGetLastErrorDescription -> zerGetLastErrorDesription
zerDeviceTranslateToIdentifier -> zerTranslateDeviceHandleToIdentifier
zerIdentifierTranslateToDeviceHandle -> zerTranslateIdentifierToDeviceHandle

Related-To: NEO-14560

Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
 2025-08-04 15:17:04 +02:00
Mateusz Hoppe
b7a73653a2 fix: set sync buffer ptr in implicit args 
Related-To: NEO-15737

Signed-off-by: Mateusz Hoppe <mateusz.hoppe@intel.com>
 2025-08-04 14:22:46 +02:00
Mateusz Jablonski
dbc00d623a refactor: allow mocking event completion timeout in ULT 
fix timing issue in L0 ULT

Related-To: NEO-15354
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
 2025-08-04 13:00:21 +02:00
Matias Cabral
1b7a26996e refactor: renaming of metrics calcOp related structures 
Resolves: NEO-15139

Signed-off-by: Matias Cabral <matias.a.cabral@intel.com>
 2025-08-04 11:09:28 +02:00
Compute-Runtime-Validation
566b6551a5 Revert "fix: correct naming of zer functions" 
This reverts commit 91a9ccaebe.

Signed-off-by: Compute-Runtime-Validation <compute-runtime-validation@intel.com>
 2025-08-02 02:09:08 +02:00
Zbigniew Zdanowicz
01889c97a5 feature: add patch preamble to level zero queue handling bb_start commands 
Related-To: NEO-15376

Signed-off-by: Zbigniew Zdanowicz <zbigniew.zdanowicz@intel.com>
 2025-08-01 14:50:56 +02:00
Mateusz Jablonski
91a9ccaebe fix: correct naming of zer functions 
zerDriverGetDefaultContext -> zerGetDefaultContext
zerDriverGetLastErrorDescription -> zerGetLastErrorDesription
zerDeviceTranslateToIdentifier -> zerTranslateDeviceHandleToIdentifier
zerIdentifierTranslateToDeviceHandle -> zerTranslateIdentifierToDeviceHandle

Related-To: NEO-14560

Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
 2025-08-01 14:25:20 +02:00
Maciej Bielski
a0231c2d22 feature: allow for cloning KernelImp instances 
From the `cloneWithStateOverride()` perspecitve the `KernelImp` has
two logical parts handled differently:
- `KernelMutableState`: separate for both regular and cloned instances,
may be optionally overriden when cloning
- rest: separate for regular instances but shared with all clones
(shallow copy), the original instance owns these resources

Related-To: NEO-15374
Signed-off-by: Maciej Bielski <maciej.bielski@intel.com>
 2025-08-01 11:51:45 +02:00
Joshua Santosh Ranjan
2fbe207328 refactor: move subdevice handling to metrics 
Related-To: NEO-15516


Signed-off-by: Joshua Santosh Ranjan <joshua.santosh.ranjan@intel.com>
 2025-08-01 11:03:23 +02:00
Michal Mrozek
9522320275 performance: improve driver getter 
- make it final to optimize virtual calls
- move to header to allow to inline

Signed-off-by: Michal Mrozek <michal.mrozek@intel.com>
 2025-08-01 08:33:18 +02:00
Mateusz Jablonski
abb00a5ce3 fix: ensure thread-safety in zeDeviceSynchronize 
get task count and flush stamp within single critical section

Related-To: NEO-14560
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
 2025-07-31 15:59:57 +02:00
Jakub Nowacki
a23647da15 feature: expose event_query_timestamps extension 
Related-To: NEO-11561

Signed-off-by: Jakub Nowacki <jakub.nowacki@intel.com>
 2025-07-31 10:51:26 +02:00
Jakub Nowacki
781b6bbee2 feature: expose sub_allocations extension 
Related-To: NEO-11561

Signed-off-by: Jakub Nowacki <jakub.nowacki@intel.com>
 2025-07-30 14:31:09 +02:00
Joshua Santosh Ranjan
280239ebec feature: add support to query metrics hw buffer size 
Related-To: NEO-13439

Signed-off-by: Joshua Santosh Ranjan <joshua.santosh.ranjan@intel.com>
 2025-07-30 13:19:19 +02:00
Compute-Runtime-Validation
705dc02f06 Revert "feature: Use zesInit on windows always" 
This reverts commit 99ee605410.

Signed-off-by: Compute-Runtime-Validation <compute-runtime-validation@intel.com>
 2025-07-30 04:38:02 +02:00
Bartosz Dunajski
d693daf339 fix: fallback copy offload to compute if max copy fill pattern exceeded 
Related-To: GSD-11436

Signed-off-by: Bartosz Dunajski <bartosz.dunajski@intel.com>
 2025-07-29 17:00:50 +02:00
Jakub Nowacki
b522da6532 feature: expose eu_count l0 extension 
Related-To: NEO-11561

Signed-off-by: Jakub Nowacki <jakub.nowacki@intel.com>
 2025-07-29 14:08:45 +02:00
shubham kumar
99ee605410 feature: Use zesInit on windows always 
Related-To: NEO-15060

Signed-off-by: shubham kumar <shubham.kumar@intel.com>
 2025-07-29 12:34:02 +02:00
Matias Cabral
b7f9d55118 feature: support excluded metrics from calc op 
Resolves: NEO-15492

Signed-off-by: Matias Cabral <matias.a.cabral@intel.com>
 2025-07-29 05:16:29 +02:00
Mateusz Jablonski
1c1028a57c refactor: remove not needed constant 
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
 2025-07-28 17:31:46 +02:00
Kindracki, Jakub Tomasz
a73d5724e3 feature: Return insufficient permissions when driver device cannot be loaded 
Return ZE_RESULT_ERROR_INSUFFICIENT_PERMISSIONS when user doesn't have
permissions to access the GPU driver device file.

Related-To: NEO-8682
Signed-off-by: Kindracki, Jakub Tomasz <jakub.tomasz.kindracki@intel.com>
 2025-07-28 16:31:07 +02:00
Mateusz Jablonski
05cf05e10e refactor: remove redundant tracking of media sampler programming 
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
 2025-07-28 15:17:46 +02:00
Bartosz Dunajski
42fbe388fb refactor: simplify bcs split code 
Related-To: NEO-14557

Signed-off-by: Bartosz Dunajski <bartosz.dunajski@intel.com>
 2025-07-28 13:25:56 +02:00
Maciej Bielski
c396367411 feature: Record&Replay support for appendLaunchKernel 
Related-To: NEO-15374
Signed-off-by: Maciej Bielski <maciej.bielski@intel.com>
 2025-07-28 12:19:30 +02:00
Bartosz Dunajski
8156c9d5aa feature: adjust required engines count for aggregated split 
Related-To: NEO-14557

Signed-off-by: Bartosz Dunajski <bartosz.dunajski@intel.com>
 2025-07-25 15:51:30 +02:00
Mateusz Hoppe
be483491bf feature: extract indirect access buffer major version from ELF's notes 
- use version from ELF to create correct layout of implcit args
Related-To: NEO-15574

Signed-off-by: Mateusz Hoppe <mateusz.hoppe@intel.com>
 2025-07-25 14:50:49 +02:00
Maciej Bielski
a755ad131d feature: implement KernelMutableState operator= 
Related-To: NEO-15374
Signed-off-by: Maciej Bielski <maciej.bielski@intel.com>
 2025-07-25 14:37:04 +02:00
Chandio, Bibrak Qamar
aeb85c202c fix: Add more return types to parseUSMType 
Related-To: NEO-15604, GSD-11403

Signed-off-by: Chandio, Bibrak Qamar <bibrak.qamar.chandio@intel.com>
 2025-07-24 21:56:21 +02:00
Mateusz Jablonski
0a347a2d8b refactor: remove not needed code related to vme usage 
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
 2025-07-24 17:54:51 +02:00
Zbigniew Zdanowicz
cdc7be1518 performance: limit number of queries when getting usm allocation data 
Related-To: NEO-13916

Signed-off-by: Zbigniew Zdanowicz <zbigniew.zdanowicz@intel.com>
 2025-07-24 17:35:33 +02:00
Zbigniew Zdanowicz
f4d238b281 performance: move main mutation object instead of copy and destruct 
Related-To: NEO-13916

Signed-off-by: Zbigniew Zdanowicz <zbigniew.zdanowicz@intel.com>
 2025-07-24 15:54:55 +02:00