intel/compute-runtime
1
0
Fork 0
mirror of https://github.com/intel/compute-runtime.git synced 2025-12-22 10:17:01 +08:00
Code Issues Packages Projects Releases Wiki Activity
20,778 Commits 42 Branches 655 Tags
98fcaf4cebc7d2bca206b695225ed1949f437f7d
Commit Graph

1679 Commits

Author SHA1 Message Date
Chandio, Bibrak Qamar
cf21cbc910 feature: Complete madvise for System Allocator 
Related-To: NEO-13715

Provides implmentation of getAtomicAccessAttribute using system
allocator.
Completes the implementation of the madvise feature using the
new IOCTLs in xe_drm header.

Signed-off-by: Chandio, Bibrak Qamar <bibrak.qamar.chandio@intel.com>
 2025-09-23 16:00:35 +02:00
Szymon Morek
5714c3858f fix: increase staging threshold for misaligned ptrs 
Related-To: NEO-16118

Signed-off-by: Szymon Morek <szymon.morek@intel.com>
 2025-09-22 12:46:58 +02:00
Jaroslaw Warchulski
195bf66a49 refactor: fix typos 
Signed-off-by: Jaroslaw Warchulski <jaroslaw.warchulski@intel.com>
 2025-09-22 12:46:41 +02:00
Grochowski, Stanislaw
93d98c528d fix: check interrupt support when allocating and releasing 
Related-To: HSD-18043299783


Signed-off-by: Grochowski, Stanislaw <stanislaw.grochowski@intel.com>
 2025-09-19 10:36:18 +02:00
Fabian Zwoliński
8cf8fe4f5a fix: use const auto& to avoid unnecessary copies 
Signed-off-by: Fabian Zwoliński <fabian.zwolinski@intel.com>
 2025-09-15 17:48:57 +02:00
Mateusz Jablonski
92316c8cd6 fix: don't setup notifyAubCapture callback when not needed 
Related-To: NEO-11080
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
 2025-09-12 11:28:37 +02:00
Filip Hazubski
0c2a8ef30b refactor: fix typo 
Signed-off-by: Filip Hazubski <filip.hazubski@intel.com>
 2025-09-11 15:55:38 +02:00
Compute-Runtime-Validation
eba3a2d685 Revert "feature: enable defer backing by default" 
This reverts commit 34d4e526e9.

Signed-off-by: Compute-Runtime-Validation <compute-runtime-validation@intel.com>
 2025-09-11 05:54:53 +02:00
Narendra Bagria
34d4e526e9 feature: enable defer backing by default 
Related-To: NEO-13403

Signed-off-by: Bellekallu Rajkiran <bellekallu.rajkiran@intel.com>
Signed-off-by: Narendra Bagria <narendra.bagria@intel.com>
 2025-09-10 12:41:30 +02:00
John Falkowski
1f281a01ee refactor: Add product helper getSharedSystemPatIndex 
Resolves: NEO-15930

Signed-off-by: John Falkowski <john.falkowski@intel.com>
 2025-09-08 17:17:35 +02:00
Szymon Morek
03357b6089 performance: adjust size threshold for staging on linux 
Related-To: NEO-15973

Signed-off-by: Szymon Morek <szymon.morek@intel.com>
 2025-09-05 18:09:49 +02:00
Szymon Morek
90be39c9c0 performance: enable low latency hint on Xe 
Related-To: NEO-14708

Signed-off-by: Szymon Morek <szymon.morek@intel.com>
 2025-09-01 13:34:29 +02:00
Aviral Nigam
c1c1f1f0af fix: Fix number of channels and bus width calculation 
Take number of memory channels from telemetry instead
of using hard coded value and calculate memory bus width
based on the formula

busWidth = numChannels * 32

Related-To: NEO-15629

Signed-off-by: Aviral Nigam <aviral.nigam@intel.com>
 2025-08-29 16:29:43 +02:00
Brandon Yates
4e8d340cf1 fix: Only primary drm context should enable eu debug 
Related-to: HSD-18043158665

Signed-off-by: Brandon Yates <brandon.yates@intel.com>
 2025-08-29 01:22:50 +02:00
Szymon Morek
228da24b38 performance: align alloc size to 2MB on XeKMD/iGPU 
Related-To: NEO-15905

Signed-off-by: Szymon Morek <szymon.morek@intel.com>
 2025-08-25 15:37:59 +02:00
Radoslaw Jablonski
393604ef80 refactor: numThreadsPerEu 
Signed-off-by: Radoslaw Jablonski <radoslaw.jablonski@intel.com>
 2025-08-25 11:31:49 +02:00
Szymon Morek
d17255fe18 performance: add low latency hint on xe 
Related-To: NEO-14708

Signed-off-by: Szymon Morek <szymon.morek@intel.com>
 2025-08-21 13:09:38 +02:00
Filip Hazubski
40ad3cf6f1 feature: Query L3 bank group info from device blob 
Related-To: HSD-14025718322, HSD-18041858866

Signed-off-by: Filip Hazubski <filip.hazubski@intel.com>
 2025-08-19 14:46:19 +02:00
Neil R. Spruit
43ebc17e17 feature: Support for opaque IPC handles on Windows and Linux 
- Added support for creating and managing opaque IPC NT handles in the
WDDM layer.
- Introduced a new flag `shareableWithoutNTHandle` to indicate if memory
can be shared without an NT handle.
- Updated the `isShareableMemory` method to accommodate this new flag.
- Added debug variable EnableShareableWithoutNTHandle to control the
behavior of sharing memory without NT handles until requested.
- Updated Linux path to enable sharing DMA Buf FDs between processes
for use in pidfd_getfd
- Updated getfd and get IPC Handle helper functions to support opaque or
previous versions

Related-To: NEO-15345 , NEO-15346 , NEO-15347, NEO-10380

Signed-off-by: Neil R. Spruit <neil.r.spruit@intel.com>
 2025-08-18 16:23:44 +02:00
Compute-Runtime-Validation
0855219087 Revert "feature: Support for opaque IPC handles on Windows and Linux" 
This reverts commit 46b1b2783b.

Signed-off-by: Compute-Runtime-Validation <compute-runtime-validation@intel.com>
 2025-08-15 02:22:02 +02:00
Brandon Yates
7c612a2a01 fix: Set MaxSlicesSupported to true max on linux 
Related-to: HSD-18042764574

Signed-off-by: Brandon Yates <brandon.yates@intel.com>
 2025-08-14 08:51:55 +02:00
Neil R. Spruit
46b1b2783b feature: Support for opaque IPC handles on Windows and Linux 
- Added support for creating and managing opaque IPC NT handles in the
WDDM layer.
- Introduced a new flag `shareableWithoutNTHandle` to indicate if memory
can be shared without an NT handle.
- Updated the `isShareableMemory` method to accommodate this new flag.
- Added debug variable EnableShareableWithoutNTHandle to control the
behavior of sharing memory without NT handles until requested.
- Updated Linux path to enable sharing DMA Buf FDs between processes
for use in pidfd_getfd

Related-To: NEO-15345 , NEO-15346 , NEO-15347, NEO-10380

Signed-off-by: Neil R. Spruit <neil.r.spruit@intel.com>
 2025-08-14 08:33:56 +02:00
Jakub Nowacki
c4ab767797 fix: add missing NonCopyableAndNonMovable asserts 
Related-To: NEO-15630

Signed-off-by: Jakub Nowacki <jakub.nowacki@intel.com>
 2025-08-13 13:13:05 +02:00
Chandio, Bibrak Qamar
6acfcb83b9 Revert "Two commits related to pStart feature" 
Related-To: NEO-15156, GSD-9939

Revert "test: fix minor errors"

This reverts commit 712dc8221e.

Revert "feature: Support for pStart"

This reverts commit 47caeda487.

Signed-off-by: Chandio, Bibrak Qamar <bibrak.qamar.chandio@intel.com>
 2025-08-13 07:10:40 +02:00
Radoslaw Jablonski
59c3b06eea refactor: xe topology 
Signed-off-by: Radoslaw Jablonski <radoslaw.jablonski@intel.com>
 2025-08-08 15:31:13 +02:00
Radoslaw Jablonski
c2d64cae6a refactor: queryDeviceCaps ioctl 
Signed-off-by: Radoslaw Jablonski <radoslaw.jablonski@intel.com>
 2025-08-08 14:43:23 +02:00
Compute-Runtime-Validation
b7472fa76f Revert "refactor: queryDeviceCaps ioctl" 
This reverts commit 3117d2d2cb.

Signed-off-by: Compute-Runtime-Validation <compute-runtime-validation@intel.com>
 2025-08-08 07:05:24 +02:00
Radoslaw Jablonski
e190e1b8ec fix: use relative subslice mapping indices in Xe DRM 
This aligns behavior with i915 and L0 spec.

Signed-off-by: Radoslaw Jablonski <radoslaw.jablonski@intel.com>
 2025-08-07 15:58:48 +02:00
Radoslaw Jablonski
9ec004ba59 fix: don't resolve MaxEuPerSubSlice from Xe topology 
All Xe platforms should provide MaxEuPerSubslice through System Info,
so there is no need to resolve it from topology.

Signed-off-by: Radoslaw Jablonski <radoslaw.jablonski@intel.com>
 2025-08-07 15:48:19 +02:00
Compute-Runtime-Validation
b40830c1e6 Revert "fix: MaxSubSlicesSupported should be set to total non-fused count" 
This reverts commit ee0022d234.

Signed-off-by: Compute-Runtime-Validation <compute-runtime-validation@intel.com>
 2025-08-07 02:15:20 +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
Radoslaw Jablonski
3117d2d2cb refactor: queryDeviceCaps ioctl 
Signed-off-by: Radoslaw Jablonski <radoslaw.jablonski@intel.com>
 2025-08-06 12:22:03 +02:00
Brandon Yates
ee0022d234 fix: MaxSubSlicesSupported should be set to total non-fused count 
Related-to: HSD-18042764574

Signed-off-by: Brandon Yates <brandon.yates@intel.com>
 2025-08-06 12:09:36 +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
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
Maciej Plewka
2c287c7c3b fix: allow to use bigger alignment than 2mb on linux 
Related-To: NEO-14082
Signed-off-by: Maciej Plewka <maciej.plewka@intel.com>
 2025-08-04 11:38:50 +02:00
Slawomir Milczarek
fa4fd26be2 fix: Add missing registration of allocation from shared handle to smem 
Related-To: NEO-12952

Fixes debug assert triggered in unregisterAllocation()
for GPU_TIMESTAMP_DEVICE_BUFFER imported from shared handle:Y
DEBUG_BREAK_IF(allocation->getUnderlyingBufferSize() > sysMemAllocsSize);

Signed-off-by: Slawomir Milczarek <slawomir.milczarek@intel.com>
 2025-08-04 09:00:11 +02:00
John Falkowski
a45f47fad1 refactor: Disable Shared System USM if CPU addr range unacceptable 
Related-To: NEO-15680

Signed-off-by: John Falkowski <john.falkowski@intel.com>
 2025-08-01 07:50:46 +02:00
Neil R. Spruit
b4f37f3a98 fix: Allocate debug surface and isa as BO instead of userptr 
Related-To: NEO-14269

Signed-off-by: Neil R. Spruit <neil.r.spruit@intel.com>
 2025-07-31 20:24:54 +02:00
Radoslaw Jablonski
2a236d079c feature: add device caps helpers 
Related-To: NEO-13039

Signed-off-by: Radoslaw Jablonski <radoslaw.jablonski@intel.com>
 2025-07-30 15:03:38 +02:00
John Falkowski
5aed2bcdce refactor: Use min GPU/CPU address range for system allocator vmBind 
Related-To:  NEO-15680

Signed-off-by: John Falkowski <john.falkowski@intel.com>
 2025-07-29 17:08:00 +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
Pawel Szymichowski
211cd3dd5d fix: Remove RTLD_DEEPBIND from dlopen call 
Resolves: NEO-15651

Signed-off-by: Pawel Szymichowski <pawel.szymichowski@intel.com>
 2025-07-28 11:02:13 +02:00
Bartosz Dunajski
1a51b7a214 refactor: remove not needed method 
Signed-off-by: Bartosz Dunajski <bartosz.dunajski@intel.com>
 2025-07-24 17:24:46 +02:00
Eric Mortensen
69f2cda2fa fix: add error checks & return to unMapVirtualMem 
Related-To: NEO-8566

Signed-off-by: Eric Mortensen <eric.mortensen@intel.com>
 2025-07-17 23:53:19 +02:00
John Falkowski
d3564fab68 refactor: Use common default pat index for vmUnbind 
Related-To: NEO-13697

Signed-off-by: John Falkowski <john.falkowski@intel.com>
 2025-07-17 10:41:38 +02:00
Maciej Plewka
12ab2131d8 fix: use alignment when allocate by linux kmd 
Related-To: NEO-14082
Signed-off-by: Maciej Plewka <maciej.plewka@intel.com>
 2025-07-14 09:29:41 +02:00
Fabian Zwoliński
0bca8782c2 fix: resolve SVM allocation size mismatch for 2MB pages 
Related-To: NEO-12287
Signed-off-by: Fabian Zwoliński <fabian.zwolinski@intel.com>
 2025-07-08 13:32:01 +02:00
Aleksandra Nizio
6975f7bb8d fix: Disable compiler cache if any IGC_ env is set 
Related-To: NEO-12721
Signed-off-by: Aleksandra Nizio <aleksandra.nizio@intel.com>
 2025-07-07 09:55:49 +02:00
Krzysztof Sprzaczkowski
5a057bbd46 feature: Add PCI barrier mmap logic for xe 
Related-To: NEO-13081
Signed-off-by: Filip Hazubski <filip.hazubski@intel.com>
Signed-off-by: Krzysztof Sprzaczkowski <krzysztof.sprzaczkowski@intel.com>
 2025-07-01 16:32:54 +02:00