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
17,457 Commits 42 Branches 655 Tags
f117f4c9388b84e0769a7972a3cd9ff3f5ee815b
Commit Graph

6062 Commits

Author SHA1 Message Date
Matias Cabral
6ddb550e05 feature: improve metrics debug messages 
Resolves: NEO-12640

Signed-off-by: Matias Cabral <matias.a.cabral@intel.com>
 2024-10-07 17:58:41 +02:00
Szymon Morek
068ceee8bd performance: improve iterating over residency allocations 
Related-To: NEO-12698,NEO-12746,NEO-12729

Signed-off-by: Szymon Morek <szymon.morek@intel.com>
 2024-10-07 16:12:22 +02:00
Katarzyna Cencelewska
42ca656edb fix: change logic to calculate available thread count 
don't use magic number, value depend on grf size

Related-To: HSD-18039369782
Signed-off-by: Katarzyna Cencelewska <katarzyna.cencelewska@intel.com>
 2024-10-07 15:46:33 +02:00
Szymon Morek
b520c64775 performance: Enable direct submission on DG2 Windows 
Signed-off-by: Szymon Morek <szymon.morek@intel.com>
 2024-10-07 15:38:20 +02:00
Dominik Dabek
6d6c4267b3 performance: enable indirect detection 
Platforms other than PVC require detection version >= 4
Vector compiler kernels require >= 6

Related-To: NEO-12491

Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
 2024-10-07 14:38:22 +02:00
Zbigniew Zdanowicz
cb3b2134ab refactor: unify programming of preferred slm size 4/n 
- remove xe hpg encode preferred slm size
- add dg2/mtl/arl release helper preferred slm array
- drop dg2 preproduction stepping values for preferred slm size
- remove obsolete product helper method

Related-To: NEO-12639

Signed-off-by: Zbigniew Zdanowicz <zbigniew.zdanowicz@intel.com>
 2024-10-07 14:28:23 +02:00
Maciej Plewka
73e4b6ae7c fix: remove w/a which disables wmtp in kernels with ray tracing 
Related-To: NEO-12872
Signed-off-by: Maciej Plewka <maciej.plewka@intel.com>
 2024-10-07 14:28:08 +02:00
Zbigniew Zdanowicz
d6016e1b91 refactor: unify programming of preferred slm size 3/n 
- add shared implementation to encode preferred slm size
- add pvc release helper preferred slm array
- drop pvc preproduction steppings values for preferred slm size
- remove obsolete product helper method

Related-To: NEO-12639

Signed-off-by: Zbigniew Zdanowicz <zbigniew.zdanowicz@intel.com>
 2024-10-07 12:10:27 +02:00
Dominik Dabek
3685852ce0 Revert "performance: device usm sets localOnly... 
Required"

This reverts commit a479afdbc8.

Related-To: NEO-12879

Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
 2024-10-07 11:59:37 +02:00
Bartosz Dunajski
9a280892f8 refactor: add timestamps to xe logs 
Signed-off-by: Bartosz Dunajski <bartosz.dunajski@intel.com>
 2024-10-07 11:42:55 +02:00
Szymon Morek
a915ef4b7b fix: Don't program redundant paging fence semaphores 
Related-To: NEO-12197

Don't program semaphore to wait for paging fence if it was
already programmed with the same value

Signed-off-by: Szymon Morek <szymon.morek@intel.com>
 2024-10-07 11:42:06 +02:00
Andrzej Koska
490b474ce7 fix: using releaseHelper to determine MSAA enablement 
Related-To: NEO-12587

Signed-off-by: Andrzej Koska <andrzej.koska@intel.com>
 2024-10-07 11:26:12 +02:00
Maciej Plewka
e73e865b61 fix: If dummy exec failed call evict from outside the lock 
Related-To: NEO-12849, HSD-18040201135
Signed-off-by: Maciej Plewka <maciej.plewka@intel.com>
 2024-10-07 10:17:16 +02:00
Mateusz Jablonski
bbffbd16a0 refactor: remove not needed code 
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
 2024-10-04 11:59:45 +02:00
Zbigniew Zdanowicz
8687a88d06 refactor: unify programming of preferred slm size 2/n 
- add interface that will provide all preferred slm sizes for all releases

Related-To: NEO-12639

Signed-off-by: Zbigniew Zdanowicz <zbigniew.zdanowicz@intel.com>
 2024-10-04 11:50:44 +02:00
Compute-Runtime-Validation
43cf7abf65 Revert "fix: reduce preemption surface size in aub tests" 
This reverts commit e56b2b257f.

Signed-off-by: Compute-Runtime-Validation <compute-runtime-validation@intel.com>
 2024-10-04 02:45:42 +02:00
Brandon Yates
013c49db02 fix: change getBindlessDebugSipKernel to also get heapless sip 
Rename function to getDebugSipKernel and select either bindless
or heapless depending on mode

Related-to: NEO-8396

Signed-off-by: Brandon Yates <brandon.yates@intel.com>
 2024-10-03 20:47:40 +02:00
Zbigniew Zdanowicz
49371cb13e refactor: unify programming of preferred slm size 1/n 
- rename function to reflect actual task function does

Related-To: NEO-12639

Signed-off-by: Zbigniew Zdanowicz <zbigniew.zdanowicz@intel.com>
 2024-10-03 18:52:01 +02:00
Mateusz Hoppe
e5658e5327 fix: fix freeing allocations in aub mode 
- free in operations handler with aub dump must be called to evict
allocations from residentAllocations list
- prior to this change - stale allocations where in the container
leading to crash in aub modes

Related-To: NEO-12869

Signed-off-by: Mateusz Hoppe <mateusz.hoppe@intel.com>
 2024-10-03 17:14:07 +02:00
Filip Hazubski
388c7dc591 fix: Update SyncBuffer size calculation 
Add a constant to describe minimal size of SyncBuffer object.

Related-To: HSD-18039952263, HSD-18039940553, HSD-18039937640

Signed-off-by: Filip Hazubski <filip.hazubski@intel.com>
 2024-10-03 15:57:24 +02:00
Compute-Runtime-Validation
31618d9e18 Revert "fix: add workaround for incorrect DRM system info" 
This reverts commit 7ac991fa3f.

Signed-off-by: Compute-Runtime-Validation <compute-runtime-validation@intel.com>
 2024-10-03 15:22:21 +02:00
Compute-Runtime-Validation
41df1a6f47 Revert "feature: support SVM heap in reserveVirtualMem" 
This reverts commit bfaeeb01d6.

Signed-off-by: Compute-Runtime-Validation <compute-runtime-validation@intel.com>
 2024-10-03 14:53:50 +02:00
Mateusz Hoppe
74432a3626 fix: use correct memoryBanks when makeResident is called in aub mode 
Related-To: NEO-12869

Signed-off-by: Mateusz Hoppe <mateusz.hoppe@intel.com>
 2024-10-03 12:02:18 +02:00
Michal Mrozek
cb3463db05 fix: update description of debug env 
Signed-off-by: Michal Mrozek <michal.mrozek@intel.com>
 2024-10-03 10:10:33 +02:00
Mateusz Jablonski
e56b2b257f fix: reduce preemption surface size in aub tests 
Related-To: NEO-12648

Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
 2024-10-03 08:29:45 +02:00
Bartosz Dunajski
0564a41919 feature: debug flag to control direct submission semaphore mode 
Related-To: HSD-18039521047

Signed-off-by: Bartosz Dunajski <bartosz.dunajski@intel.com>
 2024-10-02 21:06:53 +02:00
Dominik Dabek
ac8dcdb298 fix: do not unregister shared allocation 
Allocations imported from shared handle are not using registerAlloc in drm
manager but on free, unregister was called.

This could lead to problems with allocation size tracking.

This change will skip the unregisterAllocation call if allocation is
imported from shared handle.

Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
 2024-10-02 16:32:24 +02:00
Mateusz Jablonski
14c8f1f15d refactor: remove not used function 
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
 2024-10-02 16:14:40 +02:00
Naklicki, Mateusz
516e502b8c fix: remove obsolete workaround about memory locking 
Signed-off-by: Naklicki, Mateusz <mateusz.naklicki@intel.com>
 2024-10-02 15:55:50 +02:00
Artur Harasimiuk
22505ca3ed refactor: cleanup coding style 
Signed-off-by: Artur Harasimiuk <artur.harasimiuk@intel.com>
 2024-10-02 12:42:39 +02:00
Compute-Runtime-Validation
680e62d333 Revert "performance: Set dispatch all for small TG" 
This reverts commit 0dc2870513.

Signed-off-by: Compute-Runtime-Validation <compute-runtime-validation@intel.com>
 2024-10-02 05:41:13 +02:00
Andrzej Koska
6abc5eb1a1 fix: using releaseHelper to determine MTP enablement 
Related-To: NEO-12466

Signed-off-by: Andrzej Koska <andrzej.koska@intel.com>
 2024-10-01 15:06:07 +02:00
Mateusz Jablonski
e4be88dff0 refactor: correct location of ioctl helper files 
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
 2024-10-01 14:48:12 +02:00
Slawomir Milczarek
edeb7bdd4b refactor: Allocate copy source for work partition allocation on heap 
No perforamce impact expected since it is initialized once only,
but has the advantage of using custom allocator by overriding malloc.

Related-To: NEO-12846

Signed-off-by: Slawomir Milczarek <slawomir.milczarek@intel.com>
 2024-10-01 13:32:55 +02:00
Kamil Kopryk
1378244415 refactor: improve readability 
Signed-off-by: Kamil Kopryk <kamil.kopryk@intel.com>
 2024-10-01 12:55:53 +02:00
Mateusz Jablonski
9819b1e2e4 refactor: remove not needed flag ftrGpGpuMidThreadLevelPreempt 
the flag was used for mid thread preemption support on pre-gen12 platforms

Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
 2024-10-01 12:52:12 +02:00
Mateusz Jablonski
bb8b376d30 feature: enable WMTP on LNL 
Related-To: NEO-12648
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
 2024-10-01 12:31:20 +02:00
Szymon Morek
b2fd1972a4 fix: add cpu alloc to eviction list only once 
Related-To: NEO-12572

Also, before migration to GPU domain, remove it from this list

Signed-off-by: Szymon Morek <szymon.morek@intel.com>
 2024-10-01 11:47:32 +02:00
Mateusz Jablonski
9db83b8231 refactor: unify isMidThreadPreemptionSupported function 
mid thread preemption can be enabled only by ftrWalkerMTP flag
pre-Xe2 devices doesn't support MTP

Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
 2024-10-01 11:26:04 +02:00
Lukasz Jobczyk
f87c3b57bb refactor: Add AIL infrastructure to disable 256B Prefetch 
Signed-off-by: Lukasz Jobczyk <lukasz.jobczyk@intel.com>
 2024-10-01 11:19:04 +02:00
Mateusz Jablonski
da8ce85f0a fix: correct preemption surface size when getting wmtp sip from igc 
when SIP kernel is taken from IGC then state save area header
describes preemption surface size required for the SIP kernel

Related-To: NEO-12826
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
 2024-10-01 11:06:58 +02:00
Dominik Dabek
752f313808 fix: limit allocation cache memory wastage 
Allocations over a certain size will be checked for memory utilization
when chosen for reuse.
If utilization is below a threshold, they will not be reused.

Related-To: NEO-6893

Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
 2024-10-01 09:49:19 +02:00
Filip Hazubski
72cf384c7d refactor: Fix typo 
Signed-off-by: Filip Hazubski <filip.hazubski@intel.com>
 2024-10-01 09:31:02 +02:00
Lukasz Jobczyk
0dc2870513 performance: Set dispatch all for small TG 
Resolves: NEO-11814

Signed-off-by: Lukasz Jobczyk <lukasz.jobczyk@intel.com>
 2024-10-01 09:28:55 +02:00
Compute-Runtime-Validation
ef1b569a85 Revert "performance: Do not create global fence allocation on integrated" 
This reverts commit 6bf5183eff.

Signed-off-by: Compute-Runtime-Validation <compute-runtime-validation@intel.com>
 2024-10-01 08:14:39 +02:00
Mateusz Jablonski
8c1685b93c fix: change default gen12lp preemption mode to thread group 
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
 2024-09-30 19:19:46 +02:00
Bartosz Dunajski
b8fd1bda36 feature: use sysInfo helper to detect memory type 
Related-To: NEO-12807

Signed-off-by: Bartosz Dunajski <bartosz.dunajski@intel.com>
 2024-09-30 18:19:42 +02:00
Jitendra Sharma
733ca5ac28 fix: Write tail index even for fifo wrap around case 
Related-To: NEO-7990
Signed-off-by: Jitendra Sharma <jitendra.sharma@intel.com>
 2024-09-30 18:19:28 +02:00
Lukasz Jobczyk
6bf5183eff performance: Do not create global fence allocation on integrated 
Signed-off-by: Lukasz Jobczyk <lukasz.jobczyk@intel.com>
 2024-09-30 13:13:27 +02:00
Compute-Runtime-Validation
f03550487f Revert "fix: add cpu alloc to eviction list only once" 
This reverts commit dfc863c7c1.

Signed-off-by: Compute-Runtime-Validation <compute-runtime-validation@intel.com>
 2024-09-28 11:50:36 +02:00