Commit Graph

99 Commits

Author SHA1 Message Date
Lukasz Jobczyk
8a0c425495 feature: Mark selected resources as UC when mitigating dc flush
Related-To: NEO-10556

Signed-off-by: Lukasz Jobczyk <lukasz.jobczyk@intel.com>
2024-04-12 12:52:13 +02:00
Bartosz Dunajski
1627172e28 feature: add helper to check if device to host copy fence is required
Related-To: NEO-10417

Signed-off-by: Bartosz Dunajski <bartosz.dunajski@intel.com>
2024-04-08 11:18:44 +02:00
Kacper Nowak
b8d19ccfb5 refactor: Move logic for querying maxMemoryFillPatternSize to product helper
Signed-off-by: Kacper Nowak <kacper.nowak@intel.com>
2024-04-05 09:35:01 +02:00
Kacper Nowak
35ca2bbea9 refactor: refactor logic for querying atomic FP capabilities
- Move logic from product helper to compiler product helper
- Add method for adjusting fp16 and extra capabilities using release
helper (if present).


Signed-off-by: Kacper Nowak <kacper.nowak@intel.com>
2024-03-28 15:27:45 +01:00
Maciej Plewka
b722f3b579 feature: Add interface to bind resources as readonly
Related-To: NEO-10398
Signed-off-by: Maciej Plewka <maciej.plewka@intel.com>
2024-03-27 14:24:58 +01:00
Compute-Runtime-Validation
8e44a46983 Revert "feature: bind resources as read only"
This reverts commit f3d36d3350.

Signed-off-by: Compute-Runtime-Validation <compute-runtime-validation@intel.com>
2024-03-27 08:51:47 +01:00
Maciej Plewka
f3d36d3350 feature: bind resources as read only
Related-to: NEO-10398
Signed-off-by: Maciej Plewka <maciej.plewka@intel.com>
2024-03-26 14:11:57 +01:00
Dominik Dabek
562d6f114b performance: helper for adjust ULLS on battery
Related-To: NEO-10800

Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
2024-03-26 12:57:30 +01:00
Dominik Dabek
2b964254d6 performance: debug key for adjust ULLS on battery
ULLS controller timeout settings will be adjusted based on ac line
status and lowest queue throttle from submissions.

Lowest queue throttle is reset when controller stops ULLS.

Related-To: NEO-10800

Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
2024-03-22 14:24:00 +01:00
Dominik Dabek
5ba9308804 performance: debug flag for localPreferred
Add flag for setting localPreferred (implicit when gmm localOnly=0 and
NonLocalOnly=0) when allocating buffer, svmGpu and image.

Related-To: NEO-9695

Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
2024-03-11 10:51:49 +01:00
Mrozek, Michal
10313b7b84 refactor: remove not needed code
Signed-off-by: Mrozek, Michal <michal.mrozek@intel.com>
2024-03-07 18:50:16 +01:00
Lukasz Jobczyk
6d1a3d404e refactor: Add helper to control flat ring buffer
Signed-off-by: Lukasz Jobczyk <lukasz.jobczyk@intel.com>
2024-03-07 17:52:23 +01:00
Dominik Dabek
0120d8a58d performance: program pat index on mtl linux
Enable programming pat indexes on mtl linux for device buffers.

Change DrmMemoryManager::allocateMemoryByKMD to use gemCreateExt.

Set mmap flags based on coherency.
Map as write back on legacy and coherent.
On non-coherent map as write combined.

Changes currently disabled, to enable use debug keys:
DisableGemCreateExtSetPat=0
UseGemCreateExtInAllocateMemoryByKMD=1

Reorder BufferObject to decrease padding.

Related-To: NEO-7896

Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
2024-02-16 17:33:07 +01:00
Dominik Dabek
ed011de03e performance: program pat index on mtl linux
Enable programming pat indexes on mtl linux for device buffers.

Change DrmMemoryManager::allocateMemoryByKMD to use gemCreateExt.

Changes currently disabled, can be enabled with flag
DisableGemCreateExtSetPat=0

Related-To: NEO-7896

Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
2024-02-15 17:15:28 +01:00
Compute-Runtime-Validation
7b340775c6 Revert "performance: program pat index on mtl linux"
This reverts commit 8e0b23db84.

Signed-off-by: Compute-Runtime-Validation <compute-runtime-validation@intel.com>
2024-02-15 02:06:03 +01:00
Dominik Dabek
8e0b23db84 performance: program pat index on mtl linux
Enable programming pat indexes on mtl linux for device buffers.

Change DrmMemoryManager::allocateMemoryByKMD to use gemCreateExt.

Related-To: NEO-7896

Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
2024-02-14 18:42:04 +01:00
Kacper Nowak
cccd9a3703 refactor: Add method for retrieving extra device info
Signed-off-by: Kacper Nowak <kacper.nowak@intel.com>
2024-02-14 16:38:15 +01:00
Mateusz Jablonski
bb5f6d9660 fix: don't query vm bind support on i915 prelim for pre-Xe platforms
Related-To: HSD-18036843571
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
2024-02-09 12:50:48 +01:00
Dunajski, Bartosz
923707a509 feature: initial support for local dispatch size query
Related-To: NEO-8070

Signed-off-by: Dunajski, Bartosz <bartosz.dunajski@intel.com>
2024-02-07 14:47:05 +01:00
Kacper Nowak
999ec9b2ca refactor: Unify logic for getting atomic FP caps 1/n
- Separate logic for fp16/32/46 caps.
- Add aggregated constexprs for local & global caps of given type
- Pass arguments by reference
- Add hwInfo as argument for future refactors
- Add static_asserts in L0 to ensure there is no mismatch between
internal/external caps
Signed-off-by: Kacper Nowak <kacper.nowak@intel.com>
2024-02-07 11:39:36 +01:00
Katarzyna Cencelewska
d0b009901c fix: use proper gmm resource type for uncache resources
when new coherency model

Resolves: NEO-9657
Signed-off-by: Katarzyna Cencelewska <katarzyna.cencelewska@intel.com>
2024-01-31 15:26:11 +01:00
Mateusz Jablonski
da16dad344 fix: don't limit vm bind support based on platform
Related-To: GSD-7097
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
2024-01-29 19:43:15 +01:00
Katarzyna Cencelewska
eec01e500a fix: non-coherency issue on arl
Resolves: HSD-15015200338
Signed-off-by: Katarzyna Cencelewska <katarzyna.cencelewska@intel.com>
2024-01-26 10:26:33 +01:00
Kacper Nowak
6cf71cce81 refactor: Allow for grouping BCS1+ copy engines in more engine group types
Do not group copy engines from BCS1+ in linked copy group by default.

Signed-off-by: Kacper Nowak <kacper.nowak@intel.com>
2024-01-23 12:13:09 +01:00
Dominik Dabek
b3b402cad2 feature(zebin): read indirect detection version
Read indirect detection version intel compat notes from zebin.

This is to prepare for enabling indirect access optimization in AOT
built kernels.

Related-To: NEO-7712

Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
2024-01-12 14:20:52 +01:00
Dominik Dabek
9b52d52062 performance(ocl): enable usm pool allocator
Enable on xe hpg and lpg platforms

Related-To: NEO-9700

Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
2024-01-05 16:17:48 +01:00
Compute-Runtime-Validation
341a13ce8e Revert "performance: Limit tlb flush WA scope on DG2 Linux"
This reverts commit fa181937a4.

Signed-off-by: Compute-Runtime-Validation <compute-runtime-validation@intel.com>
2023-12-24 08:17:39 +01:00
Kacper Nowak
18fb432552 refactor: Stop using hard-coded BCS engine as default one
Do not rely on presence of BCS0 engine.

Signed-off-by: Kacper Nowak <kacper.nowak@intel.com>
2023-12-21 17:34:57 +01:00
Lukasz Jobczyk
fa181937a4 performance: Limit tlb flush WA scope on DG2 Linux
Signed-off-by: Lukasz Jobczyk <lukasz.jobczyk@intel.com>
2023-12-21 12:18:29 +01:00
Compute-Runtime-Validation
590418a588 Revert "performance: Limit tlb flush WA scope on DG2 Linux"
This reverts commit 0b85a9f256.

Signed-off-by: Compute-Runtime-Validation <compute-runtime-validation@intel.com>
2023-12-21 08:52:58 +01:00
Lukasz Jobczyk
0b85a9f256 performance: Limit tlb flush WA scope on DG2 Linux
Signed-off-by: Lukasz Jobczyk <lukasz.jobczyk@intel.com>
2023-12-20 16:53:54 +01:00
Aravind Gopalakrishnan
f5045348ad fix: Report correct DP support for platforms
Related-To: NEO-9702

- Remove older interfaces and add new release helper

Signed-off-by: Aravind Gopalakrishnan <aravind.gopalakrishnan@intel.com>
2023-12-20 10:35:49 +01:00
Mateusz Jablonski
beafea9b39 refactor: correct naming of enum class constants 2/n
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
2023-12-11 13:13:35 +01:00
Kamil Kopryk
ce7298d512 feature: Add heapless mode programming in ocl
Related-To: NEO-7621
Signed-off-by: Kamil Kopryk <kamil.kopryk@intel.com>
2023-11-24 12:53:39 +01:00
Dominik Dabek
6562828095 performance: prealloc internal heap on mtl
Preallocate 1 internal heap allocation per csr on mtl

Related-To: NEO-8152

Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
2023-11-17 13:36:21 +01:00
Dominik Dabek
8f06f3f50a performance: add override cacheable to gmm
add attribute to override cacheable attribute to gmm constructor

enable this override for command buffers on mtl

change command buffers back to allocation by kmd

this keeps the quicker allocation which is needed to keep enqueue times
low

Related-To: NEO-8152

Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
2023-11-16 10:00:45 +01:00
Aravind Gopalakrishnan
99abb40a45 feature: Add extensions to expose dot product capabilities
- Add new extension property for user to query platform DP support.
- Add new blackbox and ULTs for extension.

Related-To: NEO-8786

Signed-off-by: Aravind Gopalakrishnan <aravind.gopalakrishnan@intel.com>
2023-10-31 21:54:39 +01:00
Compute-Runtime-Validation
c9d6615231 Revert "feature: Add extensions to expose dot product capabilities"
This reverts commit e988fe66e3.

Signed-off-by: Compute-Runtime-Validation <compute-runtime-validation@intel.com>
2023-10-31 05:06:06 +01:00
Aravind Gopalakrishnan
e988fe66e3 feature: Add extensions to expose dot product capabilities
- Add new extension property for user to query platform DP support.
- Add new blackbox and ULTs for extension.

Related-To: NEO-8786

Signed-off-by: Aravind Gopalakrishnan <aravind.gopalakrishnan@intel.com>
2023-10-30 15:15:51 +01:00
Dominik Dabek
39cf653959 performance(ocl): cmd buffer prealloc per cmdqueue
Add mechanism to preallocate cmd buffer allocations in command stream
receiver reusable allocations list per command queue initialized.

This should limit additional allocations during hot loop.

Needs to be enabled in subsequent commits by setting product helper
method.

Related-To: NEO-8152

Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
2023-10-27 16:56:29 +02:00
Mateusz Jablonski
6eeb9322b9 refactor: remove not needed method
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
2023-10-06 15:22:40 +02:00
Mateusz Jablonski
038c287656 fix: avoid double free when wddm initialization fails
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
2023-10-05 13:03:49 +02:00
Naklicki, Mateusz
0461af492d fix: unify path for getting number of grfs per thread
Related-To: NEO-8043

Signed-off-by: Naklicki, Mateusz <mateusz.naklicki@intel.com>
2023-10-03 08:17:46 +02:00
Baj, Tomasz
cb903cbc03 refactor: Move disableRcs to releaseHelper
Related-To: NEO-8296

Signed-off-by: Baj, Tomasz <tomasz.baj@intel.com>
2023-09-29 14:19:02 +02:00
Mateusz Jablonski
60c30a05e7 refactor: remove not needed function
Related-To: NEO-7527

Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
2023-09-27 14:15:11 +02:00
Mateusz Jablonski
3b2c41e5ee fix: define isDirectSubmissionSupported per HW release
Related-To: NEO-8187
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
2023-09-21 09:37:48 +02:00
Mateusz Jablonski
b1808f7830 fix: correct suggested number of work groups for concurrent kernels on PVC
value depends on CCS count:
- single CCS mode (default) - 50% available
- two CCS mode - 25% available
- four CCS mode - 12.5% available

Related-To: NEO-8377
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
2023-09-20 13:40:22 +02:00
Compute-Runtime-Validation
73731d3be5 Revert "fix: correct suggested number of work groups for concurrent kernels o...
This reverts commit 6fc673b0fe.

Signed-off-by: Compute-Runtime-Validation <compute-runtime-validation@intel.com>
2023-09-15 04:21:58 +02:00
Mateusz Jablonski
6fc673b0fe fix: correct suggested number of work groups for concurrent kernels on PVC
value depends on CCS count:
- single CCS mode (default) - no limitations
- two CCS mode - 25% available
- four CCS mode - 12.5% available

Related-To: NEO-8377
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
2023-09-14 16:13:54 +02:00
Mateusz Jablonski
3eb98163a8 fix: define isCachingOnCpuAvailable per hw release
Related-To: NEO-8187
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
2023-09-13 11:13:42 +02:00