intel/compute-runtime
1
0
Fork 0
mirror of https://github.com/intel/compute-runtime.git synced 2025-12-29 00:58:39 +08:00
Code Issues Packages Projects Releases Wiki Activity
19,725 Commits 42 Branches 655 Tags
cb4fa456b7d458f962f99f06bbad88e7e20d43ca
Commit Graph

318 Commits

Author SHA1 Message Date
Jaroslaw Warchulski
2375ff89bb refactor: cleanup headers 
Related-To: NEO-5548
Signed-off-by: Jaroslaw Warchulski <jaroslaw.warchulski@intel.com>
 2025-06-23 11:23:49 +02:00
Aravind Gopalakrishnan
8e45fb880f fix: Add check if packet count is exceeded 
Related-To: NEO-13003

Signed-off-by: Aravind Gopalakrishnan <aravind.gopalakrishnan@intel.com>
 2025-06-16 20:24:29 +02:00
Compute-Runtime-Validation
4e0c765d72 Revert "performance: reorder checks for staging buffer" 
This reverts commit e030086cf2.

Signed-off-by: Compute-Runtime-Validation <compute-runtime-validation@intel.com>
 2025-06-10 23:35:58 +02:00
Szymon Morek
6a7f6f1ad6 fix: add noexcept to move operators 
Without it, copy operator is selected, even if move is explicitly defined.

Signed-off-by: Szymon Morek <szymon.morek@intel.com>
 2025-06-09 10:32:23 +02:00
Szymon Morek
e030086cf2 performance: reorder checks for staging buffer 
defer registerHostPtr call to the last check.

Signed-off-by: Szymon Morek <szymon.morek@intel.com>
 2025-06-09 10:18:35 +02:00
Fabian Zwoliński
e2228201ce fix: Avoid redundant padding in ISA allocations 
- Add isaPaddingIncluded flag to AllocationProperties to control ISA
padding
- Modify MemoryManager to skip automatic padding when flag is set
- Update ISAPoolAllocator to manage padding explicitly and set the
flag
- Use actual buffer size for HeapAllocator instead of requested size
- Add test covering different padding scenarios

This prevents double padding addition which was wasting memory when
ISAPoolAllocator managed its own padding on top of MemoryManager's
automatic padding.

Related-To: NEO-12287, HSD-18042276431
Signed-off-by: Fabian Zwoliński <fabian.zwolinski@intel.com>
 2025-05-30 15:40:38 +02:00
Lukasz Jobczyk
0e3397547c fix: Adjust waitpkg counter for non ulls light 
Resolves: NEO-14866, HSD-18042153820

Signed-off-by: Lukasz Jobczyk <lukasz.jobczyk@intel.com>
 2025-05-23 15:58:21 +02:00
Szymon Morek
a2c41e4749 fix: add mock for staging buffer size 
No need to allocate 2MB in unit tests.
This commit changes this size to 4KB.

Signed-off-by: Szymon Morek <szymon.morek@intel.com>
 2025-05-22 08:36:01 +02:00
Daniel Bermond
e0362a7c39 build: add missing headers for GCC 15 
For using fixed width integer types[1], the `<cstdint>`[2]
C++ header needs to be explicitly included with GCC 15 due
to changes[3] in libstdc++.

For details, see the documentation[4] about porting to GCC 15.

[1] https://en.cppreference.com/w/cpp/types/integer
[2] https://en.cppreference.com/w/cpp/header/cstdint
[3] https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=3a817a4a5a6d94da9127af3be9f84
[4] https://gcc.gnu.org/gcc-15/porting_to.html#cxx

Signed-off-by: Daniel Bermond <dbermond@archlinux.org>
 2025-05-19 15:31:32 +02:00
Lukasz Jobczyk
5cd5bbafc5 refactor: Split waitpkg params for ulls light and default 
Related-To: NEO-14866

Signed-off-by: Lukasz Jobczyk <lukasz.jobczyk@intel.com>
 2025-05-09 12:40:36 +02:00
Andrzej Koska
41c0546c45 fix: reset timestamps when userEvent is terminated 
Related-To: NEO-14448

Signed-off-by: Andrzej Koska <andrzej.koska@intel.com>
 2025-05-06 10:28:35 +02:00
Compute-Runtime-Validation
3fa9229483 Revert "performance: copy misaligned leftover in staging" 
This reverts commit 13610d3473.

Signed-off-by: Compute-Runtime-Validation <compute-runtime-validation@intel.com>
 2025-05-06 02:12:48 +02:00
Szymon Morek
13610d3473 performance: copy misaligned leftover in staging 
Related-To: NEO-14646

Signed-off-by: Szymon Morek <szymon.morek@intel.com>
 2025-05-05 16:56:27 +02:00
Lukasz Jobczyk
97358acabe fix: use mfence instead of sfence on discrete devices 
Related-To: NEO-14642

Signed-off-by: Lukasz Jobczyk <lukasz.jobczyk@intel.com>
 2025-04-25 13:32:25 +02:00
Lukasz Jobczyk
5d2b673ef6 fix: Adjust waitpkg threshold to 20 us 
Resolves: HSD-13012958900

Signed-off-by: Lukasz Jobczyk <lukasz.jobczyk@intel.com>
 2025-04-23 10:46:22 +02:00
Fabian Zwoliński
204577a91f fix: disable TimestampPoolAllocator 
Related-To: NEO-12287, HSD-13012945876
Signed-off-by: Fabian Zwoliński <fabian.zwolinski@intel.com>
 2025-04-18 11:30:35 +02:00
Chodor, Jaroslaw
a855ce019e refactor: pass underlying allocator by ref 
Signed-off-by: Chodor, Jaroslaw <jaroslaw.chodor@intel.com>
 2025-04-17 10:43:21 +02:00
Chodor, Jaroslaw
7744251995 refactor: explicit copy/move operations in OpaqueElementAllocator 
Signed-off-by: Chodor, Jaroslaw <jaroslaw.chodor@intel.com>
 2025-04-16 10:10:05 +02:00
Jaroslaw Warchulski
3e1aa33924 refactor: cleanup headers 
Related-To: NEO-5548
Signed-off-by: Jaroslaw Warchulski <jaroslaw.warchulski@intel.com>
 2025-04-14 14:59:40 +02:00
Marcel Skierkowski
b75fbe8e2c refactor: mock filesystem in ocloc ult pt.1 
Mocked IO operations in ./ocloc_tests application

Mocked gtest stdout capture in ocloc tests

Related-To: NEO-14084
Signed-off-by: Marcel Skierkowski <marcel.skierkowski@intel.com>
 2025-04-08 16:08:42 +02:00
Szymon Morek
ff1dcc1a8a performance: adust copy from image to host logic 
Related-To: NEO-14569

Signed-off-by: Szymon Morek <szymon.morek@intel.com>
 2025-04-08 10:02:00 +02:00
Szymon Morek
95e0244f70 fix: properly pass info about 3D image 
Related-To: NEO-14538

It's valid for 3D image to copy 2D region.
Current checks for mip map do not consider that.
This change correctly checks for mip mapped 3D image.

Signed-off-by: Szymon Morek <szymon.morek@intel.com>
 2025-04-03 16:33:57 +02:00
Szymon Morek
6ea83f322d fix: do not override user data beyond slice region 
Related-To: NEO-14538

If user passes slice pitch which is larger than region
to copy, do not override memory beyond region but within
that slice pitch.

Signed-off-by: Szymon Morek <szymon.morek@intel.com>
 2025-04-03 12:37:35 +02:00
Fabian Zwoliński
7ef3880793 feature: implement pool allocator for gpuTimestampDeviceBuffer 
The patch applies to Level Zero.
Only allocations < 2MB will be fetched from the pool.
Allocations are shared and reused within a given device.

Additionally, I added a new debug flag to control the allocator:
EnableTimestampPoolAllocator

Related-To: NEO-12287
Signed-off-by: Fabian Zwoliński <fabian.zwolinski@intel.com>
 2025-04-02 14:28:56 +02:00
Szymon Morek
8836f6df0b fix: forward mip map level for 3D images 
Related-To: NEO-14539

1D and 2D images have already mip map level set correctly.

Signed-off-by: Szymon Morek <szymon.morek@intel.com>
 2025-04-02 11:46:41 +02:00
Szymon Morek
3010af596e performance: add infrastructure for staging with 3D images 
Related-To: NEO-14026

Signed-off-by: Szymon Morek <szymon.morek@intel.com>
 2025-04-01 15:30:30 +02:00
Lukasz Jobczyk
60b551758c performance: Adjust waitpkg threshold for discrete devices 
Related-To: NEO-14336

Signed-off-by: Lukasz Jobczyk <lukasz.jobczyk@intel.com>
 2025-03-26 14:59:19 +01:00
Bartosz Dunajski
8c21a9634f refactor: add missing file to cmake list 
Signed-off-by: Bartosz Dunajski <bartosz.dunajski@intel.com>
 2025-03-24 15:19:03 +01:00
Lukasz Jobczyk
8fbde30432 fix: Adjust waitpkg threshold to 12 us 
Resolves: NEO-14434, NEO-14435
Related-To: NEO-14336

Signed-off-by: Lukasz Jobczyk <lukasz.jobczyk@intel.com>
 2025-03-24 11:43:54 +01:00
Lukasz Jobczyk
8a85a96ed2 feature: Add 3-level wait scheme with tpause intrinsic 
Related-To: NEO-14336

Signed-off-by: Lukasz Jobczyk <lukasz.jobczyk@intel.com>
 2025-03-21 12:12:57 +01:00
Chodor, Jaroslaw
0bff9def3b feature: New allocators for opaque arrays 
Adds fast allocators for opaque objects of uniform size.

Related-To: NEO-13406

Signed-off-by: Chodor, Jaroslaw <jaroslaw.chodor@intel.com>
 2025-03-20 21:59:44 +01:00
Lukasz Jobczyk
75461d190d performance: Adjust umwait counter value 
Related-To: NEO-13922, NEO-14336

Signed-off-by: Lukasz Jobczyk <lukasz.jobczyk@intel.com>
 2025-03-20 13:57:25 +01:00
Mateusz Hoppe
a86105814d fix: default initialize allocated memory when memory is overwritten 
Related-To: NEO-5093

Signed-off-by: Mateusz Hoppe <mateusz.hoppe@intel.com>
 2025-03-11 15:32:33 +01:00
Lukasz Jobczyk
be8545f7df refactor: reset all waitpkg variables on test start 
Signed-off-by: Lukasz Jobczyk <lukasz.jobczyk@intel.com>
 2025-03-11 08:44:36 +01:00
Szymon Morek
dacbce7f01 performance: introduce staging read for cl_buffer 
Related-To: NEO-14026

Signed-off-by: Szymon Morek <szymon.morek@intel.com>
 2025-03-11 08:19:28 +01:00
Fabian Zwoliński
f5e37e725c Revert "fix: configure ISA Pool params based on productHelper" 
This reverts commit bf20ae7ae8.

Signed-off-by: Fabian Zwoliński <fabian.zwolinski@intel.com>
 2025-03-10 22:41:13 +01:00
Lukasz Jobczyk
53062056ec performance: Enable wait pkg for ULLS light 
Related-To: NEO-13922

Signed-off-by: Lukasz Jobczyk <lukasz.jobczyk@intel.com>
 2025-03-07 12:18:38 +01:00
Dominik Dabek
9eb8e1812c feature: flag to log usm reuse operations 
If flag "LogUsmReuse" is set, usm reuse will log operations to csv file.
Each line will contain: timestamp, host/device, operation type,
allocation size, true/false whether operation succeeded.

This data can then be used to produce graphs and help in analyzing
usm reuse in a particular workload.

Related-To: NEO-6893

Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
 2025-03-06 11:06:27 +01:00
Szymon Morek
01a66a9347 fix: download staging buffer before reading from it 
Related-To: NEO-14154

Signed-off-by: Szymon Morek <szymon.morek@intel.com>
 2025-02-21 09:30:23 +01:00
Bartosz Dunajski
85b0d1f84d fix: clear all TS packets for combined CB event 
Related-To: NEO-13971

Signed-off-by: Bartosz Dunajski <bartosz.dunajski@intel.com>
 2025-02-20 19:24:15 +01:00
Fabian Zwoliński
bf20ae7ae8 fix: configure ISA Pool params based on productHelper 
When is2MBLocalMemAlignmentEnabled returns true,
increase pool size for builtins from 64k to 2MB.

Additionally, set appropriate alignment for kernel ISA heap allocations.
Additionally, configure isaAllocationPageSize based on productHelper

Related-To: NEO-12287
Signed-off-by: Fabian Zwoliński <fabian.zwolinski@intel.com>
 2025-02-20 08:42:35 +01:00
Filip Hazubski
b60c02d597 fix: Add asserts to ensure NonCopyable and NonMovable n/n 
Related-To: NEO-14068

Signed-off-by: Filip Hazubski <filip.hazubski@intel.com>
 2025-02-19 11:36:24 +01:00
Filip Hazubski
6b6202446b fix: Add asserts to ensure NonCopyable and NonMovable 3/n 
Related-To: NEO-14068

Signed-off-by: Filip Hazubski <filip.hazubski@intel.com>
 2025-02-18 17:16:03 +01:00
Jack Myers
c26d24e555 fix: tbx page fault manager hang issue 
- Updated `isAllocTbxFaultable` to exclude `gpuTimestampDeviceBuffer` from being
faultable.
- Replaced `SpinLock` with `RecursiveSpinLock` in `CpuPageFaultManager` and
`TbxPageFaultManager` to allow recursive locking.
- Added unit tests to verify the correct handling of `gpuTimestampDeviceBuffer`
in `TbxCommandStreamTests`.

Related-To: NEO-13748
Signed-off-by: Jack Myers <jack.myers@intel.com>
 2025-02-18 05:05:38 +01:00
Filip Hazubski
4c7900008f refactor: Change wording from NonCopyableOrMovable to NonCopyableAndNonMovable 
Signed-off-by: Filip Hazubski <filip.hazubski@intel.com>
 2025-02-17 14:19:10 +01:00
Fabian Zwoliński
1d8c8ac843 refactor: use designated initializers for SmallBuffersParams initialization 
Related-To: NEO-12287
Signed-off-by: Fabian Zwoliński <fabian.zwolinski@intel.com>
 2025-02-13 12:32:52 +01:00
Szymon Morek
8152241693 fix: don't override user data during staging read 
Related-To: NEO-13574

User can read image with row pitch larger than region width.
In such cases, don't override memory which is out of region
but within row pitch.

Signed-off-by: Szymon Morek <szymon.morek@intel.com>
 2025-02-12 16:22:48 +01:00
Compute-Runtime-Validation
116f7270be Revert "fix: tbx page fault manager hang issue" 
This reverts commit 7d4e70a25b.

Signed-off-by: Compute-Runtime-Validation <compute-runtime-validation@intel.com>
 2025-02-12 10:38:05 +01:00
Jack Myers
7d4e70a25b fix: tbx page fault manager hang issue 
- Updated `isAllocTbxFaultable` to exclude `gpuTimestampDeviceBuffer` from being
faultable.
- Replaced `SpinLock` with `RecursiveSpinLock` in `CpuPageFaultManager` and
`TbxPageFaultManager` to allow recursive locking.
- Added unit tests to verify the correct handling of `gpuTimestampDeviceBuffer`
in `TbxCommandStreamTests`.

Related-To: NEO-13748
Signed-off-by: Jack Myers <jack.myers@intel.com>
 2025-02-12 02:19:37 +01:00
Kamil Kopryk
03b56dc919 fix: untrusted input 
Do not read an unlimited size of environment variables.

Signed-off-by: Kamil Kopryk <kamil.kopryk@intel.com>
 2025-02-10 14:54:03 +01:00