1a4755694e
feat(zebin): Add support for SHT_ZEBIN_GTPIN_INFO type section
...
This commit adds support for decoding SHT_ZEBIN_GTPIN_INFO type
sections. For each section, passed data will be stored in kernel info
(for corresponding kernel).
Related-To: NEO-7689
Signed-off-by: Kacper Nowak <kacper.nowak@intel.com>
2023-02-10 16:21:59 +01:00
7790e208fd
feat(zebin): Add support for ELF section type SHT_NOBITS
...
This commit adds support for parsing SHT_NOBITS zebin's ELF sections
(containing global/constant zero-initialized data).
- Correction: in CTNI path, do not add related symbol if surface has not
been allocated.
Related-To: NEO-7196
Signed-off-by: Kacper Nowak <kacper.nowak@intel.com>
2023-02-10 16:17:16 +01:00
52234201bb
Narrowing the usDeviceID range for WA
...
This patch narrows down the scope covered by WA
to G10 machines only
Related-To: NEO-7475
Signed-off-by: Andrzej Koska andrzej.koska@intel.com
2023-02-10 13:26:00 +01:00
2887f2d936
refactor: fix typo - specified
...
Signed-off-by: Kamil Kopryk <kamil.kopryk@intel.com>
2023-02-10 10:28:32 +01:00
1740e1e747
Limit properties update for immediate command list to used in flush task
...
Related-To: NEO-7701
Signed-off-by: Zbigniew Zdanowicz <zbigniew.zdanowicz@intel.com>
2023-02-09 17:45:43 +01:00
783df81a44
Unify flush task getter implementation of product and core helpers
...
Signed-off-by: Zbigniew Zdanowicz <zbigniew.zdanowicz@intel.com>
2023-02-09 16:42:35 +01:00
2778043d67
fix(l0): check for largeGRF when computing maxWorkGroupSize
...
Sizing context (PVC):
When using LargeGRF (a.k.a GRF256) there are only 4 HW threads per EU
(instead of default 8). Together with SIMD16 that means that there can
be max 64 work-items per EU. With 8 EU per subslice this gives 512
work-items on a single subslice. For correct intra-WG synchronization
all its WIs must be executed on the same subslice (to access the same
SLM, where the synchronization primitives are stored). Thus, with SIMD16
and LargeGRF the work-group size must not exceed 512 (PVC example).
So far `maxWorkGroupSize` is taken solely from a DeviceInfo structure
both in `ModuleTranslationUnit::processUnpackedBinary()` and
`ModuleImp::initialize()`. This method does not take kernel parameters
(LargeGRF) into account. It allows to submit a kernel using LargeGRF
with SIMD16 with the work-group size set to 1024. That leads to a hang.
Fix the `.maxWorkGroupSize` computation so that it takes the kernel
parameters into consideration.
Add new (for discrete platforms >= XeHP) and adapt existing tests, fix
cosmetics by the way.
Similar check for OCL:
https://github.com/intel/compute-runtime/blob/master/opencl/source/comma
nd_queue/enqueue_kernel.h#L130
Related-To: NEO-7684
Signed-off-by: Maciej Bielski <maciej.bielski@intel.com>
2023-02-08 11:20:52 +01:00
1d675fc145
Add tests for split handling barrier
...
Signed-off-by: Lukasz Jobczyk <lukasz.jobczyk@intel.com>
2023-02-07 17:44:51 +01:00
f2be0ebfc4
Allocate and consume shared heaps atomically
...
Related-To: NEO-5055
Signed-off-by: Zbigniew Zdanowicz <zbigniew.zdanowicz@intel.com>
2023-02-07 12:52:54 +01:00
9b7b193b38
Enable recoverable page faults on PVC platform only
...
Related-To: NEO-6355
Signed-off-by: Milczarek, Slawomir <slawomir.milczarek@intel.com>
2023-02-06 14:47:36 +01:00
9f574b6fba
Introduce barrier tracking mechanism
...
Related-To: NEO-7696
Signed-off-by: Lukasz Jobczyk <lukasz.jobczyk@intel.com>
2023-02-06 14:29:23 +01:00
6ebdc51fae
Dynamic queue size limit in RelaxedOrdering mode
...
Related-To: NEO-7458
Signed-off-by: Dunajski, Bartosz <bartosz.dunajski@intel.com>
2023-02-06 12:02:02 +01:00
7e0401d280
Add improvements to heap estimation in level zero command lists
...
- add estimation parameter for interface descriptor data count
- add to the heap estimation alignment parameter for dynamic and surface heaps
- extend encode interface and implementations to allow child heaps
Related-To: NEO-5055
Signed-off-by: Zbigniew Zdanowicz <zbigniew.zdanowicz@intel.com>
2023-02-03 20:26:27 +01:00
59109a08bb
Switch device ID support to product config helper
...
This commit switches the device ID logic from the deprecated
to the new one, so that if the user passes a hex value to the -device
parameter, ocloc will use the new implementation in the product config
helper. The change also introduces a fix for setting the values in the
correct order to configure the hwIfno correctly.
Signed-off-by: Daria Hinz daria.hinz@intel.com
Related-To: NEO-7487
2023-02-03 16:55:41 +01:00
24c5352350
refactor: remove redundant including of compiler_cache.h
...
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
2023-02-03 11:16:31 +01:00
1c976f5110
test: move execution environment tests to shared
...
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
2023-02-03 11:03:18 +01:00
cab4b956eb
refactor: rename compiler product helper files
...
Related-To: NEO-6853
Signed-off-by: Kamil Kopryk <kamil.kopryk@intel.com>
2023-02-03 09:03:24 +01:00
d0c0c60205
Revert "feat(zebin): Add support for ELF section type SHT_NOBITS"
...
This reverts commit fa03aa9a40
2023-02-03 03:44:02 +01:00
606a900080
Revert "Disable EUFusion for odd work groups with DPAS on DG2"
...
This reverts commit 017d66a469
2023-02-03 02:45:21 +01:00
fa03aa9a40
feat(zebin): Add support for ELF section type SHT_NOBITS
...
This commit adds support for parsing SHT_NOBITS zebin's ELF sections
(containing global/constant zero-initialized data).
Related-To: NEO-7196
Signed-off-by: Kacper Nowak <kacper.nowak@intel.com>
2023-02-02 14:54:51 +01:00
31154dc20b
Create OS agnostic OSTime in AUB/TBX mode
...
Signed-off-by: Dunajski, Bartosz <bartosz.dunajski@intel.com>
2023-02-02 14:35:26 +01:00
5097ef4825
Change dispatch kernel interface to provide already prepared heap objects
...
Related-To: NEO-5055
Signed-off-by: Zbigniew Zdanowicz <zbigniew.zdanowicz@intel.com>
2023-02-02 14:08:43 +01:00
401344137c
fix: Correctly set UUID for non-multi-tile archs
...
Use getSubDevicesCount() from hwInfo to determine whether device
is root or not, instead of isSubDevice(), since the former does not
change with the affinity mask.
Signed-off-by: Jaime Arteaga <jaime.a.arteaga.molina@intel.com>
2023-02-02 14:00:42 +01:00
017d66a469
Disable EUFusion for odd work groups with DPAS on DG2
...
Related-To: NEO-7495, HSD-14017007475
Signed-off-by: Maciej Plewka <maciej.plewka@intel.com>
2023-02-02 13:57:42 +01:00
ac63175a0f
Add extra check for nullptr function pointer
...
Related-To: NEO-6853
Signed-off-by: Kamil Kopryk <kamil.kopryk@intel.com>
2023-02-02 12:21:23 +01:00
8da362afae
fix(l0): do not memcpy on cpu if need unlock ptr
...
Do not use cpu memory copy on windows if need to unlock locked ptr.
Related-To: NEO-7553
Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
2023-02-02 10:41:39 +01:00
2484c7ceb2
refactor: rename hw_helper files to gfx_core_helper files
...
Related-To: NEO-6853
Signed-off-by: Kamil Kopryk <kamil.kopryk@intel.com>
2023-02-01 19:37:51 +01:00
14f5a61993
Fatbinary optimization for -device release target
...
This commit is to introduce optimizations in ocloc when building
targets for release and family.
Instead of building fatbinary after all available targets in
the RTL ID table, we introduce optimizations when there is an
acronym available for the platform in the DEVICE table,
we limit to them only.
Signed-off-by: Daria Hinz <daria.hinz@intel.com>
Related-To: NEO-7582
2023-02-01 16:19:13 +01:00
1a0bc60434
refactor: rename hw_helper_tests files to gfx_core_helper_tests
...
Related-To: NEO-6853
Signed-off-by: Kamil Kopryk <kamil.kopryk@intel.com>
2023-02-01 15:58:41 +01:00
e2f7fece00
Increase RealxedOrdering queue size
...
Related-To: NEO-7458
Signed-off-by: Dunajski, Bartosz <bartosz.dunajski@intel.com>
2023-02-01 14:57:00 +01:00
7487d1450e
Move CompilerProductHelper ownership to RootDeviceEnvironment and Ocloc
...
Related-To: NEO-6853
Signed-off-by: Kamil Kopryk <kamil.kopryk@intel.com>
2023-02-01 13:09:12 +01:00
4eb443f4dc
Revert "Narrowing the usDeviceID range for WA"
...
This reverts commit be9775891candrzej.koska@intel.com
2023-02-01 11:59:50 +01:00
104126ddd7
Move ProductHelper ownership to RootDeviceEnvironment
...
Related-To: NEO-6853
Signed-off-by: Kamil Kopryk <kamil.kopryk@intel.com>
2023-02-01 08:29:46 +01:00
78cb79912c
add helper method isUnlockingLockedPtrNecessary
...
Return unlocking necessary on DG2 windows
Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
2023-01-31 17:08:45 +01:00
34b8f08fc6
Add state base address properties tracking for command lists
...
Related-To: NEO-5055
Signed-off-by: Zbigniew Zdanowicz <zbigniew.zdanowicz@intel.com>
2023-01-31 12:47:17 +01:00
2f847aa7bb
Optimize zeMemGetAllocProperties.
...
- No functional changes only performance optimizations.
- add const modifiers
- add early returns
- remove not needed code
- do fast lookup for aligned pointers
- fix incorrect tests
+21% improvement in
./api_overhead_benchmark_l0 --test=GetMemoryProperties
--api=l0 --AmountOfUsmAllocations=100 --iterations=1000
+32% improvement in
./api_overhead_benchmark_l0 --test=GetMemoryProperties
--api=l0 --AmountOfUsmAllocations=1000 --iterations=1000
Related-To: NEO-7676
Signed-off-by: Michal Mrozek <michal.mrozek@intel.com>
2023-01-30 18:00:57 +01:00
9080b0c109
Delete redundant adjustNumberOfThreadsInThreadGroup method
...
Signed-off-by: Rafal Maziejuk <rafal.maziejuk@intel.com>
2023-01-30 17:10:01 +01:00
2abbd82195
Flush task at device init
...
Resolves: NEO-7642
Signed-off-by: Lukasz Jobczyk <lukasz.jobczyk@intel.com>
2023-01-30 10:41:42 +01:00
7b7c28cb40
refactor: don't use global ProductHelper getter 19/n
...
Related-To: NEO-6853
Signed-off-by: Kamil Kopryk <kamil.kopryk@intel.com>
2023-01-30 09:02:04 +01:00
004d3e3416
refactor: don't use global ProductHelper getter 18
...
Related-To: NEO-6853
Signed-off-by: Kamil Kopryk <kamil.kopryk@intel.com>
2023-01-30 08:15:39 +01:00
29f54be816
refactor: don't use global ProductHelper getter 17
...
Related-To: NEO-6853
Signed-off-by: Kamil Kopryk <kamil.kopryk@intel.com>
2023-01-27 19:28:45 +01:00
68bfd49033
refactor: don't use global ProductHelper getter 15/n
...
Related-To: NEO-6853
Signed-off-by: Kamil Kopryk <kamil.kopryk@intel.com>
2023-01-27 17:51:57 +01:00
5e059d4b30
refactor: don't use global ProductHelper getter 16
...
Related-To: NEO-6853
Signed-off-by: Kamil Kopryk <kamil.kopryk@intel.com>
2023-01-27 17:03:46 +01:00
445706361d
refactor: don't use global ProductHelper 14/n
...
Related-To: NEO-6853
Signed-off-by: Kamil Kopryk <kamil.kopryk@intel.com>
2023-01-27 14:51:12 +01:00
414dbed5d9
Cleanup includes 46
...
Cleaned up files:
opencl/source/api/api.h
opencl/source/os_interface/windows/d3d_sharing_functions.h
opencl/test/unit_test/aub_tests/command_stream/aub_mem_dump_tests.h
opencl/test/unit_test/fixtures/ult_command_stream_receiver_fixture.h
shared/source/os_interface/linux/device_time_drm.h
shared/source/os_interface/linux/os_time_linux.h
Related-To: NEO-5548
Signed-off-by: Warchulski, Jaroslaw <jaroslaw.warchulski@intel.com>
2023-01-27 14:31:05 +01:00
026d50c7b9
refactor: don't use global ProductHelper getter 13/n
...
Related-To: NEO-6853
Signed-off-by: Kamil Kopryk <kamil.kopryk@intel.com>
2023-01-27 13:52:36 +01:00
b101f9f2f5
refactor: don't use global ProductHelper getter 12/n
...
Related-To: NEO-6853
Signed-off-by: Kamil Kopryk <kamil.kopryk@intel.com>
2023-01-27 12:29:18 +01:00
44032a4386
Revert "Flush task at device init"
...
This reverts commit 73bd55ad40
2023-01-27 01:51:55 +01:00
e9c18f8825
Add binding pool heap size parameter to state base address properties
...
Related-To: NEO-5055
Signed-off-by: Zbigniew Zdanowicz <zbigniew.zdanowicz@intel.com>
2023-01-26 18:56:27 +01:00
b7380237c2
Add adjustNumberOfThreadsInThreadGroup method to EncodeDispatchKernel
...
Related-To: NEO-7357
Signed-off-by: Rafal Maziejuk <rafal.maziejuk@intel.com>
2023-01-26 15:25:01 +01:00
Kacper Nowak