Mateusz Jablonski
d2c6283879
fix(ocl): add support for deprecated value of CL_MEM_DEVICE_ID_INTEL
...
Resolves: NEO-10184, NEO-10195, NEO-10196
Resolves: NEO-10200, NEO-10205, NEO-10207
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com >
2024-01-29 17:07:26 +01:00
Lukasz Jobczyk
524ae7713a
refactor: Add GDI profiling
...
Resolves: NEO-9236
Related-To: NEO-10036
Signed-off-by: Lukasz Jobczyk <lukasz.jobczyk@intel.com >
2024-01-29 11:36:04 +01:00
Mateusz Jablonski
327d4c4d48
fix: update OpenCL headers to v2023.12.14
...
https://github.com/KhronosGroup/OpenCL-Headers/releases/tag/v2023.12.14
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com >
2024-01-25 16:42:14 +01:00
Katarzyna Cencelewska
67b0b18be3
fix: osAgnostic path for allocate with alignment
...
Resolves: NEO-9334
Signed-off-by: Katarzyna Cencelewska <katarzyna.cencelewska@intel.com >
2024-01-25 11:03:34 +01:00
Kamil Kopryk
6d1df09e58
fix: fix exported function address when heapless mode is enabled in OCL
...
Related-To: IGC-8472
Signed-off-by: Kamil Kopryk <kamil.kopryk@intel.com >
2024-01-24 16:21:42 +01:00
Chodor, Jaroslaw
9ec9cc7ee0
fix: ocloc - Add support for verbose mode
...
with additional -v option, ocloc will output
command line it was invoked with
Signed-off-by: Chodor, Jaroslaw <jaroslaw.chodor@intel.com >
2024-01-24 15:32:28 +01:00
Dominik Dabek
6e434e0424
performance(ocl): increase buffer pool size
...
increase pool size to 2MB and threshold to 1MB
add limit to the number of pools, set to 2
Related-To: NEO-9690
Signed-off-by: Dominik Dabek <dominik.dabek@intel.com >
2024-01-24 10:55:29 +01:00
Kamil Kopryk
87d13fcb6e
fix: fix scratch programming in heapless mode
...
Related-To: NEO-10107
Signed-off-by: Kamil Kopryk <kamil.kopryk@intel.com >
2024-01-23 15:26:56 +01:00
Mateusz Jablonski
a697a3f718
refactor: create new members for storing spill and private memory in scratch
...
rename private scratch space into scratch space slot 1 as it can be generic
Related-To: NEO-9944
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com >
2024-01-23 12:42:25 +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
Compute-Runtime-Validation
f9f9035b95
Revert "refactor: create new members for storing spill and private memory in ...
...
This reverts commit 87eb5f554a .
Signed-off-by: Compute-Runtime-Validation <compute-runtime-validation@intel.com >
2024-01-23 09:13:00 +01:00
Compute-Runtime-Validation
e949ba7144
Revert "refactor: Add GDI profiling"
...
This reverts commit 8d56f8fb6b .
Signed-off-by: Compute-Runtime-Validation <compute-runtime-validation@intel.com >
2024-01-23 06:13:02 +01:00
Mateusz Jablonski
87eb5f554a
refactor: create new members for storing spill and private memory in scratch
...
rename private scratch space into scratch space slot 1 as it can be generic
Related-To: NEO-9944
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com >
2024-01-22 19:48:48 +01:00
Dunajski, Bartosz
b77e1a6a71
feature: improve reporting max cooperative group count
...
Related-To: NEO-8210
Signed-off-by: Dunajski, Bartosz <bartosz.dunajski@intel.com >
2024-01-22 14:56:15 +01:00
Lukasz Jobczyk
8d56f8fb6b
refactor: Add GDI profiling
...
Resolves: NEO-9236
Related-To: NEO-10036
Signed-off-by: Lukasz Jobczyk <lukasz.jobczyk@intel.com >
2024-01-22 14:24:08 +01:00
Dominik Dabek
f3a3ba5cb2
test: use buffer offset in tests
...
Offset will not be 0 if buffer is from pool.
Related-To: NEO-9690
Signed-off-by: Dominik Dabek <dominik.dabek@intel.com >
2024-01-22 13:32:36 +01:00
Dunajski, Bartosz
6f4ed10919
refactor: unify getMaxWorkGroupCount logic
...
Signed-off-by: Dunajski, Bartosz <bartosz.dunajski@intel.com >
2024-01-22 11:16:45 +01:00
Dominik Dabek
7a09c51f3d
fix(ocl): querying memobj offset
...
Querying buffer offset via clGetMemObjectInfo should not modify its
value.
Related-To: NEO-9690
Signed-off-by: Dominik Dabek <dominik.dabek@intel.com >
2024-01-19 15:48:46 +01:00
Zbigniew Zdanowicz
b5f698e0c5
feature: add umonitor and umwait synchronization function
...
Related-To: NEO-9737
Signed-off-by: Zbigniew Zdanowicz <zbigniew.zdanowicz@intel.com >
2024-01-19 14:20:19 +01:00
Dunajski, Bartosz
b266f1f3cc
refactor: improve implicit scaling interface
...
Signed-off-by: Dunajski, Bartosz <bartosz.dunajski@intel.com >
2024-01-19 14:09:04 +01:00
Kamil Kopryk
f1f426c15b
refactor: correct spelling
...
Signed-off-by: Kamil Kopryk <kamil.kopryk@intel.com >
2024-01-19 08:25:50 +01:00
Dunajski, Bartosz
3a5e6c0117
fix: cl_intel_subgroup_split_matrix_multiply_accumulate support on PVC
...
Related-To: GSD-7696
Signed-off-by: Dunajski, Bartosz <bartosz.dunajski@intel.com >
2024-01-18 19:40:59 +01:00
Naklicki, Mateusz
6d53bd5467
test: expect local id generation depending on kernel flag
...
Related-To: NEO-10014
Signed-off-by: Naklicki, Mateusz <mateusz.naklicki@intel.com >
2024-01-18 15:29:58 +01:00
Dominik Dabek
e0ccf22557
fix: indirect access in external functions
...
Read indirect_stateless_count in module external functions.
If greater than 0, mark all kernels that have the has_stack_calls flag
set from this module as having indirect accesses.
Related-To: NEO-7712
Signed-off-by: Dominik Dabek <dominik.dabek@intel.com >
2024-01-18 15:29:41 +01:00
Dunajski, Bartosz
ec40fdc382
fix: PVC dpas caps
...
Related-To: NEO-10083
Signed-off-by: Dunajski, Bartosz <bartosz.dunajski@intel.com >
2024-01-17 16:28:57 +01:00
Kamil Kopryk
36dcfc0b42
feature: add stateless heapless builtins
...
Related-To: NEO-7621
Signed-off-by: Kamil Kopryk <kamil.kopryk@intel.com >
2024-01-16 11:11:31 +01:00
Kamil Kopryk
ad6e02bf2d
refactor: correct spelling
...
Signed-off-by: Kamil Kopryk <kamil.kopryk@intel.com >
2024-01-16 09:16:44 +01:00
Zbigniew Zdanowicz
c49695fa6e
fix: flush L3 cache and wait for flush to complete when event is completed
...
Related-To: NEO-8395
Signed-off-by: Zbigniew Zdanowicz <zbigniew.zdanowicz@intel.com >
2024-01-16 07:05:35 +01:00
Mateusz Jablonski
556645e0c5
fix: add unrecoverable to avoid OOB access
...
Related-To: NEO-9860
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com >
2024-01-16 00:12:11 +01:00
Kamil Kopryk
ef8921247c
refactor: improve constexpr array usage
...
Gcc with o2 flag will allocate constexpr array on stack.
Static constexpr std::array will be initialized at compile time.
Signed-off-by: Kamil Kopryk <kamil.kopryk@intel.com >
2024-01-15 17:00:50 +01:00
Mateusz Jablonski
f764b6b126
test: reduce number of iterations in mt test
...
Related-To: NEO-9803
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com >
2024-01-15 10:44:30 +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
Kamil Kopryk
f99edf72e1
refactor: Add common HardwareCommandsHelper enablers
...
Signed-off-by: Kamil Kopryk <kamil.kopryk@intel.com >
2024-01-12 09:39:12 +01:00
Katarzyna Cencelewska
5d2d3ed899
refactor: modify sip kernel helpers
...
Signed-off-by: Katarzyna Cencelewska <katarzyna.cencelewska@intel.com >
2024-01-09 17:09:10 +01:00
Mateusz Jablonski
a73fb4d2fe
fix: correct reporing kernel private size on L0
...
unify the logic across APIs
Related-To: NEO-9944
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com >
2024-01-09 09:54:05 +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
Dominik Dabek
af1620a308
fix(ocl): allocation info from pool svm ptr
...
Fix querying allocation info from pooled svm ptr.
Handle requested allocation alignment.
Refactor sorted vector usage.
Do not associate device with host pool allocation.
Related-To: NEO-9700
Signed-off-by: Dominik Dabek <dominik.dabek@intel.com >
2024-01-05 15:20:01 +01:00
Kamil Kopryk
5432b57fd0
refactor: improve readability
...
Added a time unit to the variable name to make it clear.
Removed incorrect and not needed check.
Signed-off-by: Kamil Kopryk <kamil.kopryk@intel.com >
2024-01-05 14:53:52 +01:00
Dunajski, Bartosz
57900c5455
refactor: improve single slice dispatch mode handling
...
Related-To: NEO-8210
Signed-off-by: Dunajski, Bartosz <bartosz.dunajski@intel.com >
2024-01-04 15:24:09 +01:00
Kamil Kopryk
2a46350ba3
fix: fix programming indirect data pointer in heapless mode
...
Indirect data pointer should be offsetted by used offset.
Related-To: NEO-7621
Signed-off-by: Kamil Kopryk <kamil.kopryk@intel.com >
2024-01-04 07:51:31 +01:00
Dunajski, Bartosz
87857a6558
feature: add new PVC device id
...
Related-To: NEO-9736
Signed-off-by: Dunajski, Bartosz <bartosz.dunajski@intel.com >
2023-12-29 19:06:33 +01:00
Compute-Runtime-Validation
5535ef3049
Revert "performance(ocl): enable usm pool allocator"
...
This reverts commit 7bc8424a69 .
Signed-off-by: Compute-Runtime-Validation <compute-runtime-validation@intel.com >
2023-12-29 05:54:07 +01:00
Mateusz Hoppe
31e9b5e9fa
feature: add support for secondary contexts in group
...
Related-To: NEO-7824
Signed-off-by: Mateusz Hoppe <mateusz.hoppe@intel.com >
2023-12-28 13:31:08 +01:00
Dominik Dabek
7bc8424a69
performance(ocl): enable usm pool allocator
...
Enable opencl usm pool allocator by default
Related-To: NEO-9700
Signed-off-by: Dominik Dabek <dominik.dabek@intel.com >
2023-12-28 13:14:41 +01:00
Dominik Dabek
d238a68bae
fix(ocl): usm pool allocator correct size
...
Wrong debug flag was used for setting host allocation pool size
Related-To: NEO-9700
Signed-off-by: Dominik Dabek <dominik.dabek@intel.com >
2023-12-27 23:14:28 +01:00
Dunajski, Bartosz
df66a0276f
refactor: remove not used logic to check dynamic postsync layout
...
Related-To: NEO-8210
Signed-off-by: Dunajski, Bartosz <bartosz.dunajski@intel.com >
2023-12-27 13:12:11 +01:00
Dominik Dabek
2fe3804cc2
performance(ocl): add usm allocation pooling flag
...
EnableDeviceUsmAllocationPool and EnableHostUsmAllocationPool for device
and host allocations respectively.
Pool size will be set to flag value * MB.
Allocation size threshold to be pooled is 1MB.
Pools are created per context.
Related-To: NEO-9700
Signed-off-by: Dominik Dabek <dominik.dabek@intel.com >
2023-12-27 11:42:01 +01:00
Dominik Dabek
239831f097
fix: return error for 0 size usm allocations
...
according to both level zero and opencl specs, usm allocations with
size=0 should return invalid/unsupported buffer size errors
Signed-off-by: Dominik Dabek <dominik.dabek@intel.com >
2023-12-22 17:26:39 +01:00
Mateusz Jablonski
89c58b8371
test: reduce scope of debug message printing in test
...
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com >
2023-12-22 09:17:33 +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