Commit Graph

58 Commits

Author SHA1 Message Date
b6e2d2df8b Add multi return points command lists for front end reconfiguration
This change gives fine grain control over front end configuration for each
kernel.
As it gives possible to inject FE command in command queue and return to exact
place in command list.
Programming commands in queue makes patching commands in command lists
not needed as that operation is costly.
And it allows to program context information for each command list too.

Related-To: NEO-5019

Signed-off-by: Zbigniew Zdanowicz <zbigniew.zdanowicz@intel.com>
2022-09-07 12:53:24 +02:00
4da6f65d1a L0Debug - Notify cmdQueue create for subdevices
- PROCESS ENTRY triggered for first cmdQ created

Related-To: NEO-5784
Signed-off-by: Mateusz Hoppe <mateusz.hoppe@intel.com>
2022-07-22 16:08:41 +02:00
5199ea257d Revert "L0Debug - Notify cmdQueue create for subdevices"
This reverts commit ebcb3faad2.

Signed-off-by: Compute-Runtime-Validation <compute-runtime-validation@intel.com>
2022-07-21 18:50:20 +02:00
ebcb3faad2 L0Debug - Notify cmdQueue create for subdevices
- PROCESS ENTRY triggered for first cmdQ created

Related-To: NEO-5784

Signed-off-by: Mateusz Hoppe <mateusz.hoppe@intel.com>
2022-07-20 15:13:06 +02:00
5956aea18d Limit header includes from level_zero device.h
- remove including debugger_l0.h from device.h
- add getL0Debugger() to shared NEO Device

Signed-off-by: Mateusz Hoppe <mateusz.hoppe@intel.com>
2022-07-06 16:41:17 +02:00
620bb970f4 Resubmit fix for task count hang
Signed-off-by: John Falkowski <john.falkowski@intel.com>
2022-06-07 16:28:17 +02:00
747c2d3d83 Cleanup includes in cmdlist, cmdqueue, context, device, driver and module
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
2022-06-03 02:54:41 +02:00
2a4c68dc38 Remove not needed dependencies from device_imp.h
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
2022-06-02 18:35:34 +02:00
359e848dbe Revert "Fix for task count hang issue"
This reverts commit 340ba8bf56.

Signed-off-by: Jaime Arteaga <jaime.a.arteaga.molina@intel.com>
2022-05-25 20:34:00 +02:00
340ba8bf56 Fix for task count hang issue
Signed-off-by: John Falkowski <john.falkowski@intel.com>
2022-05-18 19:35:12 +02:00
b5b8762faf refactor direct submission: Pass command stream receiver to ctor
Related-To: NEO-6643

Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
2022-04-19 23:02:55 +02:00
19dded25ef Detect GPU hang in remaining calls of command queue and list
This change introduces checking of waits status in
CommandQueue and CommandList classes.

Related-To: NEO-6681
Signed-off-by: Patryk Wrobel <patryk.wrobel@intel.com>
2022-04-11 13:31:44 +02:00
a230f267e1 Poll task count indefinitely on high throttle command queue
Resolves: NEO-6781

Signed-off-by: Lukasz Jobczyk <lukasz.jobczyk@intel.com>
2022-03-25 10:06:16 +01:00
7f729b7f89 Detect GPU hang in clWaitForEvents
This change:
- moves NEO::WaitStatus to a separate file
- enables detection of GPU hang in clWaitForEvents
- adjusts most of blocking calls in CommandStreamReceiver to return WaitStatus
- adds ULTs to cover the new code

Related-To: NEO-6681
Signed-off-by: Patryk Wrobel <patryk.wrobel@intel.com>
2022-02-23 13:33:09 +01:00
60d6505932 L0 debugger - reports process ENTER/EXIT events for zeCommandQueues
- PROCESS_ENTRY - triggered by first zeCommandQueueCreate()
- PROCESS_EXIT  - triggered by last zeCommandQueueDestroy()

Resolves: NEO-6503

Signed-off-by: Igor Venevtsev <igor.venevtsev@intel.com>
2022-02-16 16:28:33 +01:00
4b0d986876 Move AllocationType enum out of GraphicsAllocation class
Signed-off-by: Bartosz Dunajski <bartosz.dunajski@intel.com>
2022-02-04 17:49:09 +01:00
498cf5e871 Implement GPU hang detection
This change uses DRM_IOCTL_I915_GET_RESET_STATS to detect
GPU hangs. When such situation is encountered, then
zeCommandQueueSynchronize returns ZE_RESULT_ERROR_DEVICE_LOST.

Related-To: NEO-5313
Signed-off-by: Patryk Wrobel <patryk.wrobel@intel.com>
2022-01-31 13:48:17 +01:00
394c0e90e1 Return error when failing on submission
Signed-off-by: Raiyan Latif <raiyan.latif@intel.com>
2022-01-12 16:42:30 +01:00
02f075c541 Power Saving Hint Support for Level Zero in Windows
- Added Functionality to pass ze_power_saving_hint_type_t to zeContextCreate
included in the pNext extensions in ze_context_desc_t.
- Enables handling a hint value 0-100 with 0 being no power savings
and 100 being maximum power savings.
- ZE_RESULT_ERROR_INVALID_ENUMERATION is returned given an invalid hint.

Related-To: LOCI-2567

Signed-off-by: Spruit, Neil R <neil.r.spruit@intel.com>
2022-01-06 23:56:59 +01:00
b2f286fc4a Reuse command buffers in L0 command queue
Signed-off-by: Lukasz Jobczyk <lukasz.jobczyk@intel.com>
2021-12-09 15:52:36 +01:00
f56773d166 Setup partition registers once per context
Related-To: NEO-6262

Signed-off-by: Zbigniew Zdanowicz <zbigniew.zdanowicz@intel.com>
2021-11-22 22:41:09 +01:00
7d109c42bc Do not execute multi-tile command list on single-tile command queue
Signed-off-by: Zbigniew Zdanowicz <zbigniew.zdanowicz@intel.com>
2021-11-03 13:27:49 +01:00
0859f99d64 Enable zeDeviceCanAccessPeer
- Propagate return value up from Batch Buffer exec
- Add zeDeviceCanAccessPeer functionality

Signed-off-by: Raiyan Latif <raiyan.latif@intel.com>
2021-10-12 09:38:22 +02:00
3b35ba504f Adapt command stream receiver to multiple active partitions
Related-To: NEO-6244

Signed-off-by: Zbigniew Zdanowicz <zbigniew.zdanowicz@intel.com>
2021-09-23 14:32:20 +02:00
a2e41716e2 Fix L0 CommandQueue command buffer allocation properties
Signed-off-by: Bartosz Dunajski <bartosz.dunajski@intel.com>
2021-09-21 16:33:48 +02:00
f6c8fb47bb Delete unused L0 function
Signed-off-by: Rafal Maziejuk <rafal.maziejuk@intel.com>
Related-To: NEO-4541
2021-09-13 14:22:59 +02:00
cd4f3c221a Synchronize switching command buffers for all partitions
Signed-off-by: Zbigniew Zdanowicz <zbigniew.zdanowicz@intel.com>
2021-09-08 12:12:23 +02:00
6b299a3ab0 Make partitioned post sync operations for partitioned workloads
Signed-off-by: Zbigniew Zdanowicz <zbigniew.zdanowicz@intel.com>
2021-09-03 20:20:29 +02:00
29c64c3dd0 Disable implicit scaling for cooperative kernels
When implicit scaling is disabled use useSingleSubdeviceValue = true.

Resolves: NEO-5757

Signed-off-by: Filip Hazubski <filip.hazubski@intel.com>
2021-08-18 14:56:37 +02:00
7b52f587da Disable timeout for inifinite synchronization time
Related-To: NEO-5845

Signed-off-by: Zbigniew Zdanowicz <zbigniew.zdanowicz@intel.com>
2021-08-03 16:34:07 +02:00
b454bcbfe7 Make KMD wait function non default and available under debug key
Related-To: NEO-5845

Signed-off-by: Zbigniew Zdanowicz <zbigniew.zdanowicz@intel.com>
2021-08-03 11:24:43 +02:00
549b458e2f Add UNRECOVERABLE_IF when bufferAllocation is nullptr
Signed-off-by: Mateusz Hoppe <mateusz.hoppe@intel.com>
2021-07-30 09:59:36 +02:00
1b4ceb3d11 Use KMD wait function for infinite timeout synchronization
Related-To: NEO-5845

Signed-off-by: Zbigniew Zdanowicz <zbigniew.zdanowicz@intel.com>
2021-07-22 19:35:36 +02:00
9e50138829 Defer ULLS initialization
Signed-off-by: Lukasz Jobczyk <lukasz.jobczyk@intel.com>
2021-07-21 15:55:16 +02:00
bcd9b2f982 Add debug flag to override command queue synchronous mode
Signed-off-by: Zbigniew Zdanowicz <zbigniew.zdanowicz@intel.com>
2021-07-02 13:03:02 +02:00
fffa22b17d Enhance wait method in L0 command queue
Signed-off-by: Lukasz Jobczyk <lukasz.jobczyk@intel.com>
2021-06-30 16:41:54 +02:00
3b41378055 Update command buffer task count in L0
Signed-off-by: Lukasz Jobczyk <lukasz.jobczyk@intel.com>
2021-06-29 16:29:23 +02:00
e1bcad51f0 Fix makeNonResident for csr residency allocations
Signed-off-by: Maciej Plewka <maciej.plewka@intel.com>
2021-06-10 20:57:15 +02:00
8abbc69fef Revert "Fix update residency task count for all submited allocations"
This reverts commit 2565223a05e410a9113dbf69dd92f68d98578329.

Signed-off-by: Maciej Plewka <maciej.plewka@intel.com>
2021-06-10 08:41:13 +02:00
925d8ad1eb Fix update residency task count for all submited allocations
Signed-off-by: Maciej Plewka <maciej.plewka@intel.com>
2021-06-07 11:54:00 +02:00
3bd4bca911 Copyright header update
Dates corrected in copyright headers to reflect original publication date
(2018 for OpenCL, 2020 for Level Zero).

Signed-off-by: lgotszal <lukasz.gotszald@intel.com>
2021-05-17 20:38:19 +02:00
f5b253207d Fix osContext initialization for L0 immediate command lists
Signed-off-by: Maciej Dziuban <maciej.dziuban@intel.com>
2021-05-17 15:05:05 +02:00
edc0a93927 Add preemption synchronization flag to command queue
Related-To: NEO-5778

Signed-off-by: Zbigniew Zdanowicz <zbigniew.zdanowicz@intel.com>
2021-04-28 13:14:18 +02:00
671d916c70 Add Debugger specific log print macros
- use DebuggerLogBitmask while printing logs for SBA tracking

Signed-off-by: Mateusz Hoppe <mateusz.hoppe@intel.com>
2021-04-02 15:00:55 +02:00
343fd602fa Add method to submit kernel on single subdevice
Signed-off-by: Jobczyk, Lukasz <lukasz.jobczyk@intel.com>
2020-12-01 22:15:05 +01:00
240563099c Clean up code for nullptr device check during cmdlist create
Signed-off-by: Vinod Tipparaju <vinod.tipparaju@intel.com>
2020-11-19 04:24:34 +01:00
b384bda99a Exclude internally used cmdLists and cmdQs from debugging
Related-To: NEO-5241

Signed-off-by: Mateusz Hoppe <mateusz.hoppe@intel.com>
2020-11-17 12:17:34 +01:00
bc1e772ec2 Set correct device bitfield for L0 command queue allocations
Signed-off-by: Bartosz Dunajski <bartosz.dunajski@intel.com>
2020-11-13 12:37:16 +01:00
f553e1e514 Return "OUT_OF_MEMORY" when gfx alloc on device fails during cmdqueue create
Signed-off-by: Vinod Tipparaju <vinod.tipparaju@intel.com>
2020-11-10 06:26:02 +01:00
083edd1a3e Thread safety violation fixes - Events & USM
Change-Id: I57de59f204d9fd4407be768d0b14bf579dae967d
Signed-off-by: Vinod Tipparaju <vinod.tipparaju@intel.com>
2020-09-03 08:54:34 +02:00