Commit Graph

169 Commits

Author SHA1 Message Date
Chandio, Bibrak Qamar bed2ad1ac0 feature: memAdise for System Allocator
Related-To: NEO-13715

Implements basic functionality for zeCommandListAppendMemoryAdvise for
system allocator

Signed-off-by: Chandio, Bibrak Qamar <bibrak.qamar.chandio@intel.com>
2025-04-16 07:11:14 +02:00
Dominik Dabek 915d657420 fix: flag to limit usm reuse based on memory usage
Host usm and device usm for igfx checks system memory usage.
Device usm for dgfx checks local memory usage.

If used memory is above limit threshold:
- no new allocations will be saved for reuse
- cleaner will use shorter hold time of 2 seconds
- cleaner will free all eligible allocations, regardless of async
deleter thread having work

Motivation: in case of gfx memory being full, making resident new
allocations will require evictions which leads to massive slowdown on
enqueue calls.
This change aims to minimize cases where extra memory usage from usm
reuse mechanism leads to above situation.

Related-To: NEO-6893, NEO-14160

Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
2025-03-27 10:25:19 +01:00
Dominik Dabek 6e998fc3c1 fix: move host usm reuse max size to mem manager
Intialize value on memory manager creation.

Related-To: NEO-6893

Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
2025-03-24 08:53:30 +01:00
John Falkowski 4d281cf51d feature: Implement appendMemoryPrefetch for Shared System USM allocations
Related-To: NEO-12989

Signed-off-by: John Falkowski <john.falkowski@intel.com>
2025-03-13 06:26:38 +01:00
Compute-Runtime-Validation fa2e3adad3 Revert "feature: Implement appendMemoryPrefetch for Shared System USM Allocat...
This reverts commit 97799b3faf.

Signed-off-by: Compute-Runtime-Validation <compute-runtime-validation@intel.com>
2025-03-12 05:55:32 +01:00
John Falkowski 97799b3faf feature: Implement appendMemoryPrefetch for Shared System USM Allocations
Related-To: NEO-12989

Signed-off-by: John Falkowski <john.falkowski@intel.com>
2025-03-11 09:12:48 +01:00
Lukasz Jobczyk 9085f6aeca fix: drain gem close worker on csr destroy
Resolves: HSD-18041493278

Signed-off-by: Lukasz Jobczyk <lukasz.jobczyk@intel.com>
2025-02-13 20:34:13 +01:00
Maciej Bielski 971b7c27a2 fix: enable usm compression on linux
Related-To: NEO-12056
Signed-off-by: Maciej Bielski <maciej.bielski@intel.com>
2025-02-04 13:09:04 +01:00
Lukasz Jobczyk 2dd9940f60 Revert "fix: count active modules for enabling per-dispatch private memory"
This reverts commit a483b361f9.

Signed-off-by: Lukasz Jobczyk <lukasz.jobczyk@intel.com>
2025-01-15 15:03:37 +01:00
Wenbin Lu a483b361f9 fix: count active modules for enabling per-dispatch private memory
Related-To: NEO-13086

Signed-off-by: Wenbin Lu <wenbin.lu@intel.com>
2025-01-10 15:03:34 +01:00
Jack Myers 7f9fadc314 fix: regression caused by tbx fault mngr
Addresses regressions from the reverted merge
of the tbx fault manager for host memory.

Recursive locking of mutex caused deadlock.

To fix, separate tbx fault data from base
cpu fault data, allowing separate mutexes
for each, eliminating recursive locks on
the same mutex.

By separating, we also help ensure that tbx-related
changes don't affect the original cpu fault manager code
paths.

As an added safe guard preventing critical regressions
and avoiding another auto-revert, the tbx fault manager
is hidden behind a new debug flag which is disabled by default.

Related-To: NEO-12268
Signed-off-by: Jack Myers <jack.myers@intel.com>
2025-01-09 07:48:53 +01:00
Compute-Runtime-Validation 124e755b9d Revert "fix: regression caused by tbx fault mngr"
This reverts commit 9a14fe2478.

Signed-off-by: Compute-Runtime-Validation <compute-runtime-validation@intel.com>
2024-12-19 17:35:03 +01:00
Jack Myers 9a14fe2478 fix: regression caused by tbx fault mngr
Addresses regressions from the reverted merge
of the tbx fault manager for host memory.

This fixes attempts by the tbx fault manager
to protect/unprotect host buffer memory, even
if the host ptr was not driver-allocated.

In the case of the smoke test that triggered
the critical regression, clCreateBuffer was
called with the CL_MEM_USE_HOST_PTR flag.
The subsequent `mprotect` calls on the
provided host ptr then failed.

Related-To: NEO-12268
Signed-off-by: Jack Myers <jack.myers@intel.com>
2024-12-18 23:16:36 +01:00
Filip Hazubski a0cc124b2e performance: Pass RootDeviceIndicesContainer by reference
Additionally pass std::map by reference in UsmMemAllocPoolsManager c-tor.

Signed-off-by: Filip Hazubski <filip.hazubski@intel.com>
2024-12-17 14:18:30 +01:00
Mateusz Hoppe 0589a70dc7 feature(sysman): reinitialize gfxPartition on reset
Related-To: NEO-13203

Signed-off-by: Mateusz Hoppe <mateusz.hoppe@intel.com>
2024-12-12 16:27:25 +01:00
Compute-Runtime-Validation 6c5d9a6ed7 Revert "feature: extend TBX page fault manager from CPU implementation"
This reverts commit 51c0e80299.

Signed-off-by: Compute-Runtime-Validation <compute-runtime-validation@intel.com>
2024-12-12 12:30:22 +01:00
Jack Myers 51c0e80299 feature: extend TBX page fault manager from CPU implementation
In TBX mode, the host could not write to host buffers after access from device
code due to the lack of a migration mechanism post-initial TBX upload.
Migration is unnecessary with real hardware, but required for TBX.

This patch introduces a new page fault manager type that extends the original
CPU fault manager, enabling automatic migration of host buffers in TBX mode.

Refactoring was necessary to avoid diamond inheritance, achieved by using a
template parameter as the base class for OS-specific fault managers.

Related-To: NEO-12268
Signed-off-by: Jack Myers <jack.myers@intel.com>
2024-12-11 09:09:50 +01:00
Fabian Zwoliński d2ce3badfc fix: bindlessHeapsHelper handle unavailable external heap
This PR handles the situation in which a component
has reserved a front window space for itself in the external heap,
so that the Compute Runtime cannot access this area.

In such a situation, we perform the following steps:
1. reserve 4GB chunk in heapStandard
2. split our chunk into 2 parts: heapFrontWindow, heapRegular
3. from this point on, map all linearStream allocations in reserved 4GB
chunk

Patch applies to Windows and WSL.
Patch only applies when the bindless global allocator is enabled.

Related-To: HSD-16025889919
Signed-off-by: Fabian Zwoliński <fabian.zwolinski@intel.com>
2024-12-05 14:18:01 +01:00
Dominik Dabek e55aa958b7 fix: track usm reuse usage in multiple contexts
Add tracking of memory used for usm reuse mechanism when multiple cl
contexts are used.
Tracking for device added to NEO::Device, for host added to
NEO::MemoryManager.

This fixes usm reuse using x% of memory per each context instead of
globally.

Related-To: NEO-13308

Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
2024-11-26 16:00:45 +01:00
Wenbin Lu 2ba80ce114 feature: support physical host memory
Related-To: NEO-11981

Signed-off-by: Wenbin Lu <wenbin.lu@intel.com>
2024-11-20 08:19:52 +01:00
Alicja Lukaszewicz 654fdc1345 feature: add query for additional device properties
Related-To: NEO-12590

Signed-off-by: Alicja Lukaszewicz <alicja.lukaszewicz@intel.com>
2024-10-29 20:40:27 +01:00
Wenbin Lu a8a40d2afd feature: support SVM heap in reserveVirtualMem
Related-To: NEO-11981

Signed-off-by: Wenbin Lu <wenbin.lu@intel.com>
2024-10-22 16:47:14 +02:00
Bartosz Dunajski f430279d62 fix: change doorbell type
Related-To: NEO-12870

Signed-off-by: Bartosz Dunajski <bartosz.dunajski@intel.com>
2024-10-14 14:58:11 +02:00
Mateusz Hoppe 5ae2552b4b fix: track shifted contextIds in bitset in bindlessHeapsHelper
- bitset is 64 bit in size, context ids may go beyond that limit
when multiple devices are available
- this change subtracts contextId of first context for a given root
device - tracked state dirty contexts ids are now zero-based

Resolves: GSD-10025

Signed-off-by: Mateusz Hoppe <mateusz.hoppe@intel.com>
2024-10-09 10:32:29 +02:00
Compute-Runtime-Validation 41df1a6f47 Revert "feature: support SVM heap in reserveVirtualMem"
This reverts commit bfaeeb01d6.

Signed-off-by: Compute-Runtime-Validation <compute-runtime-validation@intel.com>
2024-10-03 14:53:50 +02:00
Michal Mrozek dd631610b3 refactor: move memory tracking to memory manager
- remove wddm specific code
- improve total size reported to be in decimal

Signed-off-by: Michal Mrozek <michal.mrozek@intel.com>
2024-09-12 17:32:38 +02:00
Wenbin Lu bfaeeb01d6 feature: support SVM heap in reserveVirtualMem
Related-To: NEO-11981

Signed-off-by: Wenbin Lu <wenbin.lu@intel.com>
2024-09-09 23:22:04 +02:00
Compute-Runtime-Validation 99f62ac866 Revert "feature: support SVM heap in reserveVirtualMem"
This reverts commit 93cde3ee12.

Signed-off-by: Compute-Runtime-Validation <compute-runtime-validation@intel.com>
2024-09-03 20:20:25 +02:00
Wenbin Lu 93cde3ee12 feature: support SVM heap in reserveVirtualMem
Related-To: NEO-11981

Signed-off-by: Wenbin Lu <wenbin.lu@intel.com>
2024-09-03 11:38:51 +02:00
Bartosz Dunajski bca8478bd4 feature: use iotclhelper to get num media engines
Related-To: NEO-9777

Signed-off-by: Bartosz Dunajski <bartosz.dunajski@intel.com>
2024-08-08 09:36:28 +02:00
Dominik Dabek 26428d5af3 feature: track used memory by allocations
Track memory used by memory allocations. System and local per device.
Will be used for heuristics in memory pooling.

Related-To: NEO-11356

Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
2024-08-02 17:54:34 +02:00
Szymon Morek b03ac6abd1 fix: disable usm compression on linux
Related-To: NEO-12047

Signed-off-by: Szymon Morek <szymon.morek@intel.com>
2024-07-16 14:59:33 +02:00
Szymon Morek 432ecbc8f4 fix: disable compression for exported allocations
Related-To: NEO-12021

Signed-off-by: Szymon Morek <szymon.morek@intel.com>
2024-07-16 05:41:14 +02:00
Bartosz Dunajski bb21d25ed0 feature: initial support for creating media context
Related-To: NEO-11386

Signed-off-by: Bartosz Dunajski <bartosz.dunajski@intel.com>
2024-07-03 12:23:31 +02:00
Jaroslaw Warchulski daec249430 refactor: remove not needed createGraphicsAllocationFromNTHandle method
Related-To: NEO-11498
Signed-off-by: Jaroslaw Warchulski <jaroslaw.warchulski@intel.com>
2024-06-18 14:25:00 +02:00
Jaroslaw Warchulski 76a05c1cab refactor: pass arrayIndex to Wddm::openNTHandle function
Related-To: NEO-11498
Signed-off-by: Jaroslaw Warchulski <jaroslaw.warchulski@intel.com>
2024-06-12 12:11:35 +02:00
Jaroslaw Warchulski 0ac1be7669 refactor: pass arrayIndex to Wddm::openSharedHandle function
Related-To: NEO-11498
Signed-off-by: Jaroslaw Warchulski <jaroslaw.warchulski@intel.com>
2024-06-07 09:51:26 +02:00
Maciej Bielski ec21260fda refactor: extract StorageInfo::memoryBanks computation
Move the code determining the memoryBanks parameter into separate
virtual function for further memory-manager-specific customization.

Related-To: NEO-9754
Signed-off-by: Maciej Bielski <maciej.bielski@intel.com>
2024-06-05 17:50:55 +02:00
Bartosz Dunajski 64af8dd956 feature: allocating interrupt support
Related-To: NEO-8179

Signed-off-by: Bartosz Dunajski <bartosz.dunajski@intel.com>
2024-06-03 18:35:24 +02:00
Maciej Bielski dc32da2507 feature: prepare StorageInfo for Drm-specific customization
Get the local memory regions count from the primary source (MemoryInfo)
and store for further use when using DrmMemoryManager.
Add a point of dispatch (virtual `createStorageInfoFromProperties`) for
further Drm-specific customizations related to StorageInfo. As the
function became virtual, move one of its callers
(`isLocalMemoryUsedForIsa()`) from the constructor of `MemoryManager` to
respective constructors of all derivative classes. This prevents
bypassing the virtual call dispatch.

Related-To: NEO-9754
Signed-off-by: Maciej Bielski <maciej.bielski@intel.com>
2024-05-16 12:24:48 +02:00
Szymon Morek d708c5f300 performance: Allow indirect allocs as pack on windows
Related-To: NEO-11228

Signed-off-by: Szymon Morek <szymon.morek@intel.com>
2024-05-06 15:03:12 +02:00
Mateusz Hoppe d6675b14b3 fix: release secondaryContexts prior to primary context
Related-To: NEO-10920

Signed-off-by: Mateusz Hoppe <mateusz.hoppe@intel.com>
2024-04-05 16:34:39 +02:00
Maciej Plewka 564e0f0319 performance: Align host mem to 2MB when range is not limited
Signed-off-by: Maciej Plewka <maciej.plewka@intel.com>

Related-To: NEO-10217
2024-01-30 14:43:13 +01:00
Mateusz Jablonski 7b40b01f54 feature: add debug key for toggling bit in 57bit GPU VA for specific allocations
Related-To: NEO-9419
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
2024-01-15 19:37:00 +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
Mateusz Jablonski b182917d9d refactor: correct naming of allocation types
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
2023-12-11 16:23:37 +01:00
Mateusz Hoppe 97faeae16f feature: heapAssigner per root device
- create heapAssigner per root device in memory manager to allow per
device config

Related-To: NEO-7063

Signed-off-by: Mateusz Hoppe <mateusz.hoppe@intel.com>
2023-10-19 19:52:24 +02:00
Spruit, Neil R 456f07212d fix: Return pageSize/Heap for reservation based on the size and devices
Related-To: NEO-9127,NEO-9142

- Change the heap/pageSize used based on the size of the reservation
required and the device heaps available.
- Return the page size required based on the size requested by the user
for virtual reservation and the devices on the system.
- Check the size passed in by the user in memory map and physical memory
allocation is valid for the heap and page size that is required.

Signed-off-by: Spruit, Neil R <neil.r.spruit@intel.com>
2023-10-19 10:44:13 +02:00
John Falkowski 56f05303c9 feature: add support for zeMemGet/SetAtomicAccessAttributeExp
Resolves: NEO-8219

Signed-off-by: John Falkowski <john.falkowski@intel.com>
2023-10-02 15:59:17 +02:00
Mateusz Hoppe 371412a826 refactor: remove useLocalMem from allocate32BitGraphicsMemoryImpl
- this method allocates System Memory
- argument is not needed - ExternalHeap is selected inside this function
- remove unneeded ults
- allocate memory in Device Pool for external heap allocation in
OsAgnosticMemoryManager

Signed-off-by: Mateusz Hoppe <mateusz.hoppe@intel.com>
2023-09-18 10:08:09 +02:00