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>
mechanism for freeing allocations saved for reuse that have not been
used in a given time
Related-To: NEO-13425
Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
Calculate available memory for usm device reuse based as (total device
memory - used memory) * fraction for reuse.
Use sys mem allocs for devices without local memory.
Related-To: NEO-12902
Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
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>
Allocations over a certain size will be checked for memory utilization
when chosen for reuse.
If utilization is below a threshold, they will not be reused.
Related-To: NEO-6893
Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
Related-To: GSD-9385
In case of indirect allocations, we don't really know
their task count because we can't track their true usage
on GPU.
In case of non-blocking free, don't wait for latestSentTaskCount.
Signed-off-by: Szymon Morek <szymon.morek@intel.com>
Do not put into usm reuse if is internal.
Set new isInternalAllocation flag for globals allocations.
Use actual size on device for tracking memory usage.
Related-To: NEO-6893
Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
Do not put into usm reuse if is internal.
Set new isInternalAllocation flag for globals allocations.
Use actual size on device for tracking memory usage.
Related-To: NEO-6893
Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
Allocating global surface is expecting that the usm allocation is zeroed
out. Reusing allocations can be filled with junk data and this caused
errors.
Resolves: HSD-18038551036, HSD-18038551766, HSD-18038551957, HSD-18038552252
Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
Query system total memory size and limit usm host allocation recycle to
use at most x%.
x is read from ExperimentalEnableDeviceAllocationCache for device and
ExperimentalEnableHostAllocationCache for host.
Related-To: GSD-7497
Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
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>
Move code out to base class. This will allow to use the sorted vector
class with different values than only SvmData.
Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
When creating shared USM, currently default root device index
is used when accessing memoryManager.
This change fixes this issue, by using device provided by caller.
In case device is not provided, then default root device index
could be used.
Related-To: LOCI-4474
Signed-off-by: Jitendra Sharma <jitendra.sharma@intel.com>
Related-To: LOCI-4381
- Enabled support for customers to use full Virtual reservation range
with multiple physical mappings with additional allocations implicitly
included in residency.
- Buffer Surface state size extended for first allocation to stretch to
the bufferSize requested.
Signed-off-by: Neil R Spruit <neil.r.spruit@intel.com>
Related-To: LOCI-4176
- Given a Base Pointer passed into Get Peer Allocation, then the base
pointer is used in the map of the new allocation to the virtual memory.
- Enables users to use the same pointer for all devices in Peer To Peer.
- Currently unsupported on reserved memory due to mapped and exec
resiedency of Virtual addresses.
Signed-off-by: Neil R Spruit <neil.r.spruit@intel.com>
Allows the user to use alignments > 64KB in `createUnifiedMemoryAllocation`
So that the restriction in `piextUSMDeviceAlloc` of the DPC++ runtime
could be lifted
Related-To: LOCI-4168
Signed-off-by: Lu, Wenbin <wenbin.lu@intel.com>
Extended the regkey ForceMemoryPrefetchForKmdMigratedSharedAllocations
to force meory prefetch of kmd-migrated shared allocation
in clEnqueueNDRangeKernel(), clEnqueueMemFillINTEL, ...
Related-To: NEO-7841
Signed-off-by: Milczarek, Slawomir <slawomir.milczarek@intel.com>
Globals surface allocation via USM manager will have correct allocation
type set (instead of just BUFFER) and will use cpu copy when possible.
Related-To: NEO-7796
Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
Allows the user to use alignments > 64KB in `createUnifiedMemoryAllocation`
So that the restriction in `piextUSMDeviceAlloc` of the DPC++ runtime
could be lifted
Related-To: LOCI-4168
Signed-off-by: Lu, Wenbin <wenbin.lu@intel.com>
Dominik Dabek