Media/compute-runtime
1
0
Fork 0
mirror of https://github.com/intel/compute-runtime.git synced 2025-09-15 13:01:45 +08:00
Code Issues Packages Projects Releases Wiki Activity

Add note about using metric query with immediate command list

Browse Source 
Related-To: LOCI-3518

Signed-off-by: Joshua Santosh Ranjan <joshua.santosh.ranjan@intel.com>
This commit is contained in:
Joshua Santosh Ranjan
2022-12-12 12:23:29 +00:00
committed by Compute-Runtime-Automation
parent a526b91262
commit a60b5898dc
1 changed files with 6 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
programmers-guide/IMMEDIATE_COMMANDLIST.md
View File

@ -10,8 +10,9 @@ SPDX-License-Identifier: MIT
* [Introduction](#Introduction)
* [Availability](#Availability)
* [Optimizations](#Optimizations)
* [Debug Keys](#Debug-Keys)
* [Limitations](#Limitations)
* [Notes](#Notes)
* [References](#References)
# Introduction
@ -101,6 +102,10 @@ For more code samples, please refer [compute-benchmarks](https://github.com/inte
* `EnableFlushTaskSubmission=0/1` : Force enable/disable support for using optimizations in non-pipelined state filtering for immediate command lists. Defaults to 1.
* `EnableImmediateCmdListHeapSharing=0/1` : Force enable/disable support for using CSR heap resources for immediate command list. Defaults to 1 from Xe_HPG/Xe_HPC onwards. When enabled, all immediate command lists created against same ordinal share GPU heaps instead of allocating separate heaps for each command list.
# Notes
* When using L0 Metrics in Query mode to capture execution metrics of an immediate command list, the time elapsed between zetCommandListAppendMetricQueryBegin and zetCommandListAppendMetricQueryEnd points may include time elapsed in the host as well.
# References
* https://one-api.gitlab-pages.devtools.intel.com/level_zero/core/PROG.html#low-latency-immediate-command-lists