-
Notifications
You must be signed in to change notification settings - Fork 2.4k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[CPU][ARM] Enable fast math in ACL deconvolution executor #26615
[CPU][ARM] Enable fast math in ACL deconvolution executor #26615
Conversation
@@ -22,6 +22,9 @@ struct DeconvAttrs { | |||
std::vector<ptrdiff_t> paddingR; | |||
ov::CoordinateDiff outputPadding; | |||
bool withBiasesParam = false; | |||
#if defined(OV_CPU_WITH_ACL) | |||
bool aclFastMath = false; |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I would prefer to avoid backend specific params in common attributes structure.
Doesn't ACL Deconv Executor have direct access to GraphConfig?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
It doesn't.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Ok, I think we can merge this as temp solution, but in general it should be refactored.
I would expect to have some backend related properties inside ExecutorContext. @EgorDuplensky what is your opinion?
…olkit#26615) ### Details: - ACL deconvolution `fast_math` option is enabled on `PERFORMANCE` mode. - This option enables fast math computation in ACL. In case this flag were set, ACL could dispatch the fastest implementation available which may introduce a drop of accuracy as well. - Accuracy testing on dataset subset highlights some deviations from reference values. Results are attached to the ticket. ### Tickets: - CVS-152534
Details:
fast_math
option is enabled onPERFORMANCE
mode.Tickets: