Skip to content

Conversation

vikas753
Copy link

Shuffle instruction on Kepler Config uses register first and predicate second , for which default implementation of -2,-3 fails and shuffle instruction returns wrong values
Example as below :
shfl.sync.down.b32 %r14|%p1, %r5, %r12, %r11, %r13;

Shuffle instruction on Kepler Config uses register first and predicate second , for which default implementation of -2,-3 fails and shuffle instruction returns wrong values
Example as below : 
shfl.sync.down.b32 %r14|%p1, %r5, %r12, %r11, %r13;
Shuffle instruction uses r0|p1 as register sequence of operands for processing for which current implementation gives incorrect values. Proposed changes in recognizer to account for same

shfl.sync.down.b32 %r14|%p1, %r5, %r12, %r11, %r13;
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant