Potential improvement to set/restore_diag in GEQR2 #826
+120
−46
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR aims to reduce the impact of
set_diag
andrestore_diag
kernels to the runtime of GEQR2 indicated by profiling. This is achieved by:larfg
andset_diag
to reduce the number of global memory reads and writes:larfg
to write both the unit diagonal and non-unit diagonal values thus eliminating the call toset_diag
.set_diag
andrestore_diag
:set_diag
is explained above. Launch overhead ofrestore_diag
is reduced by launching the kernel once to restore all diagonal values at the expense of additional memory footprint.The following chart shows the speedup of
geqrf
with these changes on real single precision square matrices.Note:
larfb
instead oflarf
but it performed worse due to increased global memory access. I got similar results with my attempt to modifylarf
to assume implicit unit diagonal.