-
Notifications
You must be signed in to change notification settings - Fork 168
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
make the upper limit on TMP loop unrolling configurable #2971
Merged
ericniebler
merged 4 commits into
NVIDIA:main
from
ericniebler:configurable-type-algorithm-unrolling
Dec 1, 2024
Merged
make the upper limit on TMP loop unrolling configurable #2971
ericniebler
merged 4 commits into
NVIDIA:main
from
ericniebler:configurable-type-algorithm-unrolling
Dec 1, 2024
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
🟩 CI finished in 2h 52m: Pass: 100%/396 | Total: 9d 02h | Avg: 33m 08s | Max: 1h 36m | Hits: 20%/22090
|
Project | |
---|---|
CCCL Infrastructure | |
+/- | libcu++ |
CUB | |
Thrust | |
CUDA Experimental | |
python | |
CCCL C Parallel Library | |
Catch2Helper |
Modifications in project or dependencies?
Project | |
---|---|
CCCL Infrastructure | |
+/- | libcu++ |
+/- | CUB |
+/- | Thrust |
+/- | CUDA Experimental |
+/- | python |
+/- | CCCL C Parallel Library |
+/- | Catch2Helper |
🏃 Runner counts (total jobs: 396)
# | Runner |
---|---|
327 | linux-amd64-cpu16 |
28 | linux-arm64-cpu16 |
26 | linux-amd64-gpu-v100-latest-1 |
15 | windows-amd64-cpu16 |
Trying
Everything until 10 looks good. The benchmark is probably not great, because this regards only header parsing time. |
Different unrollings for
|
miscco
approved these changes
Nov 30, 2024
looks like 10 is the sweet spot |
I suspect the sweet spot is different for different compilers. |
🟩 CI finished in 2h 03m: Pass: 100%/396 | Total: 8d 13h | Avg: 31m 10s | Max: 1h 18m | Hits: 19%/22094
|
Project | |
---|---|
CCCL Infrastructure | |
+/- | libcu++ |
CUB | |
Thrust | |
CUDA Experimental | |
python | |
CCCL C Parallel Library | |
Catch2Helper |
Modifications in project or dependencies?
Project | |
---|---|
CCCL Infrastructure | |
+/- | libcu++ |
+/- | CUB |
+/- | Thrust |
+/- | CUDA Experimental |
+/- | python |
+/- | CCCL C Parallel Library |
+/- | Catch2Helper |
🏃 Runner counts (total jobs: 396)
# | Runner |
---|---|
327 | linux-amd64-cpu16 |
28 | linux-arm64-cpu16 |
26 | linux-amd64-gpu-v100-latest-1 |
15 | windows-amd64-cpu16 |
miscco
approved these changes
Dec 1, 2024
davebayer
pushed a commit
to davebayer/cccl
that referenced
this pull request
Dec 2, 2024
* make the upper limit on TMP loop unrolling configurable * drop the TMP PP unrolling limit on nvcc and nvhpc whose parsers are slow
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Description
@bernhardmgruber has discovered that the CUDA front end is spending an inordinate amount of time processing
__type_list.h
. it's possible that this is caused by excessive use of the preprocessor to unroll template recursion. this PR makes the unroll limit configurable so we can easily test that hypothesis.Checklist