ci: Add comparative GPU benchmarks on merge_group
#238
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
samuelburnham
pushed a commit
to samuelburnham/arecibo
that referenced
this pull request
Jan 5, 2024
* Digest simplifications (argumentcomputer#238) * remove unused digest computations * avoid a verifier having to recompute a digest * update crate version Restore digest computation and fix API inconsistency (argumentcomputer#242) * Revert "Digest simplifications (argumentcomputer#238)" This reverts commit 71ecb66. * upgrade neptune * make the public interface uniform wrt refs vs. copies * simplify prove_step * refactor: Adapt supernova RecursiveSNARK to Nova API changes - Updated `RecursiveSNARK` struct in `supernova/mod.rs` to include `z0_primary` and `z0_secondary` fields, simplifying method parameters. - Refactored `prove_step` method in `RecursiveSNARK` struct to leverage the new instance variables, `z0_primary` and `z0_secondary`, - Replaced all usages of `z0_primary` and `z0_secondary` in function calls with their respective instance variables. --------- Co-authored-by: Srinath Setty <srinath@microsoft.com>
samuelburnham
force-pushed
the
gpu-bench
branch
4 times, most recently
from
961fe26
to
b381177
Compare
samuelburnham
force-pushed
the
gpu-bench
branch
from
b381177
to
a621594
Compare
This was referenced Jan 8, 2024
samuelburnham
force-pushed
the
gpu-bench
branch
from
a621594
to
8af5dc3
Compare
samuelburnham
force-pushed
the
gpu-bench
branch
from
8af5dc3
to
ef6253d
Compare
samuelburnham
changed the title
merge_group (WIP)merge_group
huitseeker
approved these changes
Jan 9, 2024
This was referenced Jan 10, 2024
huitseeker
pushed a commit
to huitseeker/arecibo
that referenced
this pull request
Jan 14, 2024
* remove unused digest computations * avoid a verifier having to recompute a digest * update crate version
huitseeker
pushed a commit
to huitseeker/arecibo
that referenced
this pull request
Jan 14, 2024
…r#242) * Revert "Digest simplifications (argumentcomputer#238)" This reverts commit 71ecb66. * upgrade neptune * make the public interface uniform wrt refs vs. copies * simplify prove_step
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.
GPU benchmarks for PR regression testing
recursive-snarkbenchmark group and ID to enable formatting withcriterion-table. This will be extended tocompressed-snarkand their supernova counterparts when ready.criterion-tableoutput in a621594#commitcomment-136557829, and the basic logic works. Unfortunately, we're limited by the fact that the base benchmarks ondevdon't have the same formatting or parameters as this PR, so it's impossible to demo exactly what it will look like beyond the proposal below.bcon GPU runners gh-actions-runner#25)Proposed
criterion-tableoutput:Benchmarks
Table of Contents
Benchmark Results
RecursiveSNARK
ref=ae521be709871f94a622c85b0e5d6e8f776be144ref=ace6724dc77667506dbcd1122b9851bd6edb25d2Prove-StepCircuitSize-029.28 ms(✅ 1.00x)28.71 ms(✅ 1.02x faster)Verify-StepCircuitSize-019.48 ms(✅ 1.00x)19.10 ms(✅ 1.02x faster)Prove-StepCircuitSize-655935.57 ms(✅ 1.00x)35.27 ms(✅ 1.01x faster)Verify-StepCircuitSize-655923.84 ms(✅ 1.00x)23.89 ms(✅ 1.00x slower)Made with criterion-table
Design notes
For the other benchmarks which have more than one group, they would each appear in their own table (e.g.
CompressedSNARKandCompressedSNARK-Commitments). This layout will be shown as well when implemented.Some other display options:
StepCircuitSizeinto its own group, which is the existing behavior. This would proliferate the number of benchmark tables to a large degree if we are displaying four different sets of benchmarks in one.mdfileProveandVerifybenchmarks into their own table or group them together. This would require some finagling since the benchmarks are currently ordered sequentially in the order they are run.Future work
Helps #243