o1vm/mips: use batch_inversion for the witness generation #2813
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
dannywillems
changed the title
dannywillems
force-pushed
the
o1vm/batch-inversion
branch
from
00087d6
to
86aa2a5
Compare
dannywillems
changed the base branch from
master
to
o1vm/mips/interpreter-remove-inverse-or-zero
dannywillems
force-pushed
the
o1vm/batch-inversion
branch
2 times, most recently
from
f384222
to
e6169b1
Compare
dannywillems
force-pushed
the
o1vm/batch-inversion
branch
2 times, most recently
from
799172e
to
b45339e
Compare
dannywillems
force-pushed
the
o1vm/batch-inversion
branch
from
b45339e
to
30c9354
Compare
dannywillems
force-pushed
the
o1vm/batch-inversion
branch
2 times, most recently
from
4122619
to
8be102a
Compare
dannywillems
commented
Nov 21, 2024
| @@ -121,3 +125,31 @@ fn test_small_circuit() { | |||
| ); | |||
| assert!(verif, "Verification fails"); | |||
| } | |||
|
|
|||
| #[test] | |||
There was a problem hiding this comment.
Feel free to disagree with these changes. It is a single commit we can simply delete. I only want to keep a regression test in case arkworks changes.
dannywillems
commented
Nov 21, 2024
| @@ -50,10 +50,17 @@ pub const NUM_INSTRUCTION_LOOKUP_TERMS: usize = 5; | |||
| pub const NUM_LOOKUP_TERMS: usize = | |||
| NUM_GLOBAL_LOOKUP_TERMS + NUM_DECODING_LOOKUP_TERMS + NUM_INSTRUCTION_LOOKUP_TERMS; | |||
| // TODO: Delete and use a vector instead | |||
| // FIXME: since the introduction of the scratch size inverse, the value below | |||
There was a problem hiding this comment.
This is fixed in #2815.
dannywillems
commented
Nov 21, 2024
| scratch_state: [Fp::from(0); SCRATCH_SIZE], | ||
| scratch_state_inverse: [Fp::from(0); SCRATCH_SIZE_INVERSE], |
There was a problem hiding this comment.
In a future PR, as suggested by @mrmr1993, we can try to use BigInt/BigUInt directly to avoid computation in Montgomery representation. I do not know if it would make it better.
dannywillems
commented
Nov 21, 2024
| // MIPS + hash_counter + byte_counter + eof + num_bytes_read + chunk + bytes | ||
| // + length + has_n_bytes + chunk_bytes + preimage | ||
| pub const SCRATCH_SIZE: usize = 98; | ||
|
|
||
| /// Number of columns used by the MIPS interpreter to keep values to be |
There was a problem hiding this comment.
To verify it was the minimal value, I did try to use 11, and it failed.
From there, I run for some millions instructions the op-program, and it never failed.
Base automatically changed from
o1vm/mips/interpreter-remove-inverse-or-zero
to
master
November 21, 2024 18:19
dannywillems
force-pushed
the
o1vm/batch-inversion
branch
from
8be102a
to
4d82c40
Compare
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## master #2813 +/- ##
==========================================
+ Coverage 72.10% 72.15% +0.04%
==========================================
Files 256 256
Lines 59946 60042 +96
==========================================
+ Hits 43227 43326 +99
+ Misses 16719 16716 -3 ☔ View full report in Codecov by Sentry. 🚨 Try these New Features:
|
These fields will be used to keep track of the elements that should be inverted while generating the proof.
Checking as a regression tests that the batch inversion works with zeroes
dannywillems
force-pushed
the
o1vm/batch-inversion
branch
from
4d82c40
to
a11a350
Compare
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.
These fields will be used to keep track of the elements that should be inverted while generating the proof.
Speed from ~250kHz (commit 82464fb) to ~420kHz (head of this branch, i.e. o1vm/batch-inversion).
In a follow-up PR, we should take a look at mrmr1993 idea here to get the same expected performances. Also, we should decrease the number of columns in scratch_state as stated in the FIXME.
On commit 82464fb, getting:
On the head of this branch: