Skip to content

update(idemix): hardening - #2136

Open
adecaro wants to merge 4 commits into
mainfrom
idemix-hardening
Open

update(idemix): hardening#2136
adecaro wants to merge 4 commits into
mainfrom
idemix-hardening

Conversation

@adecaro

@adecaro adecaro commented Aug 4, 2026

Copy link
Copy Markdown
Contributor

This PR introduces idemix hardening.

@adecaro adecaro added this to the Q3/26 milestone Aug 4, 2026
@adecaro adecaro self-assigned this Aug 4, 2026
@github-actions

github-actions Bot commented Aug 4, 2026

Copy link
Copy Markdown

📊 Token Validation Benchmark

Comparison of this PR against the base branch. 🟢 improvement · 🔴 regression · ➖ within ±1.0% noise.

Variant Benchmark Params Workers TPS (base → PR) Δ TPS
csp BenchmarkAPIGRPC f=1, nc=4, w=token-validation-service 4 106 → 96 🔴 -8.7%
csp BenchmarkLocalTokenValidation out-tokens=2in-tokens=2 4 135 → 123 🔴 -9.3%
ipa BenchmarkAPIGRPC f=1, nc=4, w=token-validation-service 4 78 → 72 🔴 -6.7%
ipa BenchmarkLocalTokenValidation out-tokens=2in-tokens=2 4 68 → 64 🔴 -6.5%

AkramBitar
AkramBitar approved these changes Aug 4, 2026
View reviewed changes

@AkramBitar AkramBitar left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

@adecaro

adecaro commented Aug 6, 2026

Copy link
Copy Markdown
Contributor Author

waiting for IBM/idemix#89 to get merged

@adecaro adecaro mentioned this pull request Aug 11, 2026
Closed
@Effi-S Effi-S self-assigned this Aug 12, 2026
@LFDT-Panurus LFDT-Panurus deleted a comment from github-actions Bot Aug 12, 2026
@LFDT-Panurus LFDT-Panurus deleted a comment from github-actions Bot Aug 12, 2026
@github-actions

github-actions Bot commented Aug 12, 2026

Copy link
Copy Markdown

📊 Token Validation Benchmark

Comparison of this PR against the base branch. 🟢 improvement · 🔴 regression · ➖ within ±1.0% noise.

Variant Benchmark Params Workers TPS (base → PR) Δ TPS
csp BenchmarkAPIGRPC f=1, nc=4, w=token-validation-service 4 118 → 107 🔴 -9.3%
csp BenchmarkLocalTokenValidation out-tokens=2in-tokens=2 4 104 → 94 🔴 -9.4%
ipa BenchmarkAPIGRPC f=1, nc=4, w=token-validation-service 4 78 → 73 🔴 -6.6%
ipa BenchmarkLocalTokenValidation out-tokens=2in-tokens=2 4 76 → 71 🔴 -6.8%

@Effi-S

Effi-S commented Aug 12, 2026 •
edited
Loading

Copy link
Copy Markdown
Contributor
image

See:
From: https://github.com/IBM/mathlib/blob/v0.3.1/driver/gurvy/bls12381/bls12-381.go#L259-L263

// Mul2 computes [e]g + [f]Q via a joint Strauss-Shamir scalar multiplication.
// Benchmarked against two independent Mul calls plus an Add: allocates far less
// (1 vs ~26 allocs) but is not faster in wall-clock time, because — unlike Mul —
// it does not use the GLV endomorphism speedup, so it forgoes the ~2x speedup that
// GLV gives each individual scalar multiplication.

@Effi-S

Effi-S commented Aug 12, 2026

Copy link
Copy Markdown
Contributor
image

@Effi-S

Effi-S commented Aug 12, 2026 •
edited
Loading

Copy link
Copy Markdown
Contributor

Hi @adecaro,

Please see the full benchmarking results (Both CSP/IPA Client-Server Token Validation)
There is a clear degradation introduced by the newer mathlib version .

What do you think?

How do you want to proceed?

@Effi-S Effi-S removed their assignment Aug 12, 2026
@Effi-S
Effi-S requested a review from aaadir August 12, 2026 13:43
@aaadir

aaadir commented Aug 12, 2026

Copy link
Copy Markdown
Contributor

"Token Validation Benchmark" also shows a performance degradation, so unless the new mathlib fixes some bugs - then why would we want to take it?

I also recall when working on the proof optimization that Claud claimed that mathlib (I guess that would be the current version) computed addition with 0 just like any other addition, rather than a trivial nop. This was one of the reasons why the optimization could exploit the 0-padded witness and field values. Now that the optimization is committed, I'm not sure there are many other places that would benefit from a trivial add-0, but may be worth an investigation.

@Effi-S Effi-S closed this Aug 19, 2026
@Effi-S
Effi-S deleted the idemix-hardening branch August 19, 2026 13:33
@Effi-S
Effi-S restored the idemix-hardening branch August 19, 2026 13:36
@Effi-S Effi-S reopened this Aug 19, 2026
@LFDT-Panurus LFDT-Panurus deleted a comment from github-actions Bot Aug 19, 2026
@LFDT-Panurus LFDT-Panurus deleted a comment from github-actions Bot Aug 19, 2026
@LFDT-Panurus LFDT-Panurus deleted a comment from github-actions Bot Aug 19, 2026
@github-actions

This comment was marked as outdated.

Sign in to view
@github-actions

github-actions Bot commented Aug 28, 2026

Copy link
Copy Markdown

📊 Token Validation Benchmark

Comparison of this PR against the base branch. 🟢 improvement · 🔴 regression · ➖ within ±1.0% noise.

Variant Benchmark Params Workers TPS (base → PR) Δ TPS
csp BenchmarkAPIGRPC f=1, nc=4, w=token-validation-service 4 117 → 105 🔴 -9.9%
csp BenchmarkLocalTokenValidation out-tokens=2in-tokens=2 4 103 → 94 🔴 -9.0%
ipa BenchmarkAPIGRPC f=1, nc=4, w=token-validation-service 4 70 → 66 🔴 -6.3%
ipa BenchmarkLocalTokenValidation out-tokens=2in-tokens=2 4 100 → 92 🔴 -7.4%

adecaro added 4 commits August 28, 2026 17:10
@adecaro
idemix 9f57f06495cb
33016cb 
Signed-off-by: Angelo De Caro <adc@zurich.ibm.com>
@adecaro
idemix 03b20615e817
a27a614 
Signed-off-by: Angelo De Caro <adc@zurich.ibm.com>
@adecaro
idemix 1895a96e2b48
84ca138 
Picks up the restored always-pairwise sumOfG1Products in IBM/idemix and the
removal of the small-n dispatch threshold from IBM/mathlib's gurvy
MultiScalarMul.

Both thresholds were tuned on single-threaded microbenchmarks, where gnark's
bucket-method MultiExp loses to a pairwise Mul2+Add loop for a handful of
bases. Under concurrent validation the reasoning inverts: MultiExp's
wall-clock win comes from fanning out over runtime.NumCPU() goroutines, which
buys nothing on a saturated process and costs the bucket allocations anyway.

Measured with TestParallelBenchmarkValidatorTransfer (10 workers, 20s, Apple
M1 Max, BLS12_381_BBS_GURVY, 2 reps), allocated bytes per verification drop
~11-12%: bulletproof 618KB -> 546KB, CSP 541KB -> 477KB. CSP throughput
improves ~7% (426 -> 438 ops/s) with p50 down ~1ms; bulletproof throughput is
a wash. The single-worker latency win the thresholds bought (~4%) is given up.

Removing mathlib's threshold also stops it overriding this repo's own
crossover decision: csp.smallMSM special-cases n=1 and n=2 and hands
everything from n=3 up to MultiScalarMul, but the threshold of 7 was quietly
rerouting n=3..6 back to a pairwise loop.

Signed-off-by: Angelo De Caro <adc@zurich.ibm.com>
@adecaro
tidy
66fc495 
Signed-off-by: Angelo De Caro <adc@zurich.ibm.com>
@Effi-S

Effi-S commented Aug 31, 2026

Copy link
Copy Markdown
Contributor

@adecaro , Isolated the mathlib here: #2337

@Effi-S

Effi-S commented Sep 1, 2026

Copy link
Copy Markdown
Contributor
image

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@Effi-S Effi-S Effi-S approved these changes
@AkramBitar AkramBitar AkramBitar approved these changes
@aaadir aaadir Awaiting requested review from aaadir

Assignees

@adecaro adecaro

Labels

dep update

Projects

None yet

Milestone

Q3/26

Development

Successfully merging this pull request may close these issues.

4 participants

@adecaro @Effi-S @aaadir @AkramBitar