deps(l2): upgrade SP1 from v5.0.8 (Turbo) to v6.0.1 (Hypercube)#6188

Closed

avilagaston9 wants to merge 53 commits into

mainfrom

bench/sp1-hypercube

Commits on Feb 9, 2026

Add proving time instrumentation and log-parsing benchmark script
avilagaston9
committed
Fix stale VK file paths in deployer after guest program was moved out of prover
avilagaston9
committed
Use pending nonce in load test to support consecutive runs without nonce conflicts,
avilagaston9
committed
Enrich benchmark script with batch metadata from Prometheus metrics endpoint.
avilagaston9
committed
Make benchmark script read the log file once and exit instead of tailing.
avilagaston9
committed
Fix off-by-one in batch_size metric: a batch spanning blocks [first, last]
avilagaston9
committed
Add --timed/--no-timed flag to prover to control proving time measurement.
avilagaston9
committed
Add env var support to load test CLI for RPC URL, tx amount, and endless mode.
avilagaston9
committed
Add prover benchmarking guide and agent workflow documentation.
avilagaston9
committed
Extract load test round logic into run_round function and use it once
avilagaston9
committed
Change load_test and wait_until_all_included to take references for client,
avilagaston9
committed
Change --timed flag default to false so it must be explicitly opted in.
avilagaston9
committed
Clarify GPU acceleration option in prover benchmarking guide by showing
avilagaston9
committed
Add PROVER_ARGS variable to Makefile prover targets so extra flags like
avilagaston9
committed
Use PROVER_CLIENT_TIMED env var in Makefile prover targets instead of
avilagaston9
committed
Use PROVER_CLIENT_TIMED env var directly instead of a TIMED Makefile alias.
avilagaston9
committed
Rewrite benchmark script to output a markdown file with a table and summary
avilagaston9
committed
Remove load-test target from crates/l2/Makefile and use the existing
avilagaston9
committed
Add automatic server specs detection (CPU, RAM, GPU) to the benchmark
avilagaston9
committed
Add batches-to-prove prompt to agent setup instructions and include
avilagaston9
committed
Clarify that LOAD_TEST_RPC_URL must point to the L2 node RPC (port 1729
avilagaston9
committed
Add prover_type to BatchRequest so the proof coordinator can filter
avilagaston9
committed
Remove redundant prover_type field from Prover struct — the backend
avilagaston9
committed

Commits on Feb 10, 2026

Add missing prover_type field to TDX quote-gen BatchRequest and fix load_test formatting.
avilagaston9
committed
Clarify in prover benchmarking workflow that the load test must be started
avilagaston9
committed
Pre-compile the load test binary before starting the L2 to avoid
avilagaston9
committed
Use pre-built binaries in benchmarking workflow to eliminate redundant recompilation.
avilagaston9
committed
Replace todo!() with unimplemented!() in OpenVM and ZisK backends, rename
avilagaston9
committed

Commits on Feb 11, 2026

Skip batch assignment when the requesting prover's proof already exists,
avilagaston9
committed
Return ProverTypeNotNeeded immediately when a prover connects with a
avilagaston9
committed
Remove the all_proofs_exist loop from handle_request since it is redundant:
avilagaston9
committed
Flatten handle_request into sequential early returns instead of a nested
avilagaston9
committed
Remove the redundant contains_batch check in handle_request since
avilagaston9
committed
Move the prover type check before the store query in handle_request so
avilagaston9
committed
Fix --sp1 flag in Makefile to pass explicit 'true' value (the arg uses
avilagaston9
committed
Add SP1 Turbo (v5.0.8) proving benchmark results as baseline for comparison
avilagaston9
committed
Upgrade SP1 from v5.0.8 (Turbo) to v6.0.0-rc.1 (Hypercube) for benchmarking.
avilagaston9
committed
Add .gitignore for guest-program build outputs (ELF binaries, target dirs).
avilagaston9
committed

Commits on Feb 12, 2026

Update CI workflows to install protoc and use SP1 v6.0.0-rc.1, fix formatting
avilagaston9
committed
Revert the overly aggressive Cargo.lock regeneration (which pulled deps
avilagaston9
committed
Replace SP1 ProverClient::from_env() with compile-time feature flags
avilagaston9
committed
Add missing ProvingKey trait import in l1_proof_sender so that
avilagaston9
committed
Spawn SP1 CudaProver initialization on a separate OS thread to avoid
avilagaston9
committed
Wrap all SP1 blocking SDK operations in std::thread::scope to avoid
avilagaston9
committed

Commits on Feb 14, 2026

Add SP1 v6 GPU benchmark report documenting the failed RTX 5090 (Blackwell) attempt.
avilagaston9
committed

Commits on Feb 25, 2026

Switch SP1 from v6.0.0-rc.1 to v6.0.0 stable across all version pins, patch
avilagaston9
committed
Merge remote-tracking branch 'origin/main' into bench/sp1-hypercube
avilagaston9
committed
Regenerate Cargo.lock after merging main to resolve dependency versions
avilagaston9
committed
Bump SP1 from v6.0.0 to v6.0.1 and Rust toolchain from 1.90.0 to 1.91.0.
avilagaston9
committed
Replace expect() and unwrap() calls in the SP1 backend with proper error
avilagaston9
committed
Update Nix fenix toolchain hash for Rust 1.91.0 in service.nix. The previous
avilagaston9
committed
Apply cargo fmt to init_prover_setup — break chained method call across lines
avilagaston9
committed
Pin vergen to =9.0.6 as a direct build-dependency in cmd/ethrex to prevent
avilagaston9
committed