MauroToscano commented 9 months ago

Cairo1 proof mode

Description

This PR adds an initial working version of Cairo1 proof mode

Adds a proof mode header, working very similarly to the Cairo0 proof mode header to the code
Makes some small changes to SW Cairo header, so the pointers to the segment are properly aligned now that there are two extra instructions
Adds a RunnerMode to select from ExecutionMode, ProofModeCanonical, and ProofModeCairo1
Adds a test for dictionaries in cairo1, since they work a bit different in the headers and wasn't being tested
Reduces the number of iterations of the sample program to 100 so Cairo1 tests run faster

In the future information should be added so [fp - 2] = fp can be enforced

codecov[bot] commented 9 months ago

Codecov Report

Attention: 2 lines in your changes are missing coverage. Please review.

Comparison is base (8a2ef24) 96.82% compared to head (d777b30) 96.92%.

Files	Patch %	Lines
vm/src/vm/runners/cairo_runner.rs	97.43%	2 Missing :warning:

Additional details and impacted files

```diff @@ Coverage Diff @@ ## main #1517 +/- ## ========================================== + Coverage 96.82% 96.92% +0.10% ========================================== Files 96 96 Lines 39744 39818 +74 ========================================== + Hits 38482 38594 +112 + Misses 1262 1224 -38 ```

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

github-actions[bot] commented 9 months ago

Benchmark Results for unmodified programs :rocket:

Command	Mean [s]	Min [s]	Max [s]	Relative
`base big_factorial`	2.832 ± 0.032	2.792	2.897	1.00
`head big_factorial`	2.863 ± 0.022	2.841	2.911	1.01 ± 0.01

Command	Mean [s]	Min [s]	Max [s]	Relative
`base big_fibonacci`	2.433 ± 0.013	2.406	2.449	1.00
`head big_fibonacci`	2.434 ± 0.012	2.409	2.448	1.00 ± 0.01

Command	Mean [s]	Min [s]	Max [s]	Relative
`base blake2s_integration_benchmark`	7.046 ± 0.060	6.988	7.181	1.00 ± 0.02
`head blake2s_integration_benchmark`	7.036 ± 0.111	6.931	7.293	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
`base compare_arrays_200000`	2.541 ± 0.026	2.509	2.592	1.00
`head compare_arrays_200000`	2.549 ± 0.029	2.528	2.626	1.00 ± 0.02

Command	Mean [s]	Min [s]	Max [s]	Relative
`base dict_integration_benchmark`	1.735 ± 0.010	1.723	1.750	1.00
`head dict_integration_benchmark`	1.736 ± 0.011	1.725	1.759	1.00 ± 0.01

Command	Mean [s]	Min [s]	Max [s]	Relative
`base field_arithmetic_get_square_benchmark`	1.201 ± 0.017	1.182	1.243	1.00 ± 0.02
`head field_arithmetic_get_square_benchmark`	1.198 ± 0.009	1.184	1.214	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
`base integration_builtins`	7.037 ± 0.182	6.932	7.399	1.01 ± 0.03
`head integration_builtins`	6.981 ± 0.090	6.902	7.133	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
`base keccak_integration_benchmark`	7.096 ± 0.037	7.051	7.174	1.00
`head keccak_integration_benchmark`	7.193 ± 0.187	7.066	7.674	1.01 ± 0.03

Command	Mean [s]	Min [s]	Max [s]	Relative
`base linear_search`	2.558 ± 0.019	2.536	2.599	1.00
`head linear_search`	2.604 ± 0.101	2.549	2.889	1.02 ± 0.04

Command	Mean [s]	Min [s]	Max [s]	Relative
`base math_cmp_and_pow_integration_benchmark`	1.723 ± 0.009	1.710	1.740	1.01 ± 0.01
`head math_cmp_and_pow_integration_benchmark`	1.710 ± 0.017	1.692	1.751	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
`base math_integration_benchmark`	1.654 ± 0.011	1.639	1.681	1.00
`head math_integration_benchmark`	1.656 ± 0.013	1.634	1.672	1.00 ± 0.01

Command	Mean [s]	Min [s]	Max [s]	Relative
`base memory_integration_benchmark`	1.446 ± 0.005	1.436	1.453	1.00
`head memory_integration_benchmark`	1.461 ± 0.006	1.452	1.471	1.01 ± 0.01

Command	Mean [s]	Min [s]	Max [s]	Relative
`base operations_with_data_structures_benchmarks`	1.593 ± 0.020	1.570	1.632	1.00
`head operations_with_data_structures_benchmarks`	1.593 ± 0.009	1.582	1.605	1.00 ± 0.01

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`base pedersen`	622.0 ± 1.3	619.7	623.8	1.00
`head pedersen`	623.0 ± 5.0	618.1	633.6	1.00 ± 0.01

Command	Mean [s]	Min [s]	Max [s]	Relative
`base poseidon_integration_benchmark`	1.215 ± 0.006	1.206	1.228	1.00
`head poseidon_integration_benchmark`	1.218 ± 0.005	1.213	1.227	1.00 ± 0.01

Command	Mean [s]	Min [s]	Max [s]	Relative
`base secp_integration_benchmark`	2.138 ± 0.015	2.117	2.162	1.00 ± 0.01
`head secp_integration_benchmark`	2.130 ± 0.013	2.117	2.153	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
`base set_integration_benchmark`	1.212 ± 0.010	1.200	1.236	1.00 ± 0.01
`head set_integration_benchmark`	1.212 ± 0.006	1.204	1.223	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
`base uint256_integration_benchmark`	4.383 ± 0.039	4.339	4.468	1.00 ± 0.01
`head uint256_integration_benchmark`	4.370 ± 0.037	4.331	4.454	1.00

pefontana commented 9 months ago

should add a command line argument to set the run to proof_mode instead of just running proof_mode by default

Yes @fmoletta, that is the plan We talked with Mauro, first we can merge this to run always on proof mode, and then we can make a refactor so the user can decide if he wants to execute in proof mode or not

lambdaclass / cairo-vm