Stav/remove test compare to python vm

[Stavbe]

Initially, we planned to fill holes only in the memory used for prover_input_info, but we realized it would be better for Stone to receive a memory file without holes as well —since this they are already computed by the VM.
Therefore, we no longer intend to compare the memory against the Python VM’s memory, as they will not be identical.

---

This change is 

[github-actions]

**Hyper Thereading Benchmark results**




hyperfine -r 2 -n "hyper_threading_main threads: 1" 'RAYON_NUM_THREADS=1 ./hyper_threading_main' -n "hyper_threading_pr threads: 1" 'RAYON_NUM_THREADS=1 ./hyper_threading_pr'
Benchmark 1: hyper_threading_main threads: 1
  Time (mean ± σ):     26.330 s ±  0.006 s    [User: 25.528 s, System: 0.798 s]
  Range (min … max):   26.325 s … 26.335 s    2 runs
 
Benchmark 2: hyper_threading_pr threads: 1
  Time (mean ± σ):     25.640 s ±  0.033 s    [User: 24.768 s, System: 0.868 s]
  Range (min … max):   25.616 s … 25.663 s    2 runs
 
Summary
  hyper_threading_pr threads: 1 ran
    1.03 ± 0.00 times faster than hyper_threading_main threads: 1




hyperfine -r 2 -n "hyper_threading_main threads: 2" 'RAYON_NUM_THREADS=2 ./hyper_threading_main' -n "hyper_threading_pr threads: 2" 'RAYON_NUM_THREADS=2 ./hyper_threading_pr'
Benchmark 1: hyper_threading_main threads: 2
  Time (mean ± σ):     14.610 s ±  0.084 s    [User: 25.537 s, System: 0.835 s]
  Range (min … max):   14.551 s … 14.669 s    2 runs
 
Benchmark 2: hyper_threading_pr threads: 2
  Time (mean ± σ):     13.935 s ±  0.014 s    [User: 24.843 s, System: 0.913 s]
  Range (min … max):   13.925 s … 13.945 s    2 runs
 
Summary
  hyper_threading_pr threads: 2 ran
    1.05 ± 0.01 times faster than hyper_threading_main threads: 2




hyperfine -r 2 -n "hyper_threading_main threads: 4" 'RAYON_NUM_THREADS=4 ./hyper_threading_main' -n "hyper_threading_pr threads: 4" 'RAYON_NUM_THREADS=4 ./hyper_threading_pr'
Benchmark 1: hyper_threading_main threads: 4
  Time (mean ± σ):     10.845 s ±  0.480 s    [User: 38.708 s, System: 0.984 s]
  Range (min … max):   10.505 s … 11.184 s    2 runs
 
Benchmark 2: hyper_threading_pr threads: 4
  Time (mean ± σ):     10.431 s ±  0.528 s    [User: 36.853 s, System: 1.077 s]
  Range (min … max):   10.057 s … 10.804 s    2 runs
 
Summary
  hyper_threading_pr threads: 4 ran
    1.04 ± 0.07 times faster than hyper_threading_main threads: 4




hyperfine -r 2 -n "hyper_threading_main threads: 6" 'RAYON_NUM_THREADS=6 ./hyper_threading_main' -n "hyper_threading_pr threads: 6" 'RAYON_NUM_THREADS=6 ./hyper_threading_pr'
Benchmark 1: hyper_threading_main threads: 6
  Time (mean ± σ):     10.772 s ±  0.052 s    [User: 38.666 s, System: 0.992 s]
  Range (min … max):   10.735 s … 10.809 s    2 runs
 
Benchmark 2: hyper_threading_pr threads: 6
  Time (mean ± σ):     10.070 s ±  0.212 s    [User: 37.750 s, System: 1.062 s]
  Range (min … max):    9.920 s … 10.219 s    2 runs
 
Summary
  hyper_threading_pr threads: 6 ran
    1.07 ± 0.02 times faster than hyper_threading_main threads: 6




hyperfine -r 2 -n "hyper_threading_main threads: 8" 'RAYON_NUM_THREADS=8 ./hyper_threading_main' -n "hyper_threading_pr threads: 8" 'RAYON_NUM_THREADS=8 ./hyper_threading_pr'
Benchmark 1: hyper_threading_main threads: 8
  Time (mean ± σ):     10.467 s ±  0.079 s    [User: 39.127 s, System: 1.005 s]
  Range (min … max):   10.412 s … 10.523 s    2 runs
 
Benchmark 2: hyper_threading_pr threads: 8
  Time (mean ± σ):     10.194 s ±  0.066 s    [User: 37.636 s, System: 1.086 s]
  Range (min … max):   10.147 s … 10.241 s    2 runs
 
Summary
  hyper_threading_pr threads: 8 ran
    1.03 ± 0.01 times faster than hyper_threading_main threads: 8




hyperfine -r 2 -n "hyper_threading_main threads: 16" 'RAYON_NUM_THREADS=16 ./hyper_threading_main' -n "hyper_threading_pr threads: 16" 'RAYON_NUM_THREADS=16 ./hyper_threading_pr'
Benchmark 1: hyper_threading_main threads: 16
  Time (mean ± σ):     10.417 s ±  0.114 s    [User: 39.511 s, System: 1.097 s]
  Range (min … max):   10.337 s … 10.498 s    2 runs
 
Benchmark 2: hyper_threading_pr threads: 16
  Time (mean ± σ):     10.216 s ±  0.243 s    [User: 37.826 s, System: 1.167 s]
  Range (min … max):   10.044 s … 10.388 s    2 runs
 
Summary
  hyper_threading_pr threads: 16 ran
    1.02 ± 0.03 times faster than hyper_threading_main threads: 16

[github-actions]

Benchmark Results for unmodified programs 🚀

Command	Mean [s]	Min [s]	Max [s]	Relative
base big_factorial	2.175 ± 0.021	2.157	2.215	1.00 ± 0.01
head big_factorial	2.169 ± 0.016	2.150	2.206	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
base big_fibonacci	2.110 ± 0.025	2.091	2.175	1.00 ± 0.02
head big_fibonacci	2.105 ± 0.025	2.076	2.158	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
base blake2s_integration_benchmark	7.684 ± 0.100	7.592	7.952	1.00
head blake2s_integration_benchmark	7.944 ± 0.060	7.883	8.036	1.03 ± 0.02

Command	Mean [s]	Min [s]	Max [s]	Relative
base compare_arrays_200000	2.210 ± 0.013	2.201	2.242	1.00 ± 0.01
head compare_arrays_200000	2.201 ± 0.007	2.190	2.216	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
base dict_integration_benchmark	1.447 ± 0.011	1.434	1.470	1.01 ± 0.01
head dict_integration_benchmark	1.436 ± 0.015	1.423	1.477	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
base field_arithmetic_get_square_benchmark	1.236 ± 0.006	1.223	1.246	1.00
head field_arithmetic_get_square_benchmark	1.240 ± 0.005	1.233	1.250	1.00 ± 0.01

Command	Mean [s]	Min [s]	Max [s]	Relative
base integration_builtins	7.705 ± 0.041	7.639	7.755	1.00
head integration_builtins	7.972 ± 0.041	7.906	8.033	1.03 ± 0.01

Command	Mean [s]	Min [s]	Max [s]	Relative
base keccak_integration_benchmark	7.927 ± 0.085	7.853	8.149	1.00
head keccak_integration_benchmark	8.337 ± 0.071	8.282	8.504	1.05 ± 0.01

Command	Mean [s]	Min [s]	Max [s]	Relative
base linear_search	2.212 ± 0.014	2.199	2.244	1.00
head linear_search	2.227 ± 0.029	2.206	2.301	1.01 ± 0.01

Command	Mean [s]	Min [s]	Max [s]	Relative
base math_cmp_and_pow_integration_benchmark	1.538 ± 0.009	1.523	1.550	1.00
head math_cmp_and_pow_integration_benchmark	1.551 ± 0.014	1.538	1.587	1.01 ± 0.01

Command	Mean [s]	Min [s]	Max [s]	Relative
base math_integration_benchmark	1.473 ± 0.005	1.464	1.479	1.00
head math_integration_benchmark	1.476 ± 0.009	1.470	1.498	1.00 ± 0.01

Command	Mean [s]	Min [s]	Max [s]	Relative
base memory_integration_benchmark	1.234 ± 0.004	1.229	1.241	1.00
head memory_integration_benchmark	1.240 ± 0.016	1.228	1.274	1.00 ± 0.01

Command	Mean [s]	Min [s]	Max [s]	Relative
base operations_with_data_structures_benchmarks	1.552 ± 0.007	1.544	1.568	1.00
head operations_with_data_structures_benchmarks	1.580 ± 0.003	1.577	1.585	1.02 ± 0.00

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
base pedersen	537.5 ± 0.8	536.5	539.0	1.00
head pedersen	538.3 ± 3.2	534.7	545.9	1.00 ± 0.01

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
base poseidon_integration_benchmark	640.7 ± 7.0	634.4	658.7	1.00
head poseidon_integration_benchmark	658.5 ± 4.7	651.8	666.7	1.03 ± 0.01

Command	Mean [s]	Min [s]	Max [s]	Relative
base secp_integration_benchmark	1.853 ± 0.005	1.848	1.864	1.00
head secp_integration_benchmark	1.878 ± 0.006	1.866	1.888	1.01 ± 0.00

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
base set_integration_benchmark	635.0 ± 2.0	632.4	639.1	1.00
head set_integration_benchmark	635.7 ± 7.8	629.6	655.0	1.00 ± 0.01

Command	Mean [s]	Min [s]	Max [s]	Relative
base uint256_integration_benchmark	4.336 ± 0.036	4.301	4.429	1.00 ± 0.01
head uint256_integration_benchmark	4.325 ± 0.015	4.304	4.348	1.00

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 96.55%. Comparing base (33d75ca) to head (62792f5).

Additional details and impacted files

@@                    Coverage Diff                    @@
##           starkware-development    #2086      +/-   ##
=========================================================
- Coverage                  96.62%   96.55%   -0.08%     
=========================================================
  Files                        102      102              
  Lines                      44388    43250    -1138     
=========================================================
- Hits                       42889    41759    -1130     
+ Misses                      1499     1491       -8     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[JulianGCalderon]

Hi @Stavbe!

Is there a way to keep the memory comparison anyway? Comparing with cairo-lang is a great way for testing the Cairo VM, and ensuring that we don't break compatibility.

We came up with two solutions:

• Add support for both unfilled and filled memory (maybe with a flag --unfilled-memory). That way, we can keep the memory comparisons. The flag --memory would output the filled memory (incompatible with cairo-lang).
• Sync these changes with cairo-lang, so that cairo-lang also fills memory. That way we can update the VM behaviour, but still compare the results to cairo-lang.

[JulianGCalderon]

Another alternative could be to add a script that fills the memory holes when necessary, and leaving the VM output without holes. Would this work?

[DavidLevitGurevich]

Reviewed 6 of 6 files at r1, all commit messages.
Reviewable status: complete! all files reviewed, all discussions resolved (waiting on @Stavbe)

[Stavbe]

Hi @JulianGCalderon,
We moved the hole-filling logic to run only in proof mode, so I updated this PR to perform memory comparison checks only in the other cases.

Reviewable status: 1 of 7 files reviewed, 2 unresolved discussions (waiting on @DavidLevitGurevich and @Stavbe)

---

vm/src/tests/compare_outputs_dynamic_layouts.sh line 197 at r3 (raw file):

    echo "Running cairo-lang with case: $case"
    cairo-run --program "$full_program" \
        --layout "dynamic" --cairo_layout_params_file "$full_layout" --proof_mode \

proof mode

---

vm/src/tests/compare_factorial_outputs_all_layouts.sh line 14 at r3 (raw file):

    # Run cairo_lang
    echo "Running cairo_lang with layout $layout"
    cairo-run --layout $layout --proof_mode  --program $factorial_compiled --trace_file factorial_py.trace --memory_file factorial_py.memory --air_public_input factorial_py.air_public_input --air_private_input factorial_py.air_private_input

proof mode