Stav/disable relocate trace with flag #2133

Stavbe · 2025-07-10T15:14:42Z

Up until now, whenever we wanted to keep the trace during execution, we also performed relocation at the end.
In the new adapter flow, we want to enable the option to keep the trace without allocating it.
I’ve set the default value to true to stay consistent with the current behavior.

This change is

github-actions · 2025-07-10T15:49:20Z

**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):     25.705 s ±  0.005 s    [User: 24.943 s, System: 0.760 s]
  Range (min … max):   25.702 s … 25.709 s    2 runs
 
Benchmark 2: hyper_threading_pr threads: 1
  Time (mean ± σ):     25.660 s ±  0.009 s    [User: 24.763 s, System: 0.894 s]
  Range (min … max):   25.654 s … 25.666 s    2 runs
 
Summary
  hyper_threading_pr threads: 1 ran
    1.00 ± 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):     14.372 s ±  0.067 s    [User: 25.112 s, System: 0.839 s]
  Range (min … max):   14.324 s … 14.419 s    2 runs
 
Benchmark 2: hyper_threading_pr threads: 2
  Time (mean ± σ):     13.928 s ±  0.017 s    [User: 24.845 s, System: 0.895 s]
  Range (min … max):   13.916 s … 13.940 s    2 runs
 
Summary
  hyper_threading_pr threads: 2 ran
    1.03 ± 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):     10.874 s ±  0.015 s    [User: 37.611 s, System: 0.975 s]
  Range (min … max):   10.864 s … 10.885 s    2 runs
 
Benchmark 2: hyper_threading_pr threads: 4
  Time (mean ± σ):     10.621 s ±  0.244 s    [User: 36.958 s, System: 1.069 s]
  Range (min … max):   10.448 s … 10.794 s    2 runs
 
Summary
  hyper_threading_pr threads: 4 ran
    1.02 ± 0.02 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):     10.279 s ±  0.123 s    [User: 38.126 s, System: 0.995 s]
  Range (min … max):   10.192 s … 10.366 s    2 runs
 
Benchmark 2: hyper_threading_pr threads: 6
  Time (mean ± σ):     10.294 s ±  0.361 s    [User: 37.283 s, System: 1.113 s]
  Range (min … max):   10.039 s … 10.549 s    2 runs
 
Summary
  hyper_threading_main threads: 6 ran
    1.00 ± 0.04 times faster than hyper_threading_pr 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):     10.321 s ±  0.047 s    [User: 38.316 s, System: 1.043 s]
  Range (min … max):   10.288 s … 10.354 s    2 runs
 
Benchmark 2: hyper_threading_pr threads: 8
  Time (mean ± σ):     10.057 s ±  0.050 s    [User: 37.659 s, System: 1.082 s]
  Range (min … max):   10.022 s … 10.092 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):     10.394 s ±  0.209 s    [User: 38.877 s, System: 1.124 s]
  Range (min … max):   10.246 s … 10.541 s    2 runs
 
Benchmark 2: hyper_threading_pr threads: 16
  Time (mean ± σ):     10.104 s ±  0.172 s    [User: 37.816 s, System: 1.198 s]
  Range (min … max):    9.982 s … 10.226 s    2 runs
 
Summary
  hyper_threading_pr threads: 16 ran
    1.03 ± 0.03 times faster than hyper_threading_main threads: 16

github-actions · 2025-07-10T16:07:04Z

Benchmark Results for unmodified programs 🚀

Command	Mean [s]	Min [s]	Max [s]	Relative
`base big_factorial`	2.144 ± 0.024	2.127	2.205	1.00 ± 0.02
`head big_factorial`	2.139 ± 0.025	2.125	2.210	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
`base big_fibonacci`	2.072 ± 0.018	2.044	2.101	1.00 ± 0.01
`head big_fibonacci`	2.070 ± 0.006	2.061	2.079	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
`base blake2s_integration_benchmark`	7.618 ± 0.376	7.401	8.638	1.01 ± 0.05
`head blake2s_integration_benchmark`	7.577 ± 0.056	7.512	7.667	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
`base compare_arrays_200000`	2.210 ± 0.032	2.184	2.275	1.00 ± 0.02
`head compare_arrays_200000`	2.207 ± 0.013	2.183	2.231	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
`base dict_integration_benchmark`	1.438 ± 0.013	1.429	1.469	1.00 ± 0.01
`head dict_integration_benchmark`	1.437 ± 0.005	1.428	1.448	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
`base field_arithmetic_get_square_benchmark`	1.229 ± 0.006	1.221	1.240	1.00
`head field_arithmetic_get_square_benchmark`	1.235 ± 0.004	1.225	1.239	1.00 ± 0.01

Command	Mean [s]	Min [s]	Max [s]	Relative
`base integration_builtins`	7.606 ± 0.067	7.540	7.715	1.00
`head integration_builtins`	7.673 ± 0.059	7.611	7.772	1.01 ± 0.01

Command	Mean [s]	Min [s]	Max [s]	Relative
`base keccak_integration_benchmark`	7.776 ± 0.221	7.611	8.270	1.00
`head keccak_integration_benchmark`	7.868 ± 0.116	7.761	8.134	1.01 ± 0.03

Command	Mean [s]	Min [s]	Max [s]	Relative
`base linear_search`	2.187 ± 0.008	2.175	2.197	1.00
`head linear_search`	2.194 ± 0.009	2.186	2.215	1.00 ± 0.01

Command	Mean [s]	Min [s]	Max [s]	Relative
`base math_cmp_and_pow_integration_benchmark`	1.513 ± 0.002	1.510	1.516	1.00
`head math_cmp_and_pow_integration_benchmark`	1.513 ± 0.005	1.508	1.522	1.00 ± 0.00

Command	Mean [s]	Min [s]	Max [s]	Relative
`base math_integration_benchmark`	1.472 ± 0.007	1.463	1.486	1.01 ± 0.01
`head math_integration_benchmark`	1.464 ± 0.006	1.457	1.477	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
`base memory_integration_benchmark`	1.216 ± 0.003	1.211	1.220	1.00
`head memory_integration_benchmark`	1.227 ± 0.009	1.218	1.251	1.01 ± 0.01

Command	Mean [s]	Min [s]	Max [s]	Relative
`base operations_with_data_structures_benchmarks`	1.546 ± 0.007	1.536	1.560	1.00
`head operations_with_data_structures_benchmarks`	1.548 ± 0.007	1.537	1.559	1.00 ± 0.01

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`base pedersen`	535.4 ± 1.4	533.6	537.5	1.00 ± 0.00
`head pedersen`	535.1 ± 1.2	533.9	537.3	1.00

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`base poseidon_integration_benchmark`	629.2 ± 3.1	623.3	633.0	1.00
`head poseidon_integration_benchmark`	631.0 ± 13.5	623.4	668.9	1.00 ± 0.02

Command	Mean [s]	Min [s]	Max [s]	Relative
`base secp_integration_benchmark`	1.842 ± 0.007	1.837	1.861	1.00 ± 0.00
`head secp_integration_benchmark`	1.840 ± 0.004	1.831	1.845	1.00

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`base set_integration_benchmark`	630.4 ± 6.8	625.8	649.4	1.00 ± 0.01
`head set_integration_benchmark`	629.5 ± 2.1	627.5	634.5	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
`base uint256_integration_benchmark`	4.285 ± 0.065	4.210	4.403	1.00
`head uint256_integration_benchmark`	4.291 ± 0.042	4.262	4.404	1.00 ± 0.02

codecov · 2025-07-10T16:09:57Z

Codecov Report

❌ Patch coverage is 82.60870% with 4 lines in your changes missing coverage. Please review.
✅ Project coverage is 96.63%. Comparing base (4226352) to head (9d2256f).
⚠️ Report is 5 commits behind head on starkware-development.

Files with missing lines	Patch %	Lines
vm/src/cairo_run.rs	77.77%	4 Missing ⚠️

Additional details and impacted files

@@                    Coverage Diff                    @@
##           starkware-development    #2133      +/-   ##
=========================================================
- Coverage                  96.63%   96.63%   -0.01%     
=========================================================
  Files                        104      104              
  Lines                      43903    43915      +12     
=========================================================
+ Hits                       42427    42436       +9     
- Misses                      1476     1479       +3

☔ 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.

anatgstarkware

Reviewable status: 0 of 4 files reviewed, 1 unresolved discussion (waiting on @Stavbe)

vm/src/cairo_run.rs line 31 at r1 (raw file):

    pub trace_enabled: bool,
    pub relocate_mem: bool,
    pub relocate_trace: bool,

Please document

Code quote:

    pub relocate_mem: bool,
    pub relocate_trace: bool,

Stavbe

Reviewable status: 0 of 4 files reviewed, 1 unresolved discussion (waiting on @anatgstarkware)

vm/src/cairo_run.rs line 31 at r1 (raw file):

Previously, anatgstarkware (anatg) wrote…

Please document

Done.

anatgstarkware

Reviewable status: 0 of 4 files reviewed, 4 unresolved discussions (waiting on @Stavbe)

vm/src/cairo_run.rs line 33 at r2 (raw file):

    pub relocate_mem: bool,
    // Disable trace relocation.
    // By default, if trace is enabled, it is also relocated.

It's not clear if you refer to the default below or to the main default args. I would remove this line

Code quote:

By default, if trace is enabled, it is also relocated

vm/src/cairo_run.rs line 59 at r2 (raw file):

            trace_enabled: false,
            relocate_mem: false,
            relocate_trace: true,

This should probably be false, as the rest of the fields here

Code quote:

relocate_trace: true

vm/src/cairo_run.rs line 499 at r2 (raw file):

        let end = cairo_runner.initialize(false).unwrap();
        assert!(cairo_runner.run_until_pc(end, &mut hint_processor).is_ok());
        assert!(cairo_runner.relocate(false, true).is_ok());

Since this test runs with no trace, it makes more sense to put here false

Code quote:

true

cairo-vm-cli/src/main.rs line 195 at r2 (raw file):

        // we need air public input or trace file.
        relocate_trace: trace_enabled
            && (args.trace_file.is_some() || args.air_public_input.is_some()),

I think this main is only used for stone, so maybe these should be true here, WDYT?

Code quote:

        relocate_mem: args.memory_file.is_some() || args.air_public_input.is_some(),
        // If trace not enabled, it is not relevant. Otherwise, we need to relocate the trace only if
        // we need air public input or trace file.
        relocate_trace: trace_enabled
            && (args.trace_file.is_some() || args.air_public_input.is_some()),

Stavbe

Reviewable status: 0 of 4 files reviewed, 4 unresolved discussions (waiting on @anatgstarkware)

cairo-vm-cli/src/main.rs line 195 at r2 (raw file):

Previously, anatgstarkware (anatg) wrote…

I think this main is only used for stone, so maybe these should be true here, WDYT?

I think it’s also used to run the VM outside of any prover context, which is why prove_mode is an argument, along with the public and private inputs.
I agree that it definitely won’t be used by Stwo, so I think that the best approach is to set 'reloacte_trace' to match trace_enabled' and by that to turn off Stwo "feature" . As for the original relocate_mem` flag—I think it’s best not to touch it.

vm/src/cairo_run.rs line 33 at r2 (raw file):

Previously, anatgstarkware (anatg) wrote…

It's not clear if you refer to the default below or to the main default args. I would remove this line

Done.

vm/src/cairo_run.rs line 59 at r2 (raw file):

Previously, anatgstarkware (anatg) wrote…

This should probably be false, as the rest of the fields here

Done.

vm/src/cairo_run.rs line 499 at r2 (raw file):

Previously, anatgstarkware (anatg) wrote…

Since this test runs with no trace, it makes more sense to put here false

Done.

anatgstarkware

Reviewed 1 of 2 files at r3, 2 of 6 files at r4.
Reviewable status: 3 of 9 files reviewed, 1 unresolved discussion (waiting on @Stavbe)

cairo-vm-cli/src/main.rs line 195 at r2 (raw file):

Previously, Stavbe wrote…

I think it’s also used to run the VM outside of any prover context, which is why prove_mode is an argument, along with the public and private inputs.
I agree that it definitely won’t be used by Stwo, so I think that the best approach is to set 'reloacte_trace' to match trace_enabled' and by that to turn off Stwo "feature" . As for the original relocate_mem` flag—I think it’s best not to touch it.

Didn't the vm relocate the trace also when prove_mode was false?
What do you mean by "original relocate_mem flag"? Was it in stone?

vm/src/cairo_run.rs line 59 at r2 (raw file):

Previously, Stavbe wrote…

Done.

I see you had to change the default in many tests, which means I was wrong. Are these tests only for stone? If so, maybe the default should be the same as in main (i.e., true, as you did before, or trace_enabled), WDYT?
Also, if relocated_mem was added for stwo, then maybe its default should also change.

Stavbe

Reviewable status: 0 of 9 files reviewed, 1 unresolved discussion (waiting on @anatgstarkware)

cairo-vm-cli/src/main.rs line 195 at r2 (raw file):

Previously, anatgstarkware (anatg) wrote…

Didn't the vm relocate the trace also when prove_mode was false?
What do you mean by "original relocate_mem flag"? Was it in stone?

The trace was relocated when either args.trace_file.is_some() or args.air_public_input.is_some()
air_public_input requires proof_mode = true
trace_file does not require it.
So in theory the trace could be relocated even when proof_mode = false but I don't see why anyone would want that.
Yes

vm/src/cairo_run.rs line 59 at r2 (raw file):

Previously, anatgstarkware (anatg) wrote…

I see you had to change the default in many tests, which means I was wrong. Are these tests only for stone? If so, maybe the default should be the same as in main (i.e., true, as you did before, or trace_enabled), WDYT?
Also, if relocated_mem was added for stwo, then maybe its default should also change.

These tests are old, and they run the VM with the Stone configuration—so yes.
I think the best approach is to align it with trace_enabled, but you need to fix a value in the default configuration so it is not really an option. I changed it back to true so it won't affect the current VM expected behavior.

Relocate mem was not added for stwo

anatgstarkware

Reviewed 1 of 4 files at r1, 1 of 2 files at r3, 2 of 7 files at r5, all commit messages.
Reviewable status: 4 of 9 files reviewed, all discussions resolved (waiting on @Stavbe)

cairo-vm-cli/src/main.rs line 195 at r2 (raw file):

Previously, Stavbe wrote…

The trace was relocated when either args.trace_file.is_some() or args.air_public_input.is_some()
air_public_input requires proof_mode = true
trace_file does not require it.
So in theory the trace could be relocated even when proof_mode = false but I don't see why anyone would want that.

Yes

thanks

anatgstarkware · 2025-07-31T11:44:59Z

vm/src/cairo_run.rs line 59 at r2 (raw file):

Previously, Stavbe wrote…

These tests are old, and they run the VM with the Stone configuration—so yes.
I think the best approach is to align it with trace_enabled, but you need to fix a value in the default configuration so it is not really an option. I changed it back to true so it won't affect the current VM expected behavior.

Relocate mem was not added for stwo

Can you please document why it was set to true?

Stavbe changed the base branch from main to starkware-development July 10, 2025 15:33

Stavbe force-pushed the stav/disable_relocate_trace_with_flag branch from 2d7cfa6 to 29fc7dc Compare July 10, 2025 15:40

Stavbe force-pushed the stav/disable_relocate_trace_with_flag branch from 29fc7dc to e0eec12 Compare July 10, 2025 15:49

Stavbe requested a review from anatgstarkware July 10, 2025 15:51

Stavbe self-assigned this Jul 10, 2025

anatgstarkware requested changes Jul 17, 2025

View reviewed changes

Stavbe force-pushed the stav/disable_relocate_trace_with_flag branch from e0eec12 to b4a5c88 Compare July 27, 2025 09:28

Stavbe commented Jul 27, 2025

View reviewed changes

anatgstarkware requested changes Jul 29, 2025

View reviewed changes

Stavbe force-pushed the stav/disable_relocate_trace_with_flag branch from b4a5c88 to d7f56e9 Compare July 29, 2025 09:26

Stavbe commented Jul 29, 2025

View reviewed changes

Stavbe force-pushed the stav/disable_relocate_trace_with_flag branch 4 times, most recently from 55a5d43 to 4c4ec02 Compare July 29, 2025 13:30

anatgstarkware requested changes Jul 31, 2025

View reviewed changes

Stavbe force-pushed the stav/disable_relocate_trace_with_flag branch from 4c4ec02 to bfb6552 Compare July 31, 2025 11:29

Stavbe commented Jul 31, 2025

View reviewed changes

anatgstarkware approved these changes Jul 31, 2025

View reviewed changes

add relocate trace flag to the runner

9d2256f

Stavbe force-pushed the stav/disable_relocate_trace_with_flag branch from bfb6552 to 9d2256f Compare July 31, 2025 12:31

Stavbe merged commit 1a224f8 into starkware-development Jul 31, 2025
21 of 23 checks passed

Stavbe deleted the stav/disable_relocate_trace_with_flag branch July 31, 2025 12:32

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Stav/disable relocate trace with flag #2133

Stav/disable relocate trace with flag #2133

Uh oh!

Stavbe commented Jul 10, 2025 •

edited

Loading

Uh oh!

github-actions bot commented Jul 10, 2025 •

edited

Loading

Uh oh!

github-actions bot commented Jul 10, 2025 •

edited

Loading

Uh oh!

codecov bot commented Jul 10, 2025 •

edited

Loading

Uh oh!

anatgstarkware left a comment

Uh oh!

Stavbe left a comment

Uh oh!

anatgstarkware left a comment

Uh oh!

Stavbe left a comment

Uh oh!

anatgstarkware left a comment

Uh oh!

Stavbe left a comment

Uh oh!

anatgstarkware left a comment

Uh oh!

anatgstarkware commented Jul 31, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Stav/disable relocate trace with flag #2133

Stav/disable relocate trace with flag #2133

Uh oh!

Conversation

Stavbe commented Jul 10, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions bot commented Jul 10, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions bot commented Jul 10, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

codecov bot commented Jul 10, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

anatgstarkware left a comment

Choose a reason for hiding this comment

Uh oh!

Stavbe left a comment

Choose a reason for hiding this comment

Uh oh!

anatgstarkware left a comment

Choose a reason for hiding this comment

Uh oh!

Stavbe left a comment

Choose a reason for hiding this comment

Uh oh!

anatgstarkware left a comment

Choose a reason for hiding this comment

Uh oh!

Stavbe left a comment

Choose a reason for hiding this comment

Uh oh!

anatgstarkware left a comment

Choose a reason for hiding this comment

Uh oh!

anatgstarkware commented Jul 31, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Stavbe commented Jul 10, 2025 •

edited

Loading

github-actions bot commented Jul 10, 2025 •

edited

Loading

github-actions bot commented Jul 10, 2025 •

edited

Loading

codecov bot commented Jul 10, 2025 •

edited

Loading