Add dynamic layout basic support #1824

JulianGCalderon · 2024-08-23T20:59:40Z

Add dynamic layout support

This PR doesn't change any logic in the VM, it only changes some signatures to allow for dynamic layout parameters. The missing functionality is being developed in #1838.

It also adds a new job to the CI for testing dynamic layouts in different scenarios.

I separated it this way to make reviewing easier.

Description

This PR adds support for dynamic layout. When using dynamic layout, the actual parameters of the layout must be specified through a path to a JSON file in the command line arguments:

cairo-compile cairo_programs/abs_value_array.cairo --output cairo_programs/abs_value_array_compiled.json

target/release/cairo-vm-cli cairo_programs/abs_value_array_compiled.json \
    --layout dynamic \
    --cairo_layout_params_file cairo_layout_params_file.example.json

Or, with Cairo 1:

cargo run --bin cairo1-run cairo_programs/cairo-1-programs/array_get.cairo \
    --layout dynamic \
    --cairo_layout_params_file cairo_layout_params_file.example.json

To add this functionality, the public API had to be changed. Now both CairoRunner::new and CairoRunner::new_v2 receive as their 3rd argument an Option<CairoLayoutParams>. This must be Some only when using dynamic layout. Most changes in this PR are a side effect of this modification.

impl CairoRunner {
    /// The `cairo_layout_params` argument should only be used with dynamic layout.
    /// It is ignored otherwise.
    pub fn new_v2(
        program: &Program,
        layout: LayoutName,
        cairo_layout_params: Option<CairoLayoutParams>,
        mode: RunnerMode,
        trace_enabled: bool,
    ) -> Result<CairoRunner, RunnerError> {

Even though this is not the most "rustic" solution, it was done this way to facilitate usage and minimize changes needed.

The JSON file has the following format. I left an example in the repository's root just in case, it can be moved or removed.

{
    "rc_units": 4,
    "log_diluted_units_per_step": 4,
    "cpu_component_step": 8,
    "memory_units_per_step": 8,
    "uses_pedersen_builtin": true,
    "pedersen_ratio": 256,
    "uses_range_check_builtin": true,
    "range_check_ratio": 8,
    "uses_ecdsa_builtin": true,
    "ecdsa_ratio": 2048,
    "uses_bitwise_builtin": true,
    "bitwise_ratio": 16,
    "uses_ec_op_builtin": true,
    "ec_op_ratio": 1024,
    "uses_keccak_builtin": true,
    "keccak_ratio": 2048,
    "uses_poseidon_builtin": true,
    "poseidon_ratio": 256,
    "uses_range_check96_builtin": true,
    "range_check96_ratio": 8,
    "range_check96_ratio_den": 1,
    "uses_add_mod_builtin": true,
    "add_mod_ratio": 128,
    "add_mod_ratio_den": 1,
    "uses_mul_mod_builtin": true,
    "mul_mod_ratio": 256,
    "mul_mod_ratio_den": 1
}

Checklist

Linked to Github Issue
Unit tests added
Integration tests added.
This change requires new documentation.
- Documentation has been added/updated.
- CHANGELOG has been updated.

github-actions · 2024-08-23T21:10:03Z

**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):     30.160 s ±  0.160 s    [User: 29.397 s, System: 0.761 s]
  Range (min … max):   30.047 s … 30.273 s    2 runs
 
Benchmark 2: hyper_threading_pr threads: 1
  Time (mean ± σ):     30.293 s ±  0.100 s    [User: 29.579 s, System: 0.712 s]
  Range (min … max):   30.222 s … 30.363 s    2 runs
 
Summary
  'hyper_threading_main threads: 1' ran
    1.00 ± 0.01 times faster than 'hyper_threading_pr 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):     16.820 s ±  0.052 s    [User: 29.575 s, System: 0.805 s]
  Range (min … max):   16.783 s … 16.856 s    2 runs
 
Benchmark 2: hyper_threading_pr threads: 2
  Time (mean ± σ):     16.872 s ±  0.010 s    [User: 29.833 s, System: 0.783 s]
  Range (min … max):   16.865 s … 16.879 s    2 runs
 
Summary
  'hyper_threading_main threads: 2' ran
    1.00 ± 0.00 times faster than 'hyper_threading_pr 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):     11.791 s ±  0.533 s    [User: 41.132 s, System: 0.955 s]
  Range (min … max):   11.414 s … 12.168 s    2 runs
 
Benchmark 2: hyper_threading_pr threads: 4
  Time (mean ± σ):     11.719 s ±  0.285 s    [User: 41.396 s, System: 0.949 s]
  Range (min … max):   11.517 s … 11.921 s    2 runs
 
Summary
  'hyper_threading_pr threads: 4' ran
    1.01 ± 0.05 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):     11.468 s ±  0.163 s    [User: 41.245 s, System: 0.976 s]
  Range (min … max):   11.352 s … 11.583 s    2 runs
 
Benchmark 2: hyper_threading_pr threads: 6
  Time (mean ± σ):     11.805 s ±  0.150 s    [User: 42.139 s, System: 0.937 s]
  Range (min … max):   11.699 s … 11.911 s    2 runs
 
Summary
  'hyper_threading_main threads: 6' ran
    1.03 ± 0.02 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):     11.244 s ±  0.195 s    [User: 41.696 s, System: 0.944 s]
  Range (min … max):   11.106 s … 11.382 s    2 runs
 
Benchmark 2: hyper_threading_pr threads: 8
  Time (mean ± σ):     11.638 s ±  0.187 s    [User: 42.033 s, System: 0.957 s]
  Range (min … max):   11.506 s … 11.770 s    2 runs
 
Summary
  'hyper_threading_main threads: 8' ran
    1.04 ± 0.02 times faster than 'hyper_threading_pr 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):     11.208 s ±  0.043 s    [User: 41.925 s, System: 1.064 s]
  Range (min … max):   11.177 s … 11.239 s    2 runs
 
Benchmark 2: hyper_threading_pr threads: 16
  Time (mean ± σ):     11.195 s ±  0.092 s    [User: 42.843 s, System: 1.056 s]
  Range (min … max):   11.130 s … 11.260 s    2 runs
 
Summary
  'hyper_threading_pr threads: 16' ran
    1.00 ± 0.01 times faster than 'hyper_threading_main threads: 16'

github-actions · 2024-08-23T21:44:30Z

Benchmark Results for unmodified programs 🚀

Command	Mean [s]	Min [s]	Max [s]	Relative
`base big_factorial`	2.421 ± 0.034	2.393	2.481	1.01 ± 0.02
`head big_factorial`	2.407 ± 0.013	2.392	2.436	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
`base big_fibonacci`	2.386 ± 0.079	2.322	2.533	1.02 ± 0.04
`head big_fibonacci`	2.338 ± 0.025	2.319	2.398	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
`base blake2s_integration_benchmark`	8.757 ± 0.308	8.452	9.399	1.00
`head blake2s_integration_benchmark`	8.867 ± 0.550	8.507	10.327	1.01 ± 0.07

Command	Mean [s]	Min [s]	Max [s]	Relative
`base compare_arrays_200000`	2.515 ± 0.013	2.499	2.537	1.00
`head compare_arrays_200000`	2.580 ± 0.091	2.501	2.762	1.03 ± 0.04

Command	Mean [s]	Min [s]	Max [s]	Relative
`base dict_integration_benchmark`	1.673 ± 0.024	1.658	1.737	1.01 ± 0.02
`head dict_integration_benchmark`	1.650 ± 0.012	1.640	1.680	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
`base field_arithmetic_get_square_benchmark`	1.385 ± 0.018	1.360	1.410	1.00
`head field_arithmetic_get_square_benchmark`	1.405 ± 0.032	1.366	1.458	1.01 ± 0.03

Command	Mean [s]	Min [s]	Max [s]	Relative
`base integration_builtins`	8.928 ± 0.320	8.618	9.780	1.01 ± 0.05
`head integration_builtins`	8.800 ± 0.234	8.565	9.303	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
`base keccak_integration_benchmark`	9.096 ± 0.131	8.959	9.243	1.00
`head keccak_integration_benchmark`	9.167 ± 0.354	8.890	10.067	1.01 ± 0.04

Command	Mean [s]	Min [s]	Max [s]	Relative
`base linear_search`	2.499 ± 0.019	2.475	2.529	1.00 ± 0.01
`head linear_search`	2.498 ± 0.022	2.469	2.547	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
`base math_cmp_and_pow_integration_benchmark`	1.710 ± 0.006	1.703	1.724	1.00
`head math_cmp_and_pow_integration_benchmark`	1.711 ± 0.006	1.699	1.724	1.00 ± 0.01

Command	Mean [s]	Min [s]	Max [s]	Relative
`base math_integration_benchmark`	1.678 ± 0.009	1.665	1.698	1.00 ± 0.01
`head math_integration_benchmark`	1.678 ± 0.022	1.659	1.727	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
`base memory_integration_benchmark`	1.401 ± 0.011	1.391	1.422	1.00
`head memory_integration_benchmark`	1.419 ± 0.011	1.407	1.437	1.01 ± 0.01

Command	Mean [s]	Min [s]	Max [s]	Relative
`base operations_with_data_structures_benchmarks`	1.835 ± 0.057	1.783	1.961	1.01 ± 0.04
`head operations_with_data_structures_benchmarks`	1.808 ± 0.048	1.767	1.929	1.00

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`base pedersen`	578.0 ± 7.6	572.7	599.2	1.00 ± 0.01
`head pedersen`	576.9 ± 3.9	573.5	584.1	1.00

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`base poseidon_integration_benchmark`	703.5 ± 9.2	695.2	725.1	1.00 ± 0.01
`head poseidon_integration_benchmark`	703.3 ± 3.8	698.4	710.6	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
`base secp_integration_benchmark`	2.037 ± 0.039	2.001	2.112	1.02 ± 0.02
`head secp_integration_benchmark`	2.003 ± 0.013	1.985	2.023	1.00

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`base set_integration_benchmark`	694.8 ± 13.0	683.3	716.4	1.00 ± 0.03
`head set_integration_benchmark`	691.6 ± 11.9	682.7	723.9	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
`base uint256_integration_benchmark`	4.841 ± 0.097	4.745	5.040	1.00
`head uint256_integration_benchmark`	4.919 ± 0.149	4.754	5.180	1.02 ± 0.04

codecov · 2024-08-23T22:17:59Z

Codecov Report

Attention: Patch coverage is 88.62876% with 34 lines in your changes missing coverage. Please review.

Project coverage is 94.79%. Comparing base (63b4a55) to head (37d83fd).
Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
vm/src/types/layout.rs	83.41%	33 Missing ⚠️
vm/src/cairo_run.rs	75.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1824      +/-   ##
==========================================
- Coverage   94.83%   94.79%   -0.05%     
==========================================
  Files         101      101              
  Lines       39579    39836     +257     
==========================================
+ Hits        37536    37762     +226     
- Misses       2043     2074      +31

Flag	Coverage Δ
	`94.79% <88.62%> (-0.05%)`	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

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

…ayouts.sh

pefontana

Nice one @JulianGCalderon !
I really like the documentation
I just made some small comments an we will be good to go

pefontana · 2024-09-25T18:35:28Z

vm/src/tests/compare_outputs_dynamic_layouts.sh

+
+# Build cases to execute
+CASES=(
+    "cairo_programs/proof_programs/factorial.json;all_cairo"


Can you add 3 or 4 more programs here?

pefontana · 2024-09-25T18:42:10Z

vm/src/vm/runners/cairo_runner.rs

    pub fn new_v2(
        program: &Program,
        layout: LayoutName,
+        cairo_layout_params: Option<CairoLayoutParams>,


Can you change the name here to dynamic_layout_params, I think it is clearer

pefontana · 2024-09-25T18:46:25Z

vm/src/cairo_run.rs

@@ -26,6 +29,7 @@ pub struct CairoRunConfig<'a> {
    pub trace_enabled: bool,
    pub relocate_mem: bool,
    pub layout: LayoutName,
+    pub cairo_layout_params: Option<CairoLayoutParams>,


same here, I think dynamic_layout_params makes more clear to the user that is only needed in that cases
Also add a little comment here explaining like you did in the CairoRunner

vm/src/types/instance_definitions/builtins_instance_def.rs

cairo_layout_params_file.json

vm/src/types/instance_definitions/builtins_instance_def.rs

vm/src/vm/runners/cairo_runner.rs

vm/src/types/layout.rs

vm/src/vm/runners/cairo_runner.rs

fmoletta

Could we update the Other CLI arguments section of the README?

JulianGCalderon · 2024-09-27T19:45:52Z

@pefontana @fmoletta I made the requested changes.

pefontana

Nice one @JulianGCalderon !

Alon-Ti and others added 6 commits August 15, 2024 12:08

Fix Zero segment location.

7d14e74

Fixed has_zero_segment naming

e71b052

Fix prover input.

0ec972a

Fixed version reading when no version is supplied

8c5c924

Added change to changelog.

057bc6f

fix test_from_serializable()

64c844b

JulianGCalderon force-pushed the dynamic-layout branch 2 times, most recently from 2ee787b to 315594d Compare August 23, 2024 21:24

JulianGCalderon force-pushed the dynamic-layout branch from 4913d86 to d03618c Compare August 23, 2024 21:56

OmriEshhar1 and others added 6 commits August 25, 2024 11:48

fix panic_impl error

51d25a7

fix cairo version

88d1da1

Add dummy changelog

c5d3bd5

Pin wasm-bindgen

ea377a8

Register change in CHANGELOG

e8e7bb3

Update Cargo.lock

87a736a

JulianGCalderon changed the base branch from main to ensure-no_std-debug August 26, 2024 17:44

JulianGCalderon force-pushed the dynamic-layout branch from d03618c to 864cdc5 Compare August 26, 2024 17:44

JulianGCalderon added 8 commits August 26, 2024 14:48

Remove changes from CHANGELOG

2fe7f90

Add argument parsing for layout params file

6c7237e

Add dynamic support (no implement)

e4b162b

Add cairo_layout_params_file.example.json

43360b9

Implement dynamic layout creation

e3e31b8

Update CHANGELOG

ad97bd0

Add cli dynamic support for cairo 1

11edca3

Make wasm compatible

adbba88

JulianGCalderon force-pushed the dynamic-layout branch from 864cdc5 to adbba88 Compare August 26, 2024 17:49

Base automatically changed from ensure-no_std-debug to main August 26, 2024 20:22

JulianGCalderon added 4 commits September 19, 2024 13:48

Update rust.yml

7a44d4e

Rename compare_outputs_dynamic_layout.sh to compare_outputs_dynamic_l…

859b7fb

…ayouts.sh

Update test script

47e441b

Merge branch 'main' into dynamic-layout

9bbfc36

pefontana requested changes Sep 25, 2024

View reviewed changes