chore: add profiling command to benchmark docs

openvm-org · Jan 7, 2025 · 82a289f · 82a289f
1 parent 4d18009
commit 82a289f
Show file tree

Hide file tree

Showing 3 changed files with 25 additions and 273 deletions.
diff --git a/ci/scripts/metric_unify/main.py b/ci/scripts/metric_unify/main.py
diff --git a/docs/crates/benchmarks.md b/docs/crates/benchmarks.md
@@ -94,6 +94,24 @@ The `OUTPUT_PATH` environmental variable shouuld be set to the file path where y
 
 To run a benchmark with the leaf aggregation, add `--features aggregation` to the above command.
 
+### Markdown Output
+
+To generate a markdown summary of the collected metrics, first install `openvm-prof`:
+
+```bash
+cd <repo_root>/crates/prof
+cargo install --force --path .
+```
+
+Then run the command:
+
+```bash
+openvm-prof --json-paths $OUTPUT_PATH
+```
+
+This will generate a markdown file to the same path as $OUTPUT_PATH but with a `.md` extension. The `--json-paths` argument can take multiple files, comma separated.
+There is also an optional `--prev-json-paths` argument to compare the metrics with a previous run.
+
 ### Circuit Flamegraphs
 
 While traditional flamegraphs generated from instrumenting a proving binary run on the host machine are useful,

diff --git a/docs/repo/layout.md b/docs/repo/layout.md
@@ -6,6 +6,7 @@ The main components of the repository are:
   - [Documentation](#documentation)
   - [Benchmarks](#benchmarks)
   - [CI](#ci)
+  - [Profiling](#profiling)
   - [CLI](#cli)
   - [SDK](#sdk)
   - [Toolchain](#toolchain)
@@ -33,13 +34,17 @@ Benchmark guest programs and benchmark scripts are in [`openvm-benchmarks`](../.
 
 Scripts for CI use and metrics post-processing are in [`ci`](../../ci).
 
+### Profiling
+
+- [`openvm-prof`](../../crates/prof): Tools to post-process metrics emitted by the VM for performance profiling.
+
 ### CLI
 
 Command-line binary to compile, execute, and prove guest programs is in [`cli`](../../crates/cli).
 
 ### SDK
 
-- [`sdk`](../../crates/sdk): The developer SDK for the VM. It includes the OpenVM aggregation programs to support continuations for all VMs in the framework, and well as a local aggregation scheduling implementation. It provides the final interface for proving an arbitrary program for a target VM. The SDK includes functionality to generate the final onchain SNARK verifier contract.
+- [`openvm-sdk`](../../crates/sdk): The developer SDK for the VM. It includes the OpenVM aggregation programs to support continuations for all VMs in the framework, and well as a local aggregation scheduling implementation. It provides the final interface for proving an arbitrary program for a target VM. The SDK includes functionality to generate the final onchain SNARK verifier contract.
 
 ### Toolchain
 
@@ -122,4 +127,4 @@ The toolchain, ISA, and VM are simultaenously extendable. All non-system functio
 - [`openvm-pairing-circuit`](../../extensions/pairing/circuit): Circuit extension for optimal Ate pairing on BN254 and BLS12-381 curves.
 - [`openvm-pairing-transpiler`](../../extensions/pairing/transpiler): Transpiler extension for optimal Ate pairing on BN254 and BLS12-381.
 - [`openvm-pairing-guest`](../../extensions/pairing/guest): Guest library with optimal Ate pairing on BN254 and BLS12-381 and associated constants. Also includes elliptic curve operations for VM runtime with the `halo2curves` feature gate.
-- [`openvm-pairing-tests`](../../extensions/pairing/tests): Integration tests for the pairing extension.
+- [`openvm-pairing-tests`](../../extensions/pairing/tests): Integration tests for the pairing extension.