* Initial commit for WIP benchmarking infrastructure
* fixed lint issues and ran prettier
* added <rootDir>/scripts/bench/ to ignore paths for Jest
* tidied up code and fixed a few bugs in the runner.js
* fixed eslint
* improved the benchmark output from the runner
* fixed typo
* tided up print output in runner.js
* throw error if chrome canary is not installed on mac
* added better bench stats output (tables)
* added benchmark diff to table results
* adds bundle size comparisons to results
* tidied up the results
* fixed prettier output
* attempt to trigger bech for circleci build
* fixes flow exlclusion for lighthouse module
* added class components benchmark
* cleaned up stats.js
* stability changes
* circleci node version to 7
* added another benchmark
* added colours to the different benchmarks to check if being cached
* force no-cache headers
* added more info messages
* refactor chrome launching.
* fixed an issue where launcher.kill might fail
* Move server to runner. Launch it only once.
* tidy up
* changes the logic in how the remote repo is checked out
* removes bench from circleci build
* removed colors from benchmarks (no longer needed)
* added CI integration comment
* added hacker news benchmark
* added skipBuild functionality
* relabelled remote
* Add confidence intervals
* added first meaningful paint
* removed some unused code
* reverted code.json
* updated benchmark runs back to 10
* no longer breaks when results contain missing bundles
* adds CPU throttling
* renamed build to remote-repo
* small fix to build
* fixed bad merge
* upped runs to 10 from 2 again
* properly pulls master
* removes old-bench
* runs benchmarks in headless mode
* adds a --headless option
* improved the git build process
* added README
* updated based feedback from review
* adds merge base commit sha
* addressing more PR feedback
* remove built JS react files
* updated .gitignore
* added combined bundle load times to the metrics
Previously, the extract-components script would create the same number of layers of composites as the page it captures, but it would output a new class for each time any composite is used (since we don't want to replicate all the component logic).
I changed the script to output a single type for each type in the input -- and each generated component takes an index for which output it should return. This should be closer to how the original code behaves, especially with respect to VM function call lookups where the amount of polymorphism makes a difference.
I re-recorded the benchmarks with the new scripts. They run significantly faster:
```
Comparing old.txt (control) vs new.txt (test)
Significant differences marked by ***
% change from control to test, with 99% CIs:
* ssr_pe_cold_ms_jsc_jit
% change: -41.73% [-43.37%, -40.09%] ***
means: 39.3191 (control), 22.9127 (test)
* ssr_pe_cold_ms_jsc_nojit
% change: -44.24% [-46.69%, -41.80%] ***
means: 45.8646 (control), 25.5764 (test)
* ssr_pe_cold_ms_node
% change: -45.61% [-47.38%, -43.85%] ***
means: 90.1118 (control), 49.0116 (test)
```
This is probably in part due to the changes here, but also the page I captured has changed somewhat in the meantime and there seem to be slightly fewer components in the hierarchy, so they're not really comparable. But going forward we can use this benchmark which should be more accurate. I also included an identical copy that uses stateless functional components so we can test optimizations to those later.
Running this is left as an exercise for the reader, since my measure.py isn't designed for this at present. But something like this might work:
```diff
diff --git a/scripts/bench/measure.py b/scripts/bench/measure.py
index 4cedf47..627ec97 100755
--- a/scripts/bench/measure.py
+++ b/scripts/bench/measure.py
@@ -79,15 +79,12 @@ def _measure_ssr_ms(engine, react_path, bench_name, bench_path, measure_warm):
if (typeof React !== 'object') throw new Error('React not laoded');
report('factory_ms', END - START);
- globalEval(readFile(ENV.bench_path));
- if (typeof Benchmark !== 'function') {
- throw new Error('benchmark not loaded');
- }
+ globalEval("bm = (function(){" + readFile("bench-createclass-madman.js") + "})");
+ bm();
var START = now();
- var html = React.renderToString(React.createElement(Benchmark));
- html.charCodeAt(0); // flatten ropes
+ bm();
var END = now();
- report('ssr_' + ENV.bench_name + '_cold_ms', END - START);
+ report('cc_' + ENV.bench_name + '_cold_ms', END - START);
var warmup = ENV.measure_warm ? 80 : 0;
var trials = ENV.measure_warm ? 40 : 0;
@@ -119,7 +116,7 @@ def _main():
return 1
react_path = sys.argv[1]
- trials = 30
+ trials = 60
sys.stderr.write("Measuring SSR for PE benchmark (%d trials)\n" % trials)
for i in range(trials):
for engine in [
@@ -132,7 +129,7 @@ def _main():
sys.stderr.flush()
sys.stderr.write("\n")
- trials = 3
+ trials = 0#3
sys.stderr.write("Measuring SSR for PE with warm JIT (%d slow trials)\n" % trials)
for i in range(trials):
for engine in [
```
