Add an actor compilation benchmark #2060
Conversation
Stebalien
force-pushed
the
steb/benchmark-compile
branch
from
3ffbe35
to
ae0dc06
Compare
| use fvm_ipld_encoding::CborStore; | ||
| use fvm_shared::version::NetworkVersion; | ||
|
|
||
| fn bench_compile(c: &mut Criterion) { |
There was a problem hiding this comment.
So, I think this benchmark is correct but... I'm not able to reproduce #2058.
There was a problem hiding this comment.
Can it be that the slowdown occurs under specific actor version(s)?
There was a problem hiding this comment.
I also didn't manage to reproduce it with this benchmark on a machine that showed a significant slowdown (used this benchmark on this branch and tag fvm@v4.3.3
bench actor compile time: [3.5975 s 3.6164 s 3.6374 s]
change: [-2.3529% -1.6171% -0.8734%] (p = 0.00 < 0.05)
Change within noise threshold.
There was a problem hiding this comment.
It could be different versions. IIRC, we used to compile the actors optimizing for size instead of speed so that could be causing cranelift to take longer. I'll try pulling a specific version.
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## master #2060 +/- ##
=======================================
Coverage 75.70% 75.70%
=======================================
Files 153 153
Lines 15667 15667
=======================================
Hits 11861 11861
Misses 3806 3806 |
Stebalien
force-pushed
the
steb/benchmark-compile
branch
from
ae0dc06
to
f25a3ac
Compare
Specifically, it benchmarks compiling wasm -> native via cranelift.