-
Notifications
You must be signed in to change notification settings - Fork 5
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Gradient with DI is 2x slower than native Mooncake #272
Labels
enhancement (performance)
Would reduce the time it takes to run some bit of the code
Comments
willtebbutt
added
the
enhancement (performance)
Would reduce the time it takes to run some bit of the code
label
Sep 28, 2024
Hmm I'm struggling to replicate. On my system, I see: julia> bench_function(foo1, foo2, par1, par2)
4.096 ns (0 allocations: 0 bytes)
julia> bench_mooncake_native(foo1, foo2, par1, par2)
67.660 ns (0 allocations: 0 bytes)
julia> bench_mooncake_di(foo1, foo2, par1, par2)
67.296 ns (0 allocations: 0 bytes) My versions are: [6e4b80f9] BenchmarkTools v1.5.0
[a0c0ee7d] DifferentiationInterface v0.6.1
[da2b9cff] Mooncake v0.4.4
[90137ffa] StaticArrays v1.9.7 Could you let me know if these are the same as yours? |
With DI v0.6.1 and Mooncake v0.4.5 this is resolved indeed. It might have been on Mooncake v0.4.3 when I first ran the test. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Following the conversation on Slack around StaticArrays, I compared benchmark results between DI and native Mooncake, and there is a 2x slowdown I'm not sure I can explain. Any idea what could be causing this @willtebbutt?
The text was updated successfully, but these errors were encountered: