Skip to content
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

Question: comparison with Halide #30

Open
dkrikun opened this issue Sep 10, 2019 · 1 comment
Open

Question: comparison with Halide #30

dkrikun opened this issue Sep 10, 2019 · 1 comment

Comments

@dkrikun
Copy link

dkrikun commented Sep 10, 2019

Could you please briefly explain how enoki compares with Halide?

@wjakob
Copy link
Member

wjakob commented Sep 10, 2019

Halide was designed with image processing pipelines in mind -- it facilitates powerful structural changes (multithreading, blocking, SIMD, etc.) to achieve excellent performance for these types of methods. Communication patterns during such a computation can be fairly advanced, and Halide is built to support them. It depends on a fairly large compiler framework (LLVM).

Enoki was originally built to parallelize and differentiate code that numerically evaluates Monte Carlo integrals (i.e. lots of independent function evaluations). It is tiny in comparison, and it solves a much more specific problem. Its main purpose is to take a scalar computation and make it "wide" by performing the same computation on many elements of an array (either on the CPU or the GPU). The computation per SIMD lane/GPU thread is generally expected to be independent. The whole process can be differentiated if desired.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants