tum-ei-eda / muriscv-nn

muRISCV-NN is a collection of efficient deep learning kernels for embedded platforms and microcontrollers.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Self-contained releases

christian-herber-nxp opened this issue · comments

I tried running the provided scripts, but this proves quite difficult due to policies, proxies etc. - Just to many scripts downloading stuff from somewhere.
Is there a chance that when you create a release, you build a self-contained package that contains the necessary toolchains and simulators etc?

Would a docker-image containing all the dependencies work?

that would be a good idea. another thing is that iirc, the build scripts provided always run everything, i.e. use all compilers, all simulators and all extensions. Maybe one could modularize or parameterize this a bit. Then if I have to download things manually, I could be a bit more selective.

What kind of config are you interested in? Do you want to use the V / P ext? Which simulator would you like to run on: OVPsim / Spike? The easiest way, for now, to run only a subset would probably be to just remove the parts that you don't need from the scripts.

Beyond that and depending on your OS, you might need to build some of the dependencies yourself (toolchains, Spike simulator). All the binaries we provide to download are only tested on 18.04 (see GitHub CI).

This is less of a really blocking thing than an annoyance. Modifying git tracked files is always a bit annoying and lead sooner or later to such a modified file being part of a commit. The better solution is to parameterize the script, so that I could pass the requirement e.g. to build with V, LLVM, and Spike. The number of permutations will just grow, so I think such a solution will scale better.