Stream: general

Topic: GSoC


Akshat Garg (Jan 18 2019 at 16:46, on Zulip):

I'm a master's student working on Compilers. Is there any plan to float any compiler related project this GSoC? (Sorry, if this was not the right place for this question)

Igor Matuszewski (Jan 18 2019 at 17:05, on Zulip):

I think https://github.com/CraneStation/cranelift may be a good project

Igor Matuszewski (Jan 18 2019 at 17:05, on Zulip):

@Dan Gohman ?

Dan Gohman (Jan 18 2019 at 17:25, on Zulip):

I'm not signed for GSOC yet, but if someone's interested in working on Cranelift or Wasmtime, let's talk :-).

Akshat Garg (Jan 18 2019 at 17:58, on Zulip):

@Dan Gohman Wasmtime looks good to me, do you have some planned enhancement for it? (that can act as a GSoC project)

Dan Gohman (Jan 18 2019 at 18:33, on Zulip):

There's a lot to do, but I don't have specific GSOC project ideas yet.

Dan Gohman (Jan 18 2019 at 18:33, on Zulip):

What areas are you interested in working on?

Akshat Garg (Jan 18 2019 at 18:44, on Zulip):

These days i'm working on optimizations, but won't mind working on code generation either.

Dan Gohman (Jan 22 2019 at 18:01, on Zulip):

Ok, so I'm curious what you think then: for a compiler that doesn't have a lot of optimization yet, what do you you think would be the most valuable thing to work on first? :smiley:

Akshat Garg (Jan 25 2019 at 20:18, on Zulip):

I was working to come up with some optimization that could take advantage of the things around it. In my mind I had a flow like C -> [x] -> wasm -> cranelift -> (clif) -> wasmtime -> x86 and I thought to make some test cases targeting a particular optimization and from the end code we can infer which optimization to start with, but I could not find packages or executables to test the flow. Can you suggest something?

Dan Gohman (Jan 29 2019 at 00:55, on Zulip):

The benchmarks in https://github.com/fluencelabs/fluence/tree/master/bench/vm/tests might be useful here, as they're self-contained and easily runnable under wasmtime.

lqd (Jan 29 2019 at 07:02, on Zulip):

/me whispers VSDG, for a possibly cool optimization topic for GSoC

Akshat Garg (Feb 02 2019 at 18:13, on Zulip):

I planned to use small test cases specialized to test the presence of an optimization so the benchmarks you suggested were good but were not aligned exactly to my need. I tried to generate .wat file from a simple C program in order to compile them with wasmtime and was planning to analyze the result. I tried doing this using llvm then with emcc, converting the c to wasm and then to wat. I'm stuck since the file generated were huge. Can you suggest something to look for?

Dan Gohman (Feb 05 2019 at 16:55, on Zulip):

I unfortunately don't currently have the time to oversee an open-ended GSoC project this year.

Akshat Garg (Feb 05 2019 at 21:08, on Zulip):

No problem. Thanks anyway for your time.

Last update: Nov 21 2019 at 23:50UTC