Hugo Devillers
Hugo Devillers
We should probably have some kind of generic portable SIMD intrinsics that work regardless of the platform (RV, CUDA, AMDHSA, OpenCL, Shady...)
that link is pretty sus
Please have a look at the development branch and base any changes off it. Of note, I added a `Thorin` class that effectively holds the world, rather than the world...
This PR looks fine but keep in mind my remarks concerning computing that shared memory pool size statically using PE instead of a bespoke pass. Also why merge this into...
@madmann91 Thanks for the input! I have started a https://github.com/AnyDSL/thorin/commits/lift2cff branch where I implement this, it works on simple examples, but it's not actually very powerful on its own because...
Same issue here. Edit: On Debian sid
If the copy is asynchronous, how do you know it's finished ? Device-wide barrier ?
I have not been updating this for a while, so that's to be expected. Can you share your fork ?
No, but this is also the behavior of Rust actually... https://godbolt.org/z/MxqP15sEn I nonetheless have a patch that allows for this. I'm tempted to allow ourselves to diverge from Rust here,...
Do we document the syntax of this thing somewhere (guessing this is based on GCC's but I'm not familiar) ?