Y
Hacker News
new
|
ask
|
show
|
jobs
by
hzwanip
128 days ago
> No LLVM. No HIP translation layer. No "convert your CUDA to something else first."
What is the problem with such approaches?