I've been playing around with the next generation trait solver in the hope of seeing some compile time improvements. What I found was unfortunately quite the opposite: webgpu-tools/wesl-rs#267
TLDR:
| Solver |
naga-ext |
Version |
Total |
Frontend |
Codegen |
| Current solver |
enabled |
Before |
6.7s |
4.6s |
2.1s |
| Next solver |
enabled |
Before |
164.4s |
161.5s |
2.9s |
Even after applying a small patch to wesl (see the linked PR) to reduce the type complexity that the compiler has to deal with, it's still looking pretty bleak.
| Solver |
naga-ext |
Version |
Total |
Frontend |
Codegen |
| Current solver |
enabled |
After |
5.0s |
2.6s |
2.3s |
| Next solver |
enabled |
After |
43.0s |
40.5s |
2.5s |
cc @lcnr, you seem to be actively working on this new solver.
I've been playing around with the next generation trait solver in the hope of seeing some compile time improvements. What I found was unfortunately quite the opposite: webgpu-tools/wesl-rs#267
TLDR:
naga-extEven after applying a small patch to
wesl(see the linked PR) to reduce the type complexity that the compiler has to deal with, it's still looking pretty bleak.naga-extcc @lcnr, you seem to be actively working on this new solver.