CUDA patch #449
CUDA patch #449
Conversation
Codecov Report
@@ Coverage Diff @@
## master #449 +/- ##
===========================================
- Coverage 66.61% 30.91% -35.70%
===========================================
Files 67 31 -36
Lines 4211 1695 -2516
===========================================
- Hits 2805 524 -2281
+ Misses 1406 1171 -235 📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more |
|
@Roger-luo What's going on in this PR and can I help out? |
|
I realize I need to write a sampler on CUDA, but haven't had the chance to write it, it should be simple. Otherwise, we can dispatch single measurement to CPU. I can let you work on it if you want to try some CUDA programming in Julia. |
|
|
||
| end | ||
|
|
||
| function CuYao.cu(reg::SubspaceArrayReg{D}) where {D} |
There was a problem hiding this comment.
I think SubspaceArrayReg already overloaded adapt thus adapt(CuArray, reg) should just work?
lib/BloqadeCUDA/test/subspace.jl
Outdated
| using BloqadeCUDA | ||
|
|
||
|
|
||
| struct Sampler{A,B} |
There was a problem hiding this comment.
maybe we can start with a simple version without these abstractions
There was a problem hiding this comment.
Sure, I can't seem to figure out how to make the kernel compile without the abstractions.
It is complaining about searchsortedfirst being called inside the map kernel.
No description provided.