gitadmin/foxhunt
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

chore: enable cudarc f16 feature for half::bf16 support, remove nvrtc

Browse Source 
- cudarc features: removed "nvrtc" (no longer used, all kernels precompiled)
- cudarc features: added "f16" (enables DeviceRepr + ValidAsZeroBits for half::bf16)
- CudaSlice<half::bf16> now fully functional with all cudarc operations
- Foundation for full BF16 tensor core conversion (next session)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
jgrusewski
2026-03-27 21:31:41 +01:00
parent 136696d8b8
commit e6cbf1f622
2 changed files with 25 additions and 12 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
crates/ml/Cargo.toml
View File

@@ -129,7 +129,7 @@ rust_decimal.workspace = true
# cudarc — direct dep after candle removal, gated behind cuda feature
cudarc = { version = "0.19", optional = true, default-features = false, features = ["driver", "nvrtc", "cublas", "dynamic-linking", "std", "cuda-version-from-build-system"] }
cudarc = { version = "0.19", optional = true, default-features = false, features = ["driver", "cublas", "dynamic-linking", "std", "cuda-version-from-build-system", "f16"] }
rayon.workspace = true
crossbeam = { version = "0.8", features = ["std"] }