Resolved issue for incorrect SGEMM on Maxwell architecture.
This commit is contained in:
@@ -1,5 +1,8 @@
|
||||
# NVIDIA CUTLASS Changelog
|
||||
|
||||
## [1.2.1](https://github.com/NVIDIA/cutlass/releases/tag/v1.2.1) (2018-12-19)
|
||||
* Resolved issue with sm50 and sm52 architectures
|
||||
|
||||
## [1.2.0](https://github.com/NVIDIA/cutlass/releases/tag/v1.2.0) (2018-10-26)
|
||||
* Parallelized reductions across threadblocks ("Split-K")
|
||||
* Improved IGEMM performance
|
||||
|
||||
Reference in New Issue
Block a user