source/tensor/core/math/Binary.cu · 5bfbd041aee0670494e283f4d46486aac4d385f3 · NiuTrans / NiuTrans.Tensor · GitLab

Blame History Permalink

Support run NiuTensor on the older GPUs with Maxwell or more previous architectures. · 8178ba40
```
1. Offer macro to set whether use the half precision in cuda codes.
2. Update the manuals.
```
liyinqiao committed Apr 29, 2020
8178ba40

Binary.cu 11.3 KB