Don't have Telegram yet? Try it now!
https://hgpu.org/?p=22269
Accelerating Deep Learning Inference with Cross-Layer Data Reuse on GPUs