Have you found already that TC learning speed is directly depends from CPU and RAM frequency?


I’ve been creating workstation/server for AI Deep Learning with TensorRT.
The faster the GPU Core the bigger gains from CPU frequency.
Finally I’ve delivered setup with i9-7960X - as it was only for 4 cards.
The maximal gain was almost 30% for GTX 1080 TI and 24% with Titan V.

Does anyone of you when designing workstation/server for Tensor RT take this under consideration or nver thought about it?