[solved] Windows 376.51 vs Ubuntu 16.04 375.66

I ran a project in both of my dual-boot environments… and one is highly outperforming the other… much to my annoyance.

I’ve updated everything via the apt-get routines in linux, and standard driver downloads in windows… nothing beta

CUDA VERSION 8 with the patch

Training runs take 8 minutes in Windows, 45 minutes in Linux …

My guess is, perhaps I need to update either the nvidia driver… or tensorflow… but they all seem to match… any thoughts?

Thank you, would hate to be stuck in windows

one thing I should add

I notice Windows takes the Intel integrated graphics (this is a Lenovo P50) , so the Nvidia Quadro M1000M is not used to draw… I have a feeling this is what leads to the discrepancy…

perhaps I should reinstall, and avoid installing the nvidia graphics? I realize this is perhaps an edge case but has anyone done similar setups where the integrated graphics run 2D desktop?

I solved it… solution is pieces of another post + some more insights on my part…

https://devtalk.nvidia.com/default/topic/991849/cuda-setup-and-installation/-solved-run-cuda-on-dedicated-nvidia-gpu-while-connecting-monitors-to-intel-hd-graphics-is-this-possible-/