Ha! You're right. Honestly, I couldn't recall how many cores I had, and I figured it was faster to run `cat /proc/cpuinfo |grep processor| wc -l` than to google the specs.
It seems like using the right # cores saves ~25%. and then using the GPU for 26 layers saves another ~43%.
