Whitshad

I think I'm good, 24/7 OC there ?

Not exactly. Some parts i recover from the e-waste before they being sent to be destroyed, some parts i'm buying on e-bay or craigslist. Living in silicon valley is expensive, but if your hobby is building powerful computers - then this place rocks.

Ok, your curiosity is contagious - i have to drive this evening to the office to test it, but here it is: the single-core score is 157 (little bit lower than i7 4770k - it has 165.

The thing is i'm running a Docker swarm with the different versions of the trained model for cross-comparison and self-validation, so i'm using GPUs (i currently have two of them installed) for training the new models and CPUs to run like a 50 to a 100 Docker images with the scoring models.

It is in the office right now so i can't tell before i get there on Monday but as far as i remember it is not great because i run it in the all core overclock mode at 3.47 which means that single core turbo boost is also topped at 3.4 instead of going at 3.78 - so one core is somewhat in the ballpark of the i7 4700 single core score. I will post the exact score on Monday.

This thread is going strong i see. Hmm, hmm. Ok - here's my submission then. Slight overclock on the custom water cooling - but i mostly headed for looks on my TensorFlow personal WS than overcloackability. Not 9000 though. Not even close. Should get moar hardware!

Sure thing! 

Machine learning/Deep learning workstation. I just wanted my main workstation not to be boring plus i wanted to try myself at the custom water cooling - so it is my first try at that too.
 
My previous workstation (now retired to be my main home PC) looks like this: