GPU Build - Revision history

Ed at 17:39, 21 September 2020

2020-09-21T17:39:54Z

2020-09-21T17:34:06Z

2020-09-20T17:20:01Z

2020-09-20T17:12:14Z

2019-05-11T16:12:00Z

Final Decision

2019-05-03T20:45:00Z

2017-11-10T22:38:55Z

Double GPU Server Build

2017-11-10T22:27:55Z

K80, NVLink

2017-11-10T22:18:33Z

2017-11-10T20:26:34Z

@@ Line 1: / Line 1: @@
 {{Project
-|Has project output=
+|Has project output=Content
 |Has sponsor=McNair Center
 |Has title=GPU Build

@@ Line 1: / Line 1: @@
 {{Project
 |Has sponsor=McNair Center
 |Has title=GPU Build

@@ Line 1: / Line 1: @@
 {{Project
 |Has title=GPU Build
 |Has owner=Oliver Chang,Kyran Adams

@@ Line 1: / Line 1: @@
-{{McNair Projects
+{{Project
 |Has title=GPU Build
 |Has owner=Oliver Chang,Kyran Adams

@@ Line 7: / Line 7: @@
 ==Final Decision==
-We decided to clone the NVIDIA DIGITS DevBox: https://developer.nvidia.com/devbox
+We decided to clone the NVIDIA [[DIGITS DevBox]]: https://developer.nvidia.com/devbox
 To start with we are trying to use our existing ASUS Z10 server board, rather than switching to the Asus X99-E WS workstation class motherboard, and rather than Four TITAN X GPUs, we've got a TITAN XP and a TITAN RTX.

@@ Line 5: / Line 5: @@
 }}
 ==Single vs. Multi GPU==
 *[https://www.nvidia.com/en-us/geforce/products/10series/geforce-gtx-1080-ti/ GTX 1080 Ti Specs]
 * Since we are using Tensorflow, it doesn't scale well to multiple GPUs for a single model

@@ Line 54: / Line 54: @@
 ==Double GPU Server Build==
 [https://pcpartpicker.com/user/kyranadams/saved/gDzFdC PC Partpicker build]
 ==Double GPU Build==

@@ Line 32: / Line 32: @@
 *NVLink can link between CPU and GPU for increase in speed, but only with the CPU IBM POWER8+.
 *NVLink can link between GPU and GPU as a replacement for SLI with other CPUs, but this is not super relevant to tensorflow, even if trying to parallelize across one model.
 ==Misc. Parts==

@@ Line 28: / Line 28: @@
 Cons of multiple GPUs:
 *Adds a lot of complexity.
 ==Misc. Parts==

@@ Line 46: / Line 46: @@
 * [https://www.tooploox.com/blog/deep-learning-with-gpu Another box]
 * [http://graphific.github.io/posts/building-a-deep-learning-dream-machine/ Expensive deep learning box]
 ==Double GPU Build==
@@ Line 80: / Line 83: @@
 ===Other things to consider===
 * Water cooling? [http://timdettmers.com/2015/03/09/deep-learning-hardware-guide/ this] has a good section on cooling
-* Is case actually big enough?
+* Case is not rack mounted
 ==Software tips==
 * Setting up Ubuntu and Docker [https://medium.com/@SocraticDatum/getting-started-with-gpu-driven-deep-learning-part-2-environment-setup-fd1947aab29]