mersenneforum.org  

Go Back   mersenneforum.org > Great Internet Mersenne Prime Search > Hardware > GPU Computing

Reply
 
Thread Tools
Old 2016-04-05, 20:04   #1
Mark Rose
 
Mark Rose's Avatar
 
"/X\(‘-‘)/X\"
Jan 2013

2·31·47 Posts
Default Tesla P100 — 5.4 DP TeraFLOPS — Pascal

http://www.nvidia.com/object/tesla-p100.html

It's a beast!
  • 5.3 TeraFLOPS double-precision performance
  • 10.6 TeraFLOPS single-precision performance
  • 21.2 TeraFLOPS half-precision performance
  • 160 GB/s interconnect bandwidth with NVIDIA NVLink
  • 720 GB/s memory bandwidth with CoWoS HBM2 Stacked Memory
  • 16 GB of CoWoS HBM2 Stacked Memory

Last fiddled with by Mark Rose on 2016-04-05 at 20:06
Mark Rose is offline   Reply With Quote
Old 2016-04-05, 20:23   #2
firejuggler
 
firejuggler's Avatar
 
Apr 2010
Over the rainbow

9AB16 Posts
Default

ok, what is the speed of a titan, to compare?
firejuggler is online now   Reply With Quote
Old 2016-04-05, 20:32   #3
mackerel
 
mackerel's Avatar
 
Feb 2016
UK

39310 Posts
Default

From wikipedia:
Titan X (Maxwell): SP 6.1, DP 0.19
Titan Black (Kepler): SP 5.1, DP 1.7

Last fiddled with by mackerel on 2016-04-05 at 20:33
mackerel is offline   Reply With Quote
Old 2016-04-05, 21:30   #4
firejuggler
 
firejuggler's Avatar
 
Apr 2010
Over the rainbow

247510 Posts
Default

Digging thru the web, I found that Nvidia now offer a rack with 8 Tesla 100, named DGX-1 for 129k$
http://www.engadget.com/2016/04/05/n...supercomputer/

Quote:
NVIDIA has moved its GPU business far beyond just powering gaming computers. For several years now, the company has been focused on using its technology to power advanced machine learning technology, and today the company announced the positively wild $129,000 DGX-1 supercomputer. NVIDIA CEO Jen-Hsun Huang called the DGX-1 the equivalent of 250 servers in a box -- it has 7TB of SSD storage, eight of the just-announced Tesla P100 GPUs and two Xeon processors.

Last fiddled with by firejuggler on 2016-04-05 at 21:33
firejuggler is online now   Reply With Quote
Old 2016-04-06, 00:11   #5
tServo
 
tServo's Avatar
 
"Marv"
May 2009
near the Tannhäuser Gate

10001000102 Posts
Default

Quote:
Originally Posted by Mark Rose View Post
It's a beast!]
Indeed it is, Mark.
Also worth noting:
This is a Tesla board.
It has 3584 cuda cores.
Its speed is 1328 Mhz, boost is 1480 Mhz.
It only (!) draws 300 watts.
The memory bus width is 4096 bits !!!!! ( Previous highest was 384 bits ).
This thing from another universe has over 15 Billion transistors ( more than double Kepler ).

Zounds!
tServo is offline   Reply With Quote
Old 2016-04-06, 05:19   #6
ldesnogu
 
ldesnogu's Avatar
 
Jan 2008
France

10318 Posts
Default

This beast has the potential to crush Intel upcoming Knight Landing!

AnandTech has a chart comparing Pascal to Kepler and Maxwell.

Last fiddled with by ldesnogu on 2016-04-06 at 05:19
ldesnogu is offline   Reply With Quote
Old 2016-04-06, 17:15   #7
ixfd64
Bemusing Prompter
 
ixfd64's Avatar
 
"Danny"
Dec 2002
California

231810 Posts
Default

The consumer cards will be much more cost-effective unless you're planning to do LL testing or P-1 factoring.

Last fiddled with by ixfd64 on 2016-04-06 at 17:15
ixfd64 is online now   Reply With Quote
Old 2016-04-06, 18:11   #8
ATH
Einyen
 
ATH's Avatar
 
Dec 2003
Denmark

11×271 Posts
Default

nevermind, I see it is not a normal card but an "accelerator"

Last fiddled with by ATH on 2016-04-06 at 18:12
ATH is offline   Reply With Quote
Old 2016-04-06, 20:04   #9
Madpoo
Serpentine Vermin Jar
 
Madpoo's Avatar
 
Jul 2014

29×113 Posts
Default

Quote:
Originally Posted by firejuggler View Post
Digging thru the web, I found that Nvidia now offer a rack with 8 Tesla 100, named DGX-1 for 129k$
http://www.engadget.com/2016/04/05/n...supercomputer/
That seems... high. Averaging $16,250 USD per Tesla 100 card? Are they really expected to sell for that much individually? Probably not, meaning the advantage is having it in one dual CPU server with all the gee-whiz cooling and what-nots built in, but that seems like a hefty premium.

Whatever... you know that no matter the price, some government agency is going to buy a bunch and send taxpayers the bill.
Madpoo is offline   Reply With Quote
Old 2016-04-07, 15:54   #10
TheJudger
 
TheJudger's Avatar
 
"Oliver"
Mar 2005
Germany

111010 Posts
Default

Hi,

https://devblogs.nvidia.com/parallel...ures-revealed/
Quote:
CUDA 8 will be available in August 2016 and there will be a release candidate available around June.
IIRC correctly CUDA 7.5 can't create code for CC 6.x.


-----
Quote:
Originally Posted by tServo View Post
The memory bus width is 4096 bits !!!!! ( Previous highest was 384 bits ).
Doesn't really matter but that is not true... GT200(b) has 8x64b.

Oliver
TheJudger is offline   Reply With Quote
Old 2016-04-07, 15:56   #11
Mark Rose
 
Mark Rose's Avatar
 
"/X\(‘-‘)/X\"
Jan 2013

2·31·47 Posts
Default

This talk goes over the new architecture in detail:

http://on-demand.gputechconf.com/gtc...deo/S6176.html
Mark Rose is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Passive Pascal Xyzzy GPU Computing 1 2017-05-17 20:22
Just some fun playing with a Tesla P100 plus a question... JonRussell Hardware 9 2017-04-27 11:46
Nvidia Pascal, a third of DP firejuggler GPU Computing 12 2016-02-23 06:55
14 TeraFlops last May 2004. Now? wouter Software 8 2010-08-21 00:01
GIMPS Broke 10 Teraflops! jinydu Lounge 27 2004-01-18 05:34

All times are UTC. The time now is 01:37.

Sun Nov 29 01:37:24 UTC 2020 up 79 days, 22:48, 3 users, load averages: 0.88, 1.12, 1.20

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.