GM200 will not speed up LL

Karl M Johnson · 2015-02-12, 12:02

Source
If the rumors are true, GTX TItan "Black" of Kepler architecture will remain the fastest single GPU card for LL tests, among GeForce cards

LaurV · 2015-02-12, 13:31

Quote:

Originally Posted by Karl M Johnson

Source
If the rumors are true, GTX TItan "Black" of Kepler architecture will remain the fastest single GPU card for LL tests, among GeForce cards

That is nothing new and we commented about it long ago, otherwise how would they sell the Teslas at 5 times the price of a Titan?

Karl M Johnson · 2015-10-10, 05:43

For those who missed the news, Pascal will help GIMPS with LL tests.
Personally, I envision that some high-end GeForce variant will have uncapped FP64 performance, just like Titan X/Y/Z or Black did.
The HBM memory is nice, but it's not known how major the impact of this specific component will be.
Also, according to NV's current plans, the GeForce cards will not have over 16GB of vRAM, only the Teslas/Quadros will.
Additionally, the fancy NVLink interface will not be available on GeForce cards.

Q1 2016

2015-02-12, 12:02	#1
Karl M Johnson Mar 2010 3·137 Posts	GM200 will not speed up LL Source If the rumors are true, GTX TItan "Black" of Kepler architecture will remain the fastest single GPU card for LL tests, among GeForce cards

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Different Speed in different OS's	Dubslow	Software	11	2011-08-02 00:04
TF speed	Unregistered	Information & Answers	10	2011-07-27 12:34
Changes to the speed of light.	Flatlander	Homework Help	67	2011-01-22 13:37
CPU Speed Incorrect	AZMango	Software	8	2010-03-20 21:55
40 Times the speed of PC's	lpmurray	Hardware	2	2007-02-17 19:53

2015-10-10, 05:43	#3
Karl M Johnson Mar 2010 3×137 Posts	For those who missed the news, Pascal will help GIMPS with LL tests. Personally, I envision that some high-end GeForce variant will have uncapped FP64 performance, just like Titan X/Y/Z or Black did. The HBM memory is nice, but it's not known how major the impact of this specific component will be. Also, according to NV's current plans, the GeForce cards will not have over 16GB of vRAM, only the Teslas/Quadros will. Additionally, the fancy NVLink interface will not be available on GeForce cards. Q1 2016