Is the GTX 295 still the best card for CUDA? (Better than a GTX 470?)

Message boards : Number crunching : Is the GTX 295 still the best card for CUDA? (Better than a GTX 470?)
Message board moderation

To post messages, you must log in.

Previous · 1 · 2

AuthorMessage
Profile MadMaC
Volunteer tester
Avatar

Send message
Joined: 4 Apr 01
Posts: 201
Credit: 47,158,217
RAC: 0
United Kingdom
Message 1050400 - Posted: 22 Nov 2010, 22:57:59 UTC - in response to Message 1050107.  

Yes, the 295 is still the card to beat. Although on some projects, ATI is a killer.

Not here, though.

Two 295's in one rig are a powerful couple of crunchers.

I have 2 465's in another rig, and they are rather lame.

Very dissapointed in them, I be.



Mark, you can run upto 3 wu's per card on fermi - that will bump your RAC up :-)
ID: 1050400 · Report as offensive
kittymanProject Donor
Volunteer tester
Avatar

Send message
Joined: 9 Jul 00
Posts: 47959
Credit: 859,522,794
RAC: 155,890
United States
Message 1050406 - Posted: 22 Nov 2010, 23:24:07 UTC - in response to Message 1050400.  

Yes, the 295 is still the card to beat. Although on some projects, ATI is a killer.

Not here, though.

Two 295's in one rig are a powerful couple of crunchers.

I have 2 465's in another rig, and they are rather lame.

Very dissapointed in them, I be.



Mark, you can run upto 3 wu's per card on fermi - that will bump your RAC up :-)

I dunno....
I thought I tried that and they just took 3 times as long to run, so no real gain.
Maybe I'll give it another go when work starts flowing here again.
Always remember.....kitties are all Angels with fur.
'Cat lives matter.'

Have made friends in this life.
Most were cats.
ID: 1050406 · Report as offensive
Profile SciManStevProject Donor
Volunteer tester
Avatar

Send message
Joined: 20 Jun 99
Posts: 6132
Credit: 108,774,873
RAC: 33,313
United States
Message 1050408 - Posted: 22 Nov 2010, 23:33:54 UTC - in response to Message 1050406.  
Last modified: 22 Nov 2010, 23:35:08 UTC

Hello Mark,
You can run up to 3 at a time, and there is a definite time increase for two, but you get more out in a given time period, as that is what I did to get a RAC of 47,000, with pendings in the 700,000 range for my one rig. I am set for three now, but since I switched, SETI has not been able to provide enough constant worf to see if there is a gain. The times are not doubled, but are a little longer. That alone means that you will get more out in a given time. At the moment I am set up with hyperthreading on, and 3 wu per card. That would give 18 wu's at a time, and I'm iunterested in seeing how it performs. I believe that with the new servers, I'll be able to get enough work to see where I'm at, and tune the system. It's just too hard to get a good read on where a system will top out when there is not enough work to keep it supplied.

Steve
Warning, addicted to SETI crunching!
Crunching as a member of GPU Users Group.
GPUUG Website
ID: 1050408 · Report as offensive
Profile zoom314
Volunteer tester
Avatar

Send message
Joined: 30 Nov 03
Posts: 58757
Credit: 43,049,286
RAC: 5,527
United States
Message 1050421 - Posted: 23 Nov 2010, 0:20:37 UTC - in response to Message 1050408.  

Hello Mark,
You can run up to 3 at a time, and there is a definite time increase for two, but you get more out in a given time period, as that is what I did to get a RAC of 47,000, with pendings in the 700,000 range for my one rig. I am set for three now, but since I switched, SETI has not been able to provide enough constant worf to see if there is a gain. The times are not doubled, but are a little longer. That alone means that you will get more out in a given time. At the moment I am set up with hyperthreading on, and 3 wu per card. That would give 18 wu's at a time, and I'm iunterested in seeing how it performs. I believe that with the new servers, I'll be able to get enough work to see where I'm at, and tune the system. It's just too hard to get a good read on where a system will top out when there is not enough work to keep it supplied.

Steve

Either that, Or You'll be the first person to have figured out how to Tune a Fish. ;)

Batman: Some days you just can't get rid of a bomb.
ID: 1050421 · Report as offensive
Profile jason_gee
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 24 Nov 06
Posts: 7471
Credit: 90,103,776
RAC: 14,711
Australia
Message 1050498 - Posted: 23 Nov 2010, 6:01:49 UTC - in response to Message 1050406.  
Last modified: 23 Nov 2010, 6:03:19 UTC

Yes, the 295 is still the card to beat. Although on some projects, ATI is a killer.

Not here, though.

Two 295's in one rig are a powerful couple of crunchers.

I have 2 465's in another rig, and they are rather lame.

Very dissapointed in them, I be.



Mark, you can run upto 3 wu's per card on fermi - that will bump your RAC up :-)

I dunno....
I thought I tried that and they just took 3 times as long to run, so no real gain.
Maybe I'll give it another go when work starts flowing here again.


Ahhh, I wondered when the reasons would come out ;)

For reference on my 480, with x32f:
1 Mid AR task at a time: ~8 mins total, 8 mins each
2 Mid AR tasks at a time: ~14 mins total, 7 mins each
3 Mid AR tasks at a time: ~18 mins total, 6 mins each
4 Mid AR ... too long

So I've been running 3 at a time, and the dual core with the single 480 reached a RAC high of 26000. I'm not sure what the 465 dynamics would be, but at least 2 tasks at a time should hide some of the CPU usage that the GPUs would have been otherwise idle. The same operation wasn't workable on the GTX 260 in the other room, and dynamics of the best number of tasks to run on Fermi at a time may vary with future application optimisation levels, as well as model of card.

Jason
"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.
ID: 1050498 · Report as offensive
Previous · 1 · 2

Message boards : Number crunching : Is the GTX 295 still the best card for CUDA? (Better than a GTX 470?)


 
©2017 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.