Interesting read.

- 54,151 s: 101,850 credits - bonus missed ?

Bonus wasn't missed; the task was returned & reported inside 24h. In this case the credit is less than might be expected due to the WU type; you can see the credit doesn't line up with other WU's of similar runtime. Usually WU's grant the same credit per time (give or task a very small percentage), however some were not assessed accurately enough and the credit awarded is skewed, one way or another. This time down compared to the others. So you really have to compare like for like Work Units to get enough accuracy to compare card performances.
The tasks that were granted 60900 credits are the give away for GPU comparison. Of note is that the GTX650Ti matches a GTX560 for performance. This is a good generation on generation marker.
With more powerful cards other variables (and there are many that influence performance) can significantly influence GPU comparison. Yesterday I was looking at two same type GPU's on different OS's. I wanted to work out the relative performance influence of the operating systems, but I encountered a discrepancy - basically it was due to the system RAM; DDR2x800 vs DDR3 2133. It's been a while since I've encountered such an unbalanced system. You just can't configure your way out of slow memory.
At this stage I'm fairly convinced that a GTX660Ti matches a GTX580, so we have a few generation on generation markers. I think it's still up for debate whether or not the GTX660Ti fully matches a GTX670. The speculation is that it does, or near-enough, but without hard figures we don't know for sure.

---

FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

Many thanks for the explanations !
I will continue with long runs.

Yeah thx perfect analyis from the 650TI, i was waiting for something like that :)

---

DSKAG Austria: http://www.dskag.at

Hi, GDF:

I want to process GPU tasks, but need an NVIDIA video card. Which should I choose to suit my budget of about $150?

I have the following PCs processing various tasks:

#1 AMD-A10 HD 7660D, HD 6670, HD 6570, MOBO ASUS F2 A85V PRO;
#2 AMD-A10 HD 7660D, HD6790,MOBO ASUS F2 A85V PRO
#3 AMD 1090T, HD 6570 MOBO ASUS M5A99XEVO.

All Win7 64 bit.

Thank you,

John

----------------------------------------

John,

I think it is a fairly easy answer, at least if you are in the U.S. The GTX 650 Ti does very well, and you have several choices at that price.

I have a GIGABYTE GV-N65TOC-1GI, since I needed a short card that had good cooling and low noise. But if you can take a longer card, the Asus is also very good; there are normal speed and overclocked versions. I usually avoid the overclocked ones, since you can get errors, but the Kepler chips run cool and the Gigabyte (factory overclocked to 1032 MHz) is doing very well. It processes a long Noelia work unit in about 18 hours, essentially the same as my GTX 560 which uses more power and is longer.

But the GTX 650 Ti does reserve a whole core on my E8400 Core2 Duo, whereas the GTX 560 does not. I don't know how that would work with your CPUs.

http://www.newegg.com/Product/ProductList.aspx?Submit=ENE&DEPA=0&Order=BESTMATCH&N=-1&isNodeId=1&Description=gtx+650&x=0&y=0


Quick bunch of GTX 650 and GTX 650 Ti's on Newegg ranging from about $100 to $180 or so.

Many thanks for your help, Jim1348. Prices in US & canada appear similar these days.

Many thanks, Dylan. Both you and Jim1348 have given me enough information to start the hunt!

Be sure to get a GTX 650 Ti, as it's massively faster than a GTX 650 for crunching. Oh that sweet naming madness..

MrS

---

Scanning for our furry friends since Jan 2002

The GTX 650 Ti is twice as fast as the GTX 650, and costs about 35% more. It's well worth the extra cost.

I moved these well off topic posts from the New app is out for testing News thread, to this NVidia GTX 650 Ti Graphics cards (GPUs) thread.

---

FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

Hi,

I aborted the task 063px10x2-NOELIA_063p-0-2-RND4695 as it was showing about 45% complete after about 11.7h. I have now set my preferences for short runs, only. Should my GTX 650 Ti be able to process long run tasks without issues? I have successfully completed three, but subsequent long tasks appeared to stall.

Thanks,

John

063px10x2-NOELIA_063p-0-2-RND4695
application Long runs (8-12 hours on fastest card)
created 1 Mar 2013 | 10:12:59 UTC
minimum quorum 1
initial replication 1
max # of error/total/success tasks 7, 10, 6
Task
click for details Computer Sent Time reported
or deadline
explain Status Run time
(sec) CPU time
(sec) Credit Application
6557534 146594 1 Mar 2013 | 10:49:59 UTC 2 Mar 2013 | 11:42:49 UTC Aborted by user 42,105.23 18,527.88 --- Long runs (8-12 hours on fastest card) v6.18 (cuda42)

John;

I am running an EVGA GTX 650 Ti 2GB with no issues at all on long runs exclusively. And I usually get the bonus for coming in under 24hrs.

I'm running a light OC at 1124 graphics clock and 2722 on mem, and 1.10v.

This card was already OC from factory at 1071 to begin with.

Been stable now for weeks.

Operator

---

John, your issue is caused by the new features employed in the NOELIA WU. It's not related to your card at all.

MrS

---

Scanning for our furry friends since Jan 2002

Many thanks to all for responses. Processing short tasks and non-NOELIA tasks: all well.

John

Thought I'd follow up on GTX 650 Ti performance...

I have a Quad core (9550) running Win7x64 that is currently set to crunch long tasks only, and has been since the card was installed about two months ago.

The card came overclocked as stock (1071) and then I took it a bit higher (1124).

My current average credit with this card running long tasks exclusively is 75,446.58.

Don't know if anybody else is getting any better or worse....

Operator

---

Going by your runtimes I would say you are getting around 190K/day
http://www.gpugrid.net/results.php?hostid=144671
(86400/32000)*70800≈190,000

---

FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

Last days I clocked my GTX 650 TI with reduced OC frequencies (EVGA Precision 4.0.0): +100/+100 Mhz without extra volting (@stock: 1.087 V), runtimes are within the levels from other users.

My current average credit with this card running long tasks exclusively is 75,446.58.

Actually, I was looking at the wrong computer's stats.

The average credit for the one with the GTX 650 Ti is 150,825.35.

Sorry for any confusion.

Operator

---

This looks to be the right thread for my questions. I confess I don’t understand most of the technical aspects posted here, but I think I get the gist.

Currently I run a modest ASUS GTX 460 1GB for GPUGRID 24/7 (only – no games…) which, I understand, does not use all of the 300+ shaders available. I over-clock it to 850/1700/2000 with no problems. Recent NATHANs complete in about nine hours.

I’m about to invest in an Antec 620W PSU in preparation for a GPU upgrade. Which upgrade?

The budget puts me in the GTX 650/660 (perhaps TI…) range.

Which GPU (I do like ASUS) do you recommend that uses 100% of the shaders and lets me optimise my GPUGRID contribution?

Thank you. Tom

---

The budget puts me in the GTX 650/660 (perhaps TI…) range.

Which GPU (I do like ASUS) do you recommend that uses 100% of the shaders and lets me optimise my GPUGRID contribution?

I recommend the GTX 660. It takes about 6 hours to complete a Nathan long, which seem to be about the same (or not much longer) as for a GTX 660 Ti. And I just completed a new build, and was able to measure the power into the card as being 127 watts while crunching Nathan longs (that includes 10 watts static power, and accounts for the 91% efficiency of my power supply). The GTX 650 Ti is very efficient too, but takes about 9 or 10 hours on Nathan longs the last time I tried it a couple of months ago.

While the GTX 660 Ti adds more shaders, that does not necessarily result in correspondingly faster performance. It depends on the complexity of the work; they found that out a long time ago on Folding@home, where the more complex proteins do better on the higher-end cards, but the ordinary ones can often do better on the mid-range cards, since their clock rate is usually higher than the more expensive cards, which may more than compensate for fewer shaders.

I like the Asus cards too; they are very well built, but to be safe I would avoid the overclocked ones, though Asus seems to do a better job than most in testing the chips used in their overclocked cards. But that may not be saying much; the main market for all of these cards is for gamers, where an occasional error is not noticed, but can completely ruin a GPUGrid work unit.