I see, Cool thanks! I don't think I could do half that many.
Lets see if they average 700sec/wu thats around 125 a day the C1060 can do say 300 for a full Tesla. So even if most are shorties that gives a good amount of headroom.

---

"The most exciting phrase to hear in science, the one that heralds new discoveries, is not Eureka! (I found it!) but rather, 'hmm... that's funny...'" -- Isaac Asimov

I see, Cool thanks! I don't think I could do half that many.
Lets see if they average 700sec/wu thats around 125 a day the C1060 can do say 300 for a full Tesla. So even if most are shorties that gives a good amount of headroom.

From a quick count, my GTX295 returned 144 WU's on 7th Feb. That probably includes a fair few VLAR's which take 6 x as long as a mid-range WU. That lines up with my own estimate that each core of the 295 is equivalent to the 4 CPU cores of my o/c'd quaddie added together (or will be when the VLAR problem is sorted). So the 500 limit for each GPU core is, as you say, generous.

F.

---

@ Fred W

Why you don't use Raistmer's V7 mod with 'VLAR- kill'?

---

@ Fred W

Why you don't use Raistmer's V7 mod with 'VLAR- kill'?

Because:

1. I'm inputting into Richard's survey and, for that, we need all AR's covered.

2. The biggest problem the project has ATM is bandwidth rather than lack of crunching power and if I "kill" them then they have to be downloaded to someone else (who may well "kill" them etc...) which all increases the load on the servers. So I have decided to be public-spirited and just let them run.

F.

---

My GPU Grid experience is that the 280 and the core in the 295 are comparable in performance ... the 280 may be a shade faster, but the 295 has two cores so you get twice the work done for the same power draw ...

My GPU Grid experience is that the 280 and the core in the 295 are comparable in performance ... the 280 may be a shade faster, but the 295 has two cores so you get twice the work done for the same power draw ...

Basically if you want single process crunching, 280 is better, but we're talking about distributed computing here ;)

Tesla has 4GB of RAM for ability to crunch down through more input data.
For the CPU units, these roughly take 40-45MB of RAM each, I assume then, that it won't make much difference whether your card has 512MB or 4GB.

Who knows, maybe in few years time we will see games needing 4GB or more on graphics cards...

---

--
Get up, stand up! Don\'t give up the fight!
Credits will make everybody feel high! ;-)

YEs, which is why I will be getting GTX 295 cards or equivalent from here on in ... I am not disappointed with the 9800 GT or the GTX 280, but, the 295 is the way to go ...

The next system I build will have room for 3 though likely I will not get three cards out of the gate ... or may be I will, who knows ... :)

Personally I am leaning more toward a couple oc'ed C870's in an i7. I am thinking they make the best rac per dollar going now. At $1k each you get a far more power efficient chipset (~170w) and around 3x the rac of 295 here on seti and maybe 50k rac at ps3grid. Just sent Richard a new data dump so we can see how the c1060 did overnight on the VLAR's.. :)

---

"The most exciting phrase to hear in science, the one that heralds new discoveries, is not Eureka! (I found it!) but rather, 'hmm... that's funny...'" -- Isaac Asimov