The new applications will deliver around 30% more speed for GTX260, GTX280 cards using better hardware architecture. Approximatively the same speed as before for other cards. Thanks to Nvidia for letting us having the test cards.

gdf

That's great!
Time to get some new graphics cards... ;)

---

pixelicious.at - my little photoblog

That's great!
Time to get some new graphics cards... ;)

me too ;)

Very nice!

Now someone should tell Arctic Cooling that we need something like the Accelero S1 for the GTX 260/280 ;)

MrS

---

Scanning for our furry friends since Jan 2002

When will the new application be available? With which version number? At the moment the 6.45 is very slow, I need over 10 hours instead 7:40 with the application 6.43.

---

When will the new application be available? With which version number? At the moment the 6.45 is very slow, I need over 10 hours instead 7:40 with the application 6.43.

Do you use ncpus +1?
I've made some tests for GDF and noticed a slowdown from 45 ms/step to 72 ms/step with Vista64 and ncpus +1.

With Linux 64 there's no slowdown with ncpus +1...

---

pixelicious.at - my little photoblog

Do you use ncpus +1?
I've made some tests for GDF and noticed a slowdown from 45 ms/step to 72 ms/step with Vista64 and ncpus +1.

With Linux 64 there's no slowdown with ncpus +1...

Yes, I use ncpus=5, but I can't see any slowdown through this entry.

Time per step: 41.451 ms

Looks pretty fast. The slowdown comes with the beginning of the application 6.45. I've written the cc_config.xml 2 days later.

---

Ok, you don't have to believe me. ;)
But here are the WUs with ncpus 5 - 66472, 66791 - both with 72ms/step, and one weirdo ( 67339 ) with 180 ms/step.

After deleting the cc_config.xml the times went back to consistent 42-45 ms/step, like it was without ncpus +1.

After looking at your results I saw most of your WUs are in a range of 41-43 ms/step. I would think that your GTX280 should be a little bit more faster compared to my GTX260...

We also tried that and a few other things on a test project with another app. and it showed the same slowdown with ncpus +1 on Windows, therefore GDF was asking for a nanosleep solution for Windows here in the forums...

---

pixelicious.at - my little photoblog

Finally - it was worth buying the 260 in the end ;-) Is The app autodistributed or do we have to reset the project (when switching from 6.43 to 6.45 i had to..)

After looking at your results I saw most of your WUs are in a range of 41-43 ms/step. I would think that your GTX280 should be a little bit more faster compared to my GTX260...

I've checked again without ncpus-entry and now

# Time per step: 35.748 ms

The real running time was 8:24 instead the 7:40 with the 6.43, so it's still a little bit slower also without the ncpus-entry.

EDIT: btw, with ncpus=5 the temperature of the GPU was between 71°C and 73°C, without this entry the temperature is at 79°C.

---

Nice!

..that are great news. Because I bought a 260² one to crunch for PS3Grid... ;-)

So... let us know when the new version will be released :)


Thanks and happy crunching... :)


-
Edit:

..dont know whether this is important.

But keep in mind that there is a 2nd version of the 260 out with even more ALUs (216 instead of 192 ALUs..)

:)

But keep in mind that there is a 2nd version of the 260 out with even more ALUs (216 instead of 192 ALUs..)

and since in gaming it's just a few percent faster than the previous 260, it shouldn't be much more expensive. Right now this is not the case, but if things settle down properly than this might be the card to get.

MrS

---

Scanning for our furry friends since Jan 2002

quite right..

...never the less - it might be that someone out there has already this card... even if it is too expensive at this moment. ;)

..so my post was just a reminding to think about this 2nd editin to be implemented, too.

quite right..

...never the less - it might be that someone out there has already this card... even if it is too expensive at this moment. ;)

..so my post was just a reminding to think about this 2nd editin to be implemented, too.

The 260² you bought should already be the 2nd generation with 216 shaders.
Should be ~10% faster than the orginal 260

My crunchtime with my 260 @ stock speed was ~9.8h --> Time per step was: ~41.600 ms.
A little OC-action and I brought it down to ~8.8h --> Time per step is: ~37.500 ms
That should be the roughly estimated speed of the 260² @ stock clocking.

Cu KyleFL


Btw -- 2x 8800GTs are faster than a single GTX260 :(
2 WUs every 15.9h (time per step: 67.500ms)
My GTX260 needs 17.6h for 2 WUs.
Now I just have to get that promised 30% speed gain to strike back *g*

u r right... it is the ²(nd edition) with 216 ALUs.

:)

lets see.. after releasing the new version optimized for 260/280 GTX :)

Mehlano:

How fast is your 260² ?
Time per step and the time for finishing one WU would be interesting for me!
(stock speed or oc´ed ?)

How much did you pay?



Cu KyleFL

..so my post was just a reminding to think about this 2nd editin to be implemented, too.

Do you mean the optimization should be implemented for the new cards as well? I'm quite sure this should work automatically :)

MrS

---

Scanning for our furry friends since Jan 2002

..so my post was just a reminding to think about this 2nd editin to be implemented, too.

Do you mean the optimization should be implemented for the new cards as well? I'm quite sure this should work automatically :)

MrS

That is exactly what I meant, yepp. :) Now we got each other... :P

Uuuuh.. I bought the Zotac GTX 260 AMP² for 289 €. (ow my god..my hard earned cash :( )

..and it's not even so fast at PS3Grid as I hoped it would be. ;-)

(thats also the reason why I was so happy to hear about optimized app..)

- Time per step: 44.576 ms
- Time per step: 39.264 ms
- Time per step: 40.022 ms

..last 3 WUs. Average duration time per WU is about 9 hours.. :/

For example.. with my old GPU (8800 GTS / 640 MB / G80) and the same system I had ~11k at the 3DMark06 (!) default benchmark.

Now with the Zotac GPU it was about ~18k points at 3DMark06.

It is the stock version .. but even the AMP version. (And yes.. this one is even more expensive than the standard ² version as well... and the default version could be oc'ed to AMP version, too. :)

Hmm.

A very strange thing is happening on my host.

I´m using BOINC 6.3.10 with default settings (no messing around with ncpu+1 or something like that)

After I uploaded a new WU for GPU Grid today I found my PC crunching happily two CPU Seti units alongside one GPUgrid WU....
-- Can´t say what impact that has on the GPU-crunching time...
I´ll have to wait until the WU is finished to take a look on the time per step.



Cu KyleFL

Don't want to be rude, but this related how to "New applications coming up with 260,280 optimizations"? ;)

MrS

---

Scanning for our furry friends since Jan 2002