With PrimeGrid falling apart crunchers will be looking for greener pastures.

Currently my $100 HD5770 performs the same as my $300 GTX570 here at Moo, both completing a “192” wu in about 800 seconds

With PrimeGrid falling apart crunchers will be looking for greener pastures.

Currently my $100 HD5770 performs the same as my $300 GTX570 here at Moo, both completing a “192” wu in about 800 seconds

Why is PG "falling apart"? I had not heard that and my Nvidia cards crunch there. And I do agree my 5770 cards here just ROCK!!

PG is hardly *falling apart*. All that has happened is the credit levels will go down in stages to 50% of what they were. They will still end up the highest "paying" NVidia Project. The adjustment is not before time, the levels were rediculous. They could have managed it better, thats for sure, and some extended notice over intentions would have stopped a lot of anger from those who just bought cards.

But at the end of the day, whilst a few will drift off, there will not be many as they start to realise the realities of the NVidia credit levels at various Projects. PG are a good Project and look after the crunchers well as a general rule, good people there, it was just unfortunate the way it was handled.

Regards
Zy

They could have managed it better, thats for sure, and some extended notice over intentions would have stopped a lot of anger from those who just bought cards.

I agree I was a bit harsh because I’m also one of those.
I also wanted to share in the credit feast at PG and made a mistake when I paid a lot of money for a Nvidia a month ago.
For a bit more I could have got a HD6980
Fact is now I’m stuck with it and would like to use it here at Moo.

I also wanted to share in the credit feast at PG and made a mistake when I paid a lot of money for a Nvidia a month ago.
For a bit more I could have got a HD6980
Fact is now I’m stuck with it and would like to use it here at Moo.

well your second best bet imho would be collatz then..

They could have managed it better, thats for sure, and some extended notice over intentions would have stopped a lot of anger from those who just bought cards.

I agree I was a bit harsh because I’m also one of those.
I also wanted to share in the credit feast at PG and made a mistake when I paid a lot of money for a Nvidia a month ago.
For a bit more I could have got a HD6980
Fact is now I’m stuck with it and would like to use it here at Moo.

It doesn't work here at Moo?

Another question do Nvidia cards work here at Moo?

Another question do Nvidia cards work here at Moo?

There are applications shown on the page below

http://moowrap.net/apps.php

I have seen CUDA31 WUs going through before, but I am not aware of performance etc as I do not have NVidia. Give it a whirl, only take a couple of mins to pull one down to try out.

Regards
Zy

Here is 1 I tried a few days ago with a GTX570:

http://moowrap.net/result.php?resultid=1250199

My original question: Will there be an improvement to optimize the cuda app here

gpu / cpu / credit

AMD HD 6970
1,648.59 / 9.05 / 6,944.00 / Distributed.net Client v1.02 (ati14)

Nvidia Tesla C2070
1,568.38 / 1,516.74 / 1,632.00 / Distributed.net Client v1.02 (cuda31)

Yea it's basically a severely downclocked GTX 480 sprinkled with faerie dust. 1150MHz shader, 3000MHz mem.

It overclocks to 1400 shader and 3200 memory easily on stock voltage (don't even think there are tools to adjust for these) but the card runs almost ~ 90 Celsius while on primegrid so I'm probably not going to oc it much.

Improvements would be nice since it should only be about half as slow as AMD clocked to GTX 480/570 speeds, not four times slower, but I'm not keeping my hopes up.

---

My original question: Will there be an improvement to optimize the cuda app here

Any improvements in that area would come from our upstream (distributed.net) and I don't know if they have any ideas about improving CUDA kernels. There are no clients compiled with newer CUDA runtimes in testing at the moment either.

Wonder who's their CUDA guru/coder anyway, if any.. I personally haven't done any coding or optimizing code in CUDA/CAL/OpenCL platforms (yet) so I'm not an expert to say if current kernels can be improved.

Unfortunately, this means that I personally wouldn't hold my breath on any big CUDA improvements in the near future. :( That doesn't mean they won't happen and I'll share if I hear any news regarding CUDA.

-w