Hey all,

So as you have read, I bought a Nvidia 470 that can't currently be used with MW but can with SETI. Yes, bummer....

Please respond if you know of other applications that currently can do so and I'll add them.

Get crunchin'


Dave

---

Hi Dave,

GPUGRID has fermi-enabled wu's. If SETI (cuda23) works, you could try Einstein. But this is a not very effective app, it runs in an 1CPU/1GPU configuration.
If your card has at least 1GB of Ram, you could try to use an app_info and set it to 1CPU/0.5GPU. But still not very effective.
LATTICE HP says, they have Cuda-wu's, bur I never got one.
You could try Collatz Conjecture, they have wu's for ATI and Cuda. I never checked fermi-compatibility.

BOINC HP gives an overview about GPU-enabled Apps, but it's not complete.

Regards,
Alexander

Hi Dave,

I have tried Collatz and DNETC on a GTX470 - no problem. Also tried GPUGRID a couple of weeks ago but had problems due to the GPU combination in my PC (low end on board CUDA device and the gtx470 don't work well together). A single Fermi should be ok, though.

I haven't tried Einstein with the gtx470 but from what I hear the Einstein guys where the only ones who got their CUDA app right from the beginning - no recompiles were necessary where Fermi was released. It just worked. Keep in mind, though, that the Einstein CUDA app is no very effective since they are only doing an FFT on th GPU. Everything else is still done on CPU. That's why it needs 1CPU/1GPU as Alexander already mentioned. There are several threads in the Einstein forum that discusses this subject.

Regards,
Michael

Hi Werkstatt,


GPUGRID is running perfect on my 470. Thanks for the advice!

Get crunchin'


BDDave

---

Hi Dave,

A few minutes ago I've found a post somewhere here, that 'They' are working on an fermi-enabled app for MW.
Maybe it's a good idea to come along from time to time and check that.

Regards,

Alexander

You can calculate SETI@home Enhanced (MultiBeam) on a GTX4xx series GPU.
They have a 6.10_cuda_fermi_app.
AFAIK, you can let run 3 SETI@home CUDA WUs simultaneously (with app_info.xml file) on a GTX4xx for best performance.

Have a look in the SETI@home NC subforum for more infos.

---

For sure, I haven't given up on my MW bar yet!

BDDave

---

At for SETI I've got about 30 WU waiting to be received. The are currently off line from Tuesday to Fridays now.


BDDave

lol
einstein was the last who wanted to test cuda and still has it not working very effective.
you gain maybe half an hour while it is constantly running on your cpu and your gpu, points wise i think hardly any change either.
The projects cuda wise are gpugrid, collatz, milkyway and seti.
But i wonder if it is wise to use a fermi on milkyway since ati rulez there.
Nvidia cards don't do double precision work very well.
Seti has the problem that it is hard to get units for cuda

---

Its new, its relative fast... my new bicycle

Seti has the problem that it is hard to get units for cuda

No, today I crunched ~100 wu's, yesterday ~150.
They stopped their servers now, but on friday they distribute new wu's.

Alexander

I crunch (normally) only SETI@home.

Between Tuesday afternoon and Friday afternoon my time (MEST) the server are down.
Between Friday afternoon and Monday afternoon you have a 'WU limit in progress'. The enabled actually limit is published in the Tech News subforum.
Between Monday afternoon and Tuesday afternoon the limit is much increased or disabled, depend of server usage.

My BOINC DLed ~ 1,400 CUDA and ~ 400 CPU WUs between Monday afternoon to for a few hours. ~ 10 day WU cache.
Intel Core2 Duo E7600 & OCed GTX260-216.
Too much CPU WUs, because the DCF and flops (app_info.xml- and server-flops) are not finished balanced.

---