What is everybody crunching lately? And which is a better GPU for such, AMD or NV?

[VirtualLarry]

I recently purchased two 7790 1GB cards, and I'm contemplating getting two Asus 2GB GTX650 Ti Boost cards.

Primary projects for me would be Einstein@home, Milkyway@home, PrimeGrid, and WCG. Might get into F@H, but I've gotten semi-bored with that. It seems the biggest points in F@H go to those guys with 16-core systems. I guess I'd rather crunch prime numbers on mid-range GPUs.

Which GPU brands do better at those projects?

And what are you folks all crunching these days?

Any projects that TA is involved in, that could use some GPU grunt?

 

[Vertaxis]

What you need to ask is what projects are using single precision floats vs. double precision floats. Both AMD and nVidia are good for single precision. For double precision, like with some Milkyway@home apps, AMD Radeons are the better cards.

You should google for the comparison charts for these cards. Wikipedia has a good set of charts.

 

[ZipSpeed]

I'm still full steam on WCG. I couldn't resist, and bought a second 7950 thanks to all the fire sales happening with retailers trying to get rid of 7000 series stock. I've been putting the 7950s on POEM but work seems to be sporadic and few in between. I prefer to stick with humanitarian/biology projects but it seems to be quite limited for AMD GPUs.

 

[VirtualLarry]

I'm doing a mix of Einstein@home, primegrid, WCG on my HTPC (LLano quad-core).

The only WUs I get for WCG are Clean Energy Project, and only one at a time.

 

[Rattledagger]

Which GPU brands do better at those projects?

Well, while waiting for WCG to release a new project, I've been mainly crunching Climateprediction.net with SIMAP as back-fill in case runs-out of CPDN-work, and neither project uses GPU so GPU-type doesn't matter.

 

[Sunny129]

well i'm a regular Einstein@Home and Milkyway@Home cruncher. TeAm member Faxon is bringing two 5870s online to crunching for the winter, but we can always use more grunt. your 7790s have pretty dismal FP64 performance, but they are capable, so you can crunch MW@H with them.

i've also found that AMD parts run better on Einstein@Home these days too...i used to have a GTX 680 and dual GTX 580s on the project, but soon replaced them all with 7970s due to substantially better run times. i have no idea how the 7790 stacks up against its direct nVidia competitor in this arena.

i'm sure nVidia parts run better on a number of projects...i just don't know if any of them interest you...and i don't know that nVidia parts will run better on any of those projects (MW@H and E@H i know for sure, and PG and WCG i have no idea)...

 

[salvorhardin]

I recently bought a 7950 and currently using it to complete my 50 million in milkyway and will then switch it over to f@h so I can have it set up and ready to go for the december race. I'm mostly been running projects in order to complete milestones and the ones I do run I check to see how much more work can be done with a gpu compared to a cpu and plan my resources accordingly.

 

[biodoc]

I have my GTX570 and GTX660Ti on GPUgrid which is an Nvidia/cuda only project.

Here's a link to find out more about the project: http://www.gpugrid.net/about.php

 

[Bubbleawsome]

I only do F@H but AMD and GK110 are the kings here.

 

[Markfw]

I only do F@H but AMD and GK110 are the kings here.

According to this:
http://www.overclock.net/t/475163/gpu-projects-ppd-database-revived

Nvidia are kings, and I think $/ppd 660ti at a little over $200 each are kings for 72k ppd

 

[Bubbleawsome]

According to this:
http://www.overclock.net/t/475163/gpu-projects-ppd-database-revived

Nvidia are kings, and I think $/ppd 660ti at a little over $200 each are kings for 72k ppd

Dang, last I checked we had 7950s as the $/ppd king with nothing but titan rivaling even the 7970.
http://www.anandtech.com/show/7457/the-radeon-r9-290x-review/18
I've been choosing my cards based on that. :/

 

[VirtualLarry]

Just for confirmation, if I were to start Folding again, I can't do "bigadv" WUs on a Thuban (hex-core AMD), right? And there's no early-completion bonus for SMP WUs? Nor for GPU WUs?

Just wondering what kind of PPD I could get out of a box with a 3.5Ghz Thuban, DDR3-1600 RAM (Gskill, won't overclock worth beans), and either a pair of 7790 1GB cards (which I already have), or a pair of GTX650 Ti Boost cards (which I might acquire).

 

[kb3edk]

I recently purchased two 7790 1GB cards, and I'm contemplating getting two Asus 2GB GTX650 Ti Boost cards.

Primary projects for me would be Einstein@home, Milkyway@home, PrimeGrid, and WCG...

(snip)

I got a GTX650 myself a few weeks ago on clearance sale at the local computer store and have been using it on PrimeGrid for the TeAm, it's easily over 100Kppd most days.

In fact, I just got an exciting E-mail from PrimeGrid yesterday, over the weekend my card found what was (at the time) the 298th largest prime number in recorded history!

[size=-1]Dear Primefinder,

Congratulations! Our records indicate that a computer registered by you has found a unique prime number. This computer is running BOINC, is attached to the PrimeGrid project, and is assigned to the Proth Prime Search. What makes this prime unique is that it's large enough to enter the Top 5000 List in Chris Caldwell's The Largest Known Primes Database.

Since you have auto-reporting selected, the following prime was submitted on your behalf:

Added 116125 : 731*2^1968039+1 (592442 digits)

If you have any questions or concerns, please contact us and we will surely resolve any problems.

Once again, congratulations on your find! Thank you for participating in PrimeGrid.[/size]

 

[Traxan]

I'm back with Einstein@Home. But since my PC is in my bedroom, I don't run it at night. The GTX670 fan is annoyingly loud.

I tried Seti@Home for a while, but I feel that the constant XYZ rotation of the animation is an awful waste of GPU cycles. They really should take that out and hold it steady, not rotating in all 3 dimensions while animating the crunching at the same time.

 

[PCTC2]

Still at F@H after all these years. To be honest, I haven't folding with GPUs in a few years (GTX 260's were the last GPU I folded on). I have 3x SMP (generally around 32 threads) servers that I'm currently running for around 220k PPD each.