Bionic and 7970

Milkyway gives the most points for AMD/ATI cards.

I think Poem now tops that, they've only recently introduced GPU folding, and even more recent on Nvidia cards, my single GTX 570 at stock was doing 470,000 pts/day running 3 instances at once, I don't about ATi cards but most of the top producers are using them. CPU crunching on POEM is next to useless in comparison.
 
I was up to 34-36s yesterday, but have reset my OC to stock (1050) since I was getting driver crashes, and also the heat was a bit much with a volt OC. WUs now taking 41s , but I am watching videos and using my computer more, so I guess this variance is to be expected.

So Biffa does that mean a single 7970 can output as much as your MangyCours rig? 400k on MW vs 400k folding@home? Would love stanford to make a gpu client that was as good as the bionc ones, they really should be on top of it, since they're leading the distributed computing charge.
 
Gah ever since I installed bionc yesterday I've had nothing but hassle with my network. Eariler today I was running it fine, but now my network speed and latency has just dropped off a cliff. I've stopped running the bionc client and rebooted my pc+router and it hasn't restored my broadband speed.

I think the bionc client has changed my network settings or something, how can I fix this?
 
So Biffa does that mean a single 7970 can output as much as your MangyCours rig? 400k on MW vs 400k folding@home? Would love stanford to make a gpu client that was as good as the bionc ones, they really should be on top of it, since they're leading the distributed computing charge.

You would have to compare something that is similar to folding doing the same work I guess, MW is doing different science on a different platform, you can't compare the two :)

Plus its all relative to the other work being done in that project by the people in the project, not relative to what going on in other projects, even boinc ones really.

As you probably know gpus lend themselves to doing a certain type of work whilst CPUs lend themselves to other types of work, thats why there are projects that still don't run on GPUs because its not the right sort of work. Folding and the other projects that do both CPU and GPU usually have different types of work units doing different things with differing load levels, e.g. I know gpu work units in F@H have much smaller molecules than the big CPU work units.

Also depends on what the project wants to get done, and how they reward people for crunching certain types of data, if they really want a certain type crunched early they will boost the bonus points on that type of work unit to make it more attractive to get done quickly.
 
Just a thought on your network - is there any chance you're incurring the wrath of your ISP and bumping up against a fair usage policy?
If WU's are completed every 40s or so there must be a nearly constant stream of data being sent and received from your end.
 
Think I've found out the cause of my network outage everytime BOINC is in use.....HEAT

My small PCI wireless card 1-2 inches above the 7970 and my router downstairs is fine, but my wireless network card above the gpu feels like it's about to melt. I assume that's the cause of the drastic dip in bandwidth when I'm running GPU MW@home ?

Got the case open now and temps have fallen and my network has gone back to normal. If I close it up and it drops off again it will most likely be the cause.

I can't connect to the net via other means since my router is downstairs and no upstairs line...


I guess I could underclock my GPU to lower the temps. Better than nothing?
 
Last edited:
yeah I just tried that out and it seems to work, setting my fan to 100% before I run the client and it keeps the temps, and the pc from getting too toasty and the network from falling off a cliff. Only problem is the racket :mad:
 
Back
Top Bottom