Big advanced core and deadline changes

[dekez]

Planned changes to "Big Advanced" (BA) projects, effective January 16, 2012

Big Advanced (BA) is an experimental type of Folding@home WUs intended for the most powerful machines in FAH. However, as time goes on, technology advances, and the characteristics associated with the most powerful machines changes. Due to these advances in hardware capabilities, we will need to periodically change the BA minimum requirements. Thus, we are shortening the deadlines of the BA projects. As a result, assignments will have a 16 core minimum. To give donors some advance warning, we are announcing this now, but the change will take place in 2 months: no earlier than on Monday January 16, 2012.

We understand that any changes to how FAH works is a disruption for donors, and we have been trying to minimize such changes. For that reason, we are not changing the points system at this time.

However, we want to emphasize that the BA program is experimental and that donors should expect changes in the future, potentially without a lot of notice (although we will try our best to give as much notice as we can). In particular, as hardware evolves, it is expected that we will need to change the nature of the BA WUs again in the future.

Link

I'm assuming big advanced is big beta? I guess this has been introduced to stop people using the hack to allow 8 thread machines to run ba units. If so bad news in terms of points for people like me, but if it is harming the science then a change is definitely needed (I'm not convinced it is but hey ho, another excuse to re-jig the hardware ).

 

[Grimgoth]

i feel this really goes against the spirit of distributed computing AND shows that their programming is lazy....
the tenet of DC for me is that if you need something crunched faster, split it up more and spread it out. If the next WU is dependent on a previous result, then what you need is more done in parallel. If you need the same machine to crunch the same data set, because of inherent randomness in machine crunching (is there such a thing??), then your results are completely random in the first place.
Eg; when I was running molecular docking simulations, you'd hunt for a distribution of probable solutions- and the more times you ran with the same data set, the more likely the "correct" candidate dropped out more times (assuming normalish distribution). You could run it as 1 job x 1000 random seeds, or 4 jobs x 250 random seeds, or even 10 jobs of 100 random seeds; depending on your computing resources available.
I fail to understand the scenario where they justify requiring 16 cores for parallel work.

 

[hanluc]

Link

I'm assuming big advanced is big beta?

No, BIG ADVanced = BIGADV, not just these new super 6903/4 bigadvs. So no one with a machine with less than 16 threads will EVER be assigned any bigadv work unit after the change

 

[dekez]

No, BIG ADVanced = BIGADV, not just these new super 6903/4 bigadvs. So no one with a machine with less than 16 threads will EVER be asigned any bigadv work unit after the change

There's going to be very few people doing bigadv then, I don't see how that benefits anyone

 

[Frozennova]

There's going to be very few people doing bigadv then, I don't see how that benefits anyone

At least team ocuks two biggest hitters won't be harmed, they will probably benefit as the bonus will be better with a shorter deadline/higher hardware requirement.

 

[hanluc]

Here's my current HFM screeshot:

The top one, "main" is a 980x@ 4.53ghz, producing over 100K ppd, that will now be relegated to normal smp, I'd expect the ppd to be less than half that. The one below is a 970, again that too will join the cull and the very bottom one I've just built, a new media PC, thought that could do occasional bigadv too, so I bought a 2700K, I needn't of bothered.

I'm not prepared to fork out running these rigs 24/7 when PG devalue my contribution in half, and at the moment I don't feel like speculating on a upgrade when I've no idea what surprises PG will do next.

Not happy

 

[Biffa]

Similar thing happened in February. They lopped 20% off the bonus on bigadv.

I wouldn't worry about it the # of cores stipulation, the same hack that makes a 8 thread 2600K look like its got 12 threads will make it look like its got 16 threads as well.

Whether the return time will let it get in on time is another matter, that would probably depend on how ninja your overclocking skillz were

Would also be good if they upped the points on SMP and maybe GPUs as well, mind you with GPU's you already get a huge point per atom advantage anyway:

GPUs gets around 10ppd per atom in a wu. (Say 16k gtx 470,avg )
BIGADV gets 0.17 ppd per atom in a wu. (450k my 4p ,6903)

Thats because there are over 2,600,000 atoms to crunch in a 6903 WU and under 2,000 in a 7621 GPU WU

Bigadv does more complex calculations with less wu overlap.

Bit more info:

Originally Posted by Vijay

Sorry for not being more clear in the post. It's 16 physical cores in terms of speed (to make the deadlines), but since the client just looks for logical cores (and can't tell the difference between logical and physical cores), the client hard minimum will be 16 logical cores (if we made the client minimum 32 cores, then there would be cases of people with 16 physical cores not working). Since this is a bit confusing to say, I tried to stress the speed needed (16 physical cores), but as before, the key determining issue will be making the deadlines (especially since some people often spoof the number of cores the client detects anyway).

We very much understand how upsetting any change is, especially since donors work hard to optimize their setups. That is why we tried to give advance notice, especially since people might buy new equipment over the holidays. Please keep in mind that BA is very much experimental and that future changes not just could happen but most likely will. We are making this change since as time marches on, pretty soon all new machines will be BA-capable, which defeats the purpose of the whole BA project.

Moreover, we understand that change is upsetting for donors, so we are trying to be much more conservative with changes to other parts of FAH.

Basically the cost to get into legit bigadv folding is now just under a grand and thats if you buy new.

That covers entry level dual 8 core G34 opty setup or 1366 Xeon setup (including memory, mobo, VAT and delivery)

 

[dekez]

When this comes in I'll probably switch back to cpus on boinc and gpus on folding, not good news for my seti output but at least the extra heat that folding on gpus creates is in time for winter.

 

[Grimgoth]

When this comes in I'll probably switch back to cpus on boinc and gpus on folding, not good news for my seti output but at least the extra heat that folding on gpus creates is in time for winter.

you actually telling me that your choice of application to run on your gpu is determined by how much you want to heat your room??

 

[dekez]

you actually telling me that your choice of application to run on your gpu is determined by how much you want to heat your room??

Well it's one of the deciding factors , in the summer it's to hot to run folding on the 460s as they end up either fans ridiculously loud or the gpus crashing / clocking back. Seti doesn't seem to stress them as much, but in the winter folding on them is doable and heats the house more (haven't had to turn the heating on yet!)

One day I'll build my dream house, centrally heated by a massive farm in the basement I hate the idea of using energy just to create heat, when it be a by-product of something useful like crunching.

 

[Jamiman]

you actually telling me that your choice of application to run on your gpu is determined by how much you want to heat your room??

My PC heats my room. That's why I'm sticking with 4 series cards