Change in BA requirements

[mdk777]

OK,
So that confirms that the imbalance in BA to smp ratio was caused by lower smp participation, and not increased BA participation?

[7im]

anyone claiming to know why the client count dropped without a 4 page explanation of all the variables just doesn't have a leg to stand on. Could be people don't like V7. Could be the crappy economy. Could be everyone is moving to mobile devices. Could be an alien invasion. Could be GPUs. Could be an AMD stats glitch. Could be the increasing cost of electricity. Could be Global warming, so they turned off their FAH heaters.

Regarding the client count: we have had a large company donating computer time anonymously and that donation time naturally ran its course (they don't care about BA, etc). That covers about ~30,000 CPUs or so. It's unfortunate timing that that ended around the new year, coincidentally with some of the rough server backend issues we had and the BA discussion.

I am hoping that this group will let us publicly acknowledge their contribution soon as what they've done (and the work we've been able to do on those machines) has been pretty exciting for us.

Well, anyone except him. Kind of changes things, doesn't it.

[Rattledagger]

Rattledagger, what user name are you folding under now?

I've been using Baldurs_Gate in the yearly FAH-Christmas-races and as mentioned is normally blissfully ignorant about whatever happens at FAH for 10+ months/year.

Not having read the 1000+ forum-posts in 2013 does put me at a disadvantage, since apparently the FAH-blog does not seem to mention Core_78 was fully retired in August, but of course it's possible have overlooked a Blog-post.

Not running FAH continuously also gives me the "benefit" of having to search-through the install-instructions and various FAQ's trying to find answers to things that should have been trivial to do without any FAQ or install-instructions like "do not pre-download next wu at 98%".


In any case, being a little more on-topic, Pande mentions a loss of roughly 30k computers but it's a little unclear if they've responsible for the large drop in windows-computers resently or if there'll be an adittional large drop in February...

[PantherX]

FYI, please read the new announcement regarding bigadv experiment (viewtopic.php?f=24&t=25598).

[-alias-]

We'll see how it ends up , it is still 54 weeks until that BA-stop date! As I understand BA will disappears in the form we know today , and all 2P/4P servers can then download smp or whatever PG will let us have, which in practice means that the PPD / Watt for us with such machines drops to 1/3 or less of what it is today.

I've experienced this before when the previous generation BA machines were made to scrap by PG almost over night. Then I was left with 2 pcs . i7-980X and 4 pcs . i7 2600K machines that I could no longer use as before. The investment for this machines where $8,400 at the time, and those machines were sold in parts for a few $. It's practically the same thing that is happening now, just at an even greater cost for the individual BA-donor. For my own part , we are talking about over $30,000 invested hardware that you definitely can not sell after today. Meanwhile, PG have now also announced that no one is going to build 2P/4P for dedicated folding in the future. It will be interesting to see what is happening from now on. I do not think that there will be a full stop from several of the major BA - folders, but it is not impossible. Although I am very unsure of how long I still participate in the big picture beyond the summer. It depends on what the rest of the top 20 list-donors is doing, and I will certainly think very carefully about it.

Edit: Corrected (misreading) error in time until BA stops!

[tear]

Ok, so here's what we know:

BA changes were not dictated by the need to satisfy specific need but to satisfy pretty much
an arbitrary cap (5%).

The cap basically means (meant?) not that the machines meeting specific criteria (feasibility
for the needs) are rewarded but machines in top 5%.
This is huge gap compared to the way the project marketed bigadv ("we need fast machines").
Turns out, it's something akin to "candy for the top 5%!!".

No wonder no rationale was provided, the change had nothing to do with actual needs of the person
running bigadv simulations or amount of non-bigadv work whatsoever [sic!]. Someone just checked
the numbers and said, "oh... bigadv has grown too much, time to close the valve a little bit".

In other words, both changes to thread-number threshold were means to reduce relative number of
bigadv machines, not to take care of any specific need. That is pretty darn sad.

Now the project decided to drop one of the changes, certainly not because it was warranted by
change of needs (since there aren't any) but because, reportedly, donors are angry.

The project decided to bargain instead of realizing and admitting that threshold policy is fundamentally
wrong and needs to be changed to something actually driven by merit.

While announced end of bigadv may seem to indicate planning on project's part, it's really a side
effect of the withdrawal. So, has Folding@Home really learned anything here? I have no idea.

[Bill1024]

We'll see how it ends up , it is still over 16 months until that BA-stop date! unsure of how long I still participate in the big picture beyond the summer.

January 31 2015, is only 12 months and 16 days away from today, not 16 months is it not?


Still all in all, if the PPD for SMP after 1/31/15 is still not close to what people used to get doing BA.
They may not fold SMP. They will build GPU farms or go to WCG.

[kromberg]

Goodbye FAH, you will not get another CPU/GPU cycle from me ever again.

[Leonardo]

Going forward, the next steps will include a discussion of the change of the QRB formula and possibly an update of the benchmark machine. Our plan is to seek more input from donors for both changes.

The context shows seeking input from donors to be in reference to QRB and the official benchmarking machine(s). I strongly encourage Pande Group to seek donor input for hardware considerations in employments additional to QRB and benchmarking. Let me provide input, now: Please consider the not insignificant number of server class machines that your donors already have. Many of us would enjoy continuing to employ these machines in FAH beyond the end of BigADV. Whatever type of work units are available come January 2015, attempt to ensure scalability for 24 to 48+ core machines.


Regardless of the specific content of the 12-month BA notice, thanks for the advance warning. That's definitely a step in the right direction.

[craigyas]

This is a good move on the part of Pande....

maybe they can rig up a system that still gives a similar bonus to the big folders who can complete units quite fast.... it would be a shame to lose all those machines!!

Ill keep my 4p 24 core running until the deadline next year! and maybe beyond!

fold on!

[orion]

Ill keep my 4p 24 core running until the deadline next year! and maybe beyond!

If your intent is to keep running BA that may be a problem. With the 5/01/2014 core count change there will also be a deadline change for the WU per Dr. Pande's post.

1) The posted change in BA requirements will be revised. The only change in requirements going forward will be to require 24 cores (with according changes in deadlines) and that will occur on May 1, 2014.

So your 8431's may not make the BA deadline but if you're only running SMP then you should be good to go.

[craigyas]

Ill keep my 4p 24 core running until the deadline next year! and maybe beyond!

If your intent is to keep running BA that may be a problem. With the 5/01/2014 core count change there will also be a deadline change for the WU per Dr. Pande's post.

1) The posted change in BA requirements will be revised. The only change in requirements going forward will be to require 24 cores (with according changes in deadlines) and that will occur on May 1, 2014.

So your 8431's may not make the BA deadline but if you're only running SMP then you should be good to go.

yeah, definitely took notice of that.... hoping i can still make the deadlines though, fingers crossed!

[Bill1024]

What would be the sense of keeping it at 24 cores and then tighten the return time so they don't make it?
I think they will make the slowest ones just make it and tweek the QRB for BA and SMP.
That would make sense.

[craigyas]

What would be the sense of keeping it at 24 cores and then tighten the return time so they don't make it?
I think they will make the slowest ones just make it and tweek the QRB for BA and SMP.
That would make sense.

That's a good point, even these 16 core processors are still <3ghz. 24 good cores are twenty four good cores, whether it be in a dual 12 core system, or a quad hex core system....hopefully x.x

[Jesse_V]

Goodbye FAH, you will not get another CPU/GPU cycle from me ever again.

I'm sorry to hear you go.

I've enjoyed contributing with my GPU and CPU, and I could only hope to reach bigadv requirements. I'm happy contributing what I can though, I'm sorry that you don't feel that way too.