Bigadv points change

Moderators: Site Moderators, FAHC Science Team

Jester
Posts: 102
Joined: Sun Mar 30, 2008 1:03 pm

Re: Bigadv points change

Post by Jester »

Thanks,
Might have saved me an exersise in futility.... :ewink:
But now that leaves only two choices....... :?
RoomateoYo
Posts: 30
Joined: Tue Oct 19, 2010 5:11 pm
Hardware configuration: i7-2600k @ 4.4 GHz
GTX780

i7-2630QM
GT555M
Location: Jackson, MI

Re: Bigadv points change

Post by RoomateoYo »

I just noticed on the Project Summary that the bigadv base point values have gone up.
Jester
Posts: 102
Joined: Sun Mar 30, 2008 1:03 pm

Re: Bigadv points change

Post by Jester »

RoomateoYo wrote:I just noticed on the Project Summary that the bigadv base point values have gone up.

Hmm,
I'll have to check that out,
All Bigadv Wu's or only the "Big" Bigadv ?
I can't see why they would though....
Last edited by Jester on Fri Jul 08, 2011 3:38 pm, edited 1 time in total.
Racer43
Posts: 35
Joined: Tue Jan 05, 2010 1:31 pm

Re: Bigadv points change

Post by Racer43 »

IMHO, I feel that the supply and demand issue is real. As Jester noted, machines not originally targeted for bigadv were found to be capable of running bigadv (4 & 6 core/HT i7, a few hex AMD) via tricking the download server into thinking there were enough cores to run bigadv. Hence, more machines were available to run bigadv than anticipated. This, then, led to a short supply of bigadv, which then started to drain the availability of other smp units. I don't think the Pande group anticipated this turn of events, or this quick of a migration of more capable machines.

My reasoning is this; how many of us have migrated from uni to smp, or smp to bigadv? I have. Even with a slow dual core, I am able to return small smp with 1/3 of the preferred deadline to spare, which then negated the need to use dual uni, especially after the newer version of smp came out a few years ago. If we check the server status, we see a column listing those work units available. Of the over 3 million units available as I post this, over 1 million are classic units, over 1 million are GPU, and about 1/2 million are PS3. That leaves not that many left for smp/bigadv; most of the smp servers show a very critical shortage.

To quote Dr. Pande, "...my group's expertise is in computational methods, biology, and drug design... " (viewtopic.php?f=16&t=19098). In any business, the ability to perceive the needs of the future is difficult beyond belief; all we can base future needs upon are past data and information. Note that the Federal Government is having a problem trying to fund the Federal Highway Administration; with the big push for fuel efficient automobiles, less in gasoline tax is being collected, hence less revenue which can go into the roads fund, but the constant the demand for more road availability. Nobody really thought about this back in 2007/2008 when fuel prices were skyrocketing. These are scientists, not MBAs. Does Dr. Pande see that there are more contributions available (demand) than units available (supply), and as SK Thinker said, these contributions (demand) do not have a shelf life? I believe he does. However, is there such a thing as a quick fix in this matter? I do not think so. Hence, I feel Dr. Pande's group needed a stopgap measure, which turns out to be the change QRB. Note how also uni points have increased? Classic invisible hand of supply and demand (thank you Adam Smith). To me, Dr. Pande is trying to encourage migration to that area where more help is needed, to wit, classic units, while trying to project what is needed now and what will be needed in six months, one year, two years, etc.

Does this seem like a minimization of our investment in higher end machines, this lack of full utilization? On the surface, it does. Can this be rectified? Dr. Pande's group is pedaling as fast as they can. Is Dr. Pande and his group indifferent to our wants? No, yet he has to balance science with our needs; at times, science takes the front seat over our needs. I feel he may not have seen just how fully committed we are, even without access to farms or servers, and just how willing we were to upgrade our personal systems with F@H in the forefront and our needs as secondary. IMHO, this is where we are right now. Evolution brings new understanding, understanding which may not have been apparent in times past. Time will tell, frustratingly enough as that phrase is. But we are not forgotten; Dr. Pande is righting the ship while trying to maintain that ever importance balance.

My .02 cents.
Team 163828, Always Broke Folding :mrgreen:
Folding towards the Top 1000 teams with: a Phenom II X2 laptop and an Athlon 64 X2 5000+ with a Sapphire HD 6570 :shock:
Jester
Posts: 102
Joined: Sun Mar 30, 2008 1:03 pm

Re: Bigadv points change

Post by Jester »

I can find little in any of that to argue with, :ewink:
But i still feel that a heavy, across the board reduction will have less of the desired effect than reducing deadlines for more intensive work units,
That way at least, only those systems that are more than capable, ie: the original target of 8 cores or 12 with the new Wu's will be viable,
those that are more suited to regular smp work can be used for such,
I'm sure those using "marginal" machines for Bigadv Wu's always thought, just a little at least, that things would change if availability became
limited, but I don't think those with high end, dedicated systems anticipated such a devaluation.....
With the hyperthreaded 4 and 6 core i7 and "SB"Cpu's there was no real "tricking" done as the O/S rightly or wrongly reported the hyperthreaded cores
as real, where the "tricks" come in is having less real or hyperthreaded cores than minimum and having something report more...
Grandpa_01
Posts: 1122
Joined: Wed Mar 04, 2009 7:36 am
Hardware configuration: 3 - Supermicro H8QGi-F AMD MC 6174=144 cores 2.5Ghz, 96GB G.Skill DDR3 1333Mhz Ubuntu 10.10
2 - Asus P6X58D-E i7 980X 4.4Ghz 6GB DDR3 2000 A-Data 64GB SSD Ubuntu 10.10
1 - Asus Rampage Gene III 17 970 4.3Ghz DDR3 2000 2-500GB Segate 7200.11 0-Raid Ubuntu 10.10
1 - Asus G73JH Laptop i7 740QM 1.86Ghz ATI 5870M

Re: Bigadv points change

Post by Grandpa_01 »

Jester wrote:Thanks,
Might have saved me an exersise in futility.... :ewink:
But now that leaves only two choices....... :?
Yes you do have a couple of choices. The question to ask yourself is why did I do this to begin with. Was it for the competition or the science or both. If it was for the competition then that is still there. X machine is still going to make X amount of points more than Y machine is and Y will never catch X even if they were making the exact same amount of points per WU. X is always going to produce more WU's than Y is.

Was it for the science if it was then you are still producing the same amount of science as before it just has a different arbitrary number assigned to it which has no actual value.

Was it both guess what both reasons are still there X is going to out produce Y in points / value, the science value is still there. So to me there is only 1 clear choice keep folding and continue to have fun doing it. Feel good about maybe helping find a cure for some disease and being a part of it. All pieces of the puzzle are needed to put the puzzle together no 1 piece is greater than the other it is just that the pieces along the border are easier to find. So in reality how do you define a value to any piece. It pretty much depends on which piece you need at that moment. :wink:
Image
2 - SM H8QGi-F AMD 6xxx=112 cores @ 3.2 & 3.9Ghz
5 - SM X9QRI-f+ Intel 4650 = 320 cores @ 3.15Ghz
2 - I7 980X 4.4Ghz 2-GTX680
1 - 2700k 4.4Ghz GTX680
Total = 464 cores folding
Racer43
Posts: 35
Joined: Tue Jan 05, 2010 1:31 pm

Re: Bigadv points change

Post by Racer43 »

To true, Jester. I agree with deadline reduction over point reduction scenario, yet there would be a whole different section vocalizing against the change, stating almost the same arguments. If I had to go back to classic, I'd probably stop again myself.

No true tricking, but remember when the HT i7 came out and the whole argument over logical versus physical cores? To me, same issue here. I feel for you and your investment. I, too, am looking to add another system in my house (HTPC), yet every time I come to a decision, I look at the impact that this system will have folding. As of yet, I still have not purchased the second system. I could very easily get another A64 X2 and a PCI or AGP card and be done for under $200, yet I keep coming back to hex core AMD and waiting for prices to be where I am comfortable so as to pull the trigger, because the hex core is way beyond the HTPC needs.
Team 163828, Always Broke Folding :mrgreen:
Folding towards the Top 1000 teams with: a Phenom II X2 laptop and an Athlon 64 X2 5000+ with a Sapphire HD 6570 :shock:
bruce
Posts: 20824
Joined: Thu Nov 29, 2007 10:13 pm
Location: So. Cal.

Re: Bigadv points change

Post by bruce »

Racer43 wrote:My .02 cents.
Thank for a very coherent summary of the state of affairs. Let me go one step further with some facts and some hypothetical extrapolation. (Not that I expect anything like this to actually happen . . . but when we're talking about a long-term solution, nobody knows what might happen.)

Though a client that can automatically reconfigure itself has not been proposed, I can conceive that it might be possible to write one. Suppose a machine with N cores (N presumed to be large) can't get bigadv and it's reassigned to standard SMP. That works, but there's constant bitching. If it went one step further and automatically kicked off N uniprocessor WU if, for some reason, it couldn't get SMP WUs, I can't imagine the uproar we would hear. We certainly need a points system that encourages big hardware to work on tough problems but we need acceptable solutions to what to do with those systems when the current tough problems have been solved and we're waiting for the scientists to propose new problems, generate projects, perform appropriate stability and validity testing on those projects, and release them to FAH. (This second case is probably more appropriately applied if N=2 or 3 or 4 than if N=32 or 48, but the concept would still be the same.)

It's fair to say that there's always important science to be done but sometimes there's a shortage of a certain level of complexity. Considering donor attitudes, project availability, hardware availability, points, bonuses, deadlines, etc. (and anything else that might be considered part of the big-picture aspects of FAH) what can the Pande Group do to make the overall system operate smoothly under changing conditions?

I think my example demonstrates that somewhere along the line there's a disconnect between various portions of the points system. That point was made very clearly by "Racer43"
If I had to go back to classic, I'd probably stop again myself.
Jester
Posts: 102
Joined: Sun Mar 30, 2008 1:03 pm

Re: Bigadv points change

Post by Jester »

Grandpa_01 wrote:
Jester wrote:Thanks,
Might have saved me an exersise in futility.... :ewink:
But now that leaves only two choices....... :?
Yes you do have a couple of choices. The question to ask yourself is why did I do this to begin with. Was it for the competition or the science or both. If it was for the competition then that is still there. X machine is still going to make X amount of points more than Y machine is and Y will never catch X even if they were making the exact same amount of points per WU. X is always going to produce more WU's than Y is.

Was it for the science if it was then you are still producing the same amount of science as before it just has a different arbitrary number assigned to it which has no actual value.

Was it both guess what both reasons are still there X is going to out produce Y in points / value, the science value is still there. So to me there is only 1 clear choice keep folding and continue to have fun doing it. Feel good about maybe helping find a cure for some disease and being a part of it. All pieces of the puzzle are needed to put the puzzle together no 1 piece is greater than the other it is just that the pieces along the border are easier to find. So in reality how do you define a value to any piece. It pretty much depends on which piece you need at that moment. :wink:
That makes the most sense of anything posted in the last week or so on the subject,
and had me looking back at exactly those things, strongest are the memories of those
I Fold for, three of them are here:

27/10/2002
Folding on.

So another week over,
a few more Wu's to the score,
thinking and planning,
how to fold a few more,
another machine in the pipeline,
would sure help the fight,
have to consider the cost,
with a budget that's tight,
the costs may be high,
to keep folding on,
but the rewards are so great,
if we can save only one,
One mother of a child,
who can't understand,
where mummy has gone,
we must lend a hand,
One husband and father,
the loss is so great,
to the family left behind,
we must concentrate,
One innocent child,
the chance to grow old,
taken away,
how can we not fold,
so, fellow folders,
the costs may be great,
but not folding at all,
I will not contemplate.

All those years ago I had already answered my own question.
Only one option left now...... :ewink:
bruce
Posts: 20824
Joined: Thu Nov 29, 2007 10:13 pm
Location: So. Cal.

Re: Bigadv points change

Post by bruce »

Racer43 wrote:I agree with deadline reduction over point reduction scenario, yet there would be a whole different section vocalizing against the change, stating almost the same arguments.
The arguments would be similar but as you say, not quite the same.

Suppose I own a machine that can marginally meet the deadlines of whatever class of WUs we're talking about. Suppose I'm creating demand and I'm essentially slowing down a project because there may be plenty of faster machines that are not able to get these WUs because takeing some WUs and taking longer to process them. From a PPD standpoint, I'm doing just what I should be doing. Scientifically, though, the project would benefit if I moved to a lower class of WUs.

From Stanford's perspective, this situation might be improved by requiring me to have more cores, by shortening the deadlines, and probably by several other methods. No matter what method they choose, when I'm forced to choose a lower class, I'm going to get lower points and I'm going to be unhappy. If they are somewhat lower, it's going to be a lot more palatable than if it a radical reduction in points -- but in both cases, I'm probably going to gripe.

There's one more aspect associated with shortening the deadlines, though. I must be able to opt-out so that Stanford will assign WUs with deadlines that are consistent with my hardware.

If I'm on the cusp between standard SMP and bigadv, I can choose to remove the bigadv (once I figure out that exceeding the deadline on 90% of my assignments is costing me more points that removing the -bigadv flag). The same has been true for the slowest of dual-core machines that might be better off running uniprocessor WUs than SMP WUs but most folks have forgotten about that case. Of course if I've used one of the tricks to report an incorrect number of cores, I can also become an honest man and un-hack my machine.

Note that there is a limit to how much deadlines can be lowered. Let's suppose that deadlines for Big-bigadv are shortened. If I have a real 12-core machine that can't meet the new deadlines, I cannot opt out because there's no -bigadv12 flag that I can remove. This would cause real damage to both my attitude and to the science. The same thing happened recently with some uniprocessor projects, but I won't go into that here. The problem is understood and I'm sure they'll find a way to fix it.
Racer43
Posts: 35
Joined: Tue Jan 05, 2010 1:31 pm

Re: Bigadv points change

Post by Racer43 »

To Grandpa_01: Amen. I coach high school football (linemen). How often do then get the joy and glory of scoring the touchdown, but if their contribution had not been felt, would a touchdown have been scored in the first place?

To Jester: Amen, again. I am in the same boat. Although I walked away a while ago, I couldn't get this out of my blood, and here I am again. What if it is my grandchild who is saved by the work we do today? And it is another father's son? At least we were able to help keep that devastating breach in another father's heart from opening.

As has been said before, "Fold On".
Team 163828, Always Broke Folding :mrgreen:
Folding towards the Top 1000 teams with: a Phenom II X2 laptop and an Athlon 64 X2 5000+ with a Sapphire HD 6570 :shock:
Jester
Posts: 102
Joined: Sun Mar 30, 2008 1:03 pm

Re: Bigadv points change

Post by Jester »

Another possibility that's often overlooked is something "discovered by accident",
how many times in history have all sorts of breakthrough's been made while researching something else,
When we simply look, who knows what we may find. :ewink:
bess43
Posts: 2
Joined: Thu Jun 30, 2011 6:46 am

Re: point system is getting ridiculous...

Post by bess43 »

There is another problem with zeroing out stuff every year besides just the points, and I haven't seen it mentioned much in this thread. However there are folks out there with not so many points, but they have run a huge number of work units which in itself is just as much of a badge to people as the actual "points". The # of work units crunched also attests to the devotion and efforts of a great number of contributors/volunteers as it shows they have accomplished a lot of the grunt work that some of the big points crowd don't fully appreciate the value of. The point being there is much more to this project than just points per se. No shame at all in not having many points but having crunched a lot of work units as that takes a lot of time. :-) Just sayin.... :-)
the animal
Posts: 41
Joined: Mon Feb 25, 2008 5:04 pm
Location: TeAm Anandtech

Re: Bigadv points change

Post by the animal »

RoomateoYo wrote:I just noticed on the Project Summary that the bigadv base point values have gone up.
I'm not seeing any change. Link?
noorman
Posts: 270
Joined: Sun Dec 02, 2007 2:26 pm
Hardware configuration: Folders: Intel C2D E6550 @ 3.150 GHz + GPU XFX 9800GTX+ @ 765 MHZ w. WinXP-GPU
AMD A2X64 3800+ @ stock + GPU XFX 9800GTX+ @ 775 MHZ w. WinXP-GPU
Main rig: an old Athlon Barton 2500+ @2.25 GHz & 2* 512 MB RAM Apacer, Radeon 9800Pro, WinXP SP3+
Location: Belgium, near the International Sea-Port of Antwerp

Re: point system is getting ridiculous...

Post by noorman »

.
Very true; in the Team I was part of for nearly 6 years (and in which I started Folding) we always pointed that out; the actual science is measured in finished Wu's, not in the credits awarded to them!
.
- stopped Linux SMP w. HT on i7-860@3.5 GHz
....................................
Folded since 10-06-04 till 09-2010
Post Reply