Donors Advisory Board information and discussion

[Xilikon]

Now that I got the DAB member position after Tobit decided to retire from the DAB and after discussing with Vijay about what must be done to create the DAB, I'm making this thread to explain what is the exact purpose of the DAB and to answer any questions. Please don't turn this in a nice woodlog where we flame each other like a bunch of drama queen bitches (blame Marty9876 for that )...

First of all, what is a Donors Advisory Board ? It's a board grouping the Pande Group, Bruce (representing the smaller teams without representative) and the representative of the top 5 teams. The DAB doesn't take decisions and doesn't vote on proposal so forget it if you expect it to override the PG choices. However, the representatives can discuss about the various "hot" topics of today and tomorrow with the PG to find a good way to resolve at the satisfaction of the general folding crowd. When a solution is agreed between the PG and the DAB, Vijay will then do the necessary to implement the solution if the solution is beneficial for the whole project form the participation and the science view. The DAB will also participate by explaining the decisions to their respective teams and to answer questions/concerns (sometime, we can use that to polish a solution which is agreed in the DAB and to remove any potential issue).

Let's use the example of the benchmarking issue which is a longtime hot topic : The PG always use a reference machine to determine the value of each workunit but that will bring out lots of flaws when we see the same workunit getting a big ppd swing among a broader range of hardware even if the hardware is very close to the reference benchmark machine. The DAB could propose to collaborate in the validation of the points value by running the same units on a few differents machines then propose a better points value which reduce the discrepancy and represent the value better. Remeber this is a example and the solution might not be that but that's the goal of the DAB to find out with the PG what is the best solution which can satisfy the PG (science production output) and the general folding crowd (points output) at the same time.

I know some people ask what is the level of the DAB in the folding world. For me, the DAB is a "political" board who use diplomacy to find the best solutions because there is no perfect solution and we must satisfy the needs of the folders while satisfying the needs of the PG at the same time even with a opposed view on the project. It's not a easy task to find the best compromise and to explain it to everyone so it take a good levelheaded set of people to be able to do this. However, the DAB will have a lot of weight when we need to make decisions that the PG cannot ignore. The PG will benefit from that weight in the sense they can get more connected to the needs of folders instead of being accused of ignoring them.

I hope this will help clarify the goals of the DAB and if you have legitimate questions, feel free to ask and I'll answer to the best of my knowledge. If I'm unsure, I will check with Vijay and others DAB members to provide a correct and honest answer.

 

[osrk]

Now that I got the DAB member position after Tobit decided to retire from the DAB and after discussing with Vijay about what must be done to create the DAB, I'm making this thread to explain what is the exact purpose of the DAB and to answer any questions. Please don't turn this in a nice woodlog where we flame each other like a bunch of drama queen bitches (blame Marty9876 for that )...
.

Kyle should make a separate forum where people can flame the shit out of each other.

Also the point system needs to be redone. I feel this is main because they use such a mainstream box.

 

[jebo_4jc]

sounds like you're off to a good start Xilikon

The benchmark machine stuff is probably #1 on my list. I think that would iron out a lot of the points issues.

#2 for me would be a more "regular joe" way for us to be informed about the specific scientific progress our contributions are making. Like specifically "Project 6701 contributed to progress in the area of huntington's disease" or the like.

 

[Agent_N]

Repost from other post started by Tobit:

"I would like to see Pande Group put some rules in place so that we know where the boundaries are to play within. There are several gray areas (fermi flag on g92 hardware; bigadv on X6's and i7's, etc.) that need to be addressed. I realize that the official answer to bigadv on X6's is discouraging the practice of, though it can technically meet the deadline, but when such experiments are borderline detrimental to the goal of F@H I feel that PG needs to take an official stance of yes or no. If they are ok with it, fine, I'll be ok with it since it is their project and they have apparently weighed the risks before stating their position. However, I don't see a problem with teams experimenting to see of X or Y works and determining what rule to put in place for the official stance regarding that hardware experiment and some reasons why or why not, even if it can be technically done. The idea is for the bettering of scientific understanding and some "unfair" rules would have to be put in place. If you don't like it, buy better hardware! I'd like to run bigadv, but I don't have money to throw at that, so I keep plugging along with a 435 X3, but that is starting to concern me with some of the newer SMP units that I can't finish with 80% time remaining.

I see a lot of ambiguity and non committed responses; that doesn't bode well for telling members of one's team that X is not allowed for specific reasons. PG is running the F@H project and if we don't like a rule, too bad. If it were your project, then by all means, do what you want with it, otherwise shut up and pedal with what hardware you have while following what PG has decided regarding gray area issues.

I am venting a bit because these issues have caused some strife within the team and it has been exacerbated with the recent, imminent takeover, of the #1 spot by EVGA. It seems that team vs. team and team member competition has led to many of the issues that were never a problem when everyone was folding anonymously.

Edit: The 980X and it's Xeon derivatives seem to handle bigadv just fine even though they don't have 8 physical cores. This is another reason I feel that teams should be allowed to experiment, but under supervision and with the intent to of reporting to PG the results so that they can determine yays and nays. Some have commented that certain CPU's with 8 physical cores have difficulty meeting bigadv deadlines. The 8 core restriction seems to be less than ideal for determining what hardware is eligible what what hardware is not eligible."

 

[may i be worthy]

Congrats Xilikon, nobody doubts you will do a great job.

I have mentioned my thoughts plenty in the previous threads, suffice to say the most important thing is to make the concept of the board work and be a success, keep it flexible and the rules as simple as possible.

I am a believer in letting things stay organic and flexible - get in and solve problems, rather than design a bureaucracy that burns time and effort - if it is not working, there is no shortage of oversight from the teams (the "Five Families" to borrow from the Godfather) to kick up a fuss.

Communication, points, benchmarking, "marketing" the project to the donors, Project descriptions, early and open betatesting or roadmaps to help donors investment wisely in hardware - these are the things to start with.

 

[brycejones]

No drama? Dang it the fun police have shown up.

I guess my biggest concern would really be around communicating better the value of the results we are returning. How are our not insignifcant monetary and time contributions moving the research ball forward?

 

[osrk]

No drama? Dang it the fun police have shown up.

I guess my biggest concern would really be around communicating better the value of the results we are returning. How are our not insignifcant monetary and time contributions moving the research ball forward?

we're doing research?! I thought it was a long term PC burn in tool

 

[sirmonkey1985]

we're doing research?! I thought it was a long term PC burn in tool

didnt they tell you. these burn in tools are a secret government project to help create a super weapon.

 

[APOLLO]

However, I don't see a problem with teams experimenting to see of X or Y works and determining what rule to put in place for the official stance regarding that hardware experiment and some reasons why or why not, even if it can be technically done.

The problem with experimentation is when it is taken too far or when it approaches violating the EULA. It's not always clear cut when it's OK and when it's not. Better to err on the side of discretion. Don't get me wrong, I'm all for experimenting where it yields positive results for the research, and nothing is tampered with including circumventing project infrastructure for personal benefit. I deployed multiple VMs when we were limited to two cores per VM, despite PG's frowning at everyone for doing that. I switched over as soon as I could utilize more than two cores. Would I attempt something similar today? Almost certainly not, because I now know how the research functions better than I did before. Back then, I was under the false impression that volume is superior to speed. With time comes wisdom.

The idea is for the bettering of scientific understanding and some "unfair" rules would have to be put in place. If you don't like it, buy better hardware!

Yes, that or simply drop folding altogether. There are plenty of other DC projects that can use the processing power.

I am venting a bit because these issues have caused some strife within the team and it has been exacerbated with the recent, imminent takeover, of the #1 spot by EVGA. It seems that team vs. team and team member competition has led to many of the issues that were never a problem when everyone was folding anonymously.

I don't even remember a time when F@H didn't have teams but I started folding in Q1/Q2 '01.

On a serious note, a lot of our most contentious threads and topics occurred around the EVGA overtake and even more recently. It's not so much the competition itself as it is a fundamental philosophical rift that's dividing this team and the community at large. We find ourselves in a year when there's a coalescence of many new developments, changes and emergent issues hitherto not seen that have all conspired to make 2010 one of the most disputatious and controversial years in the history of F@H. It is the combination of so many factors - new clients, new types of WUs, new hardware not to mention all the controversies involving participatory conduct, intra/inter-team conflicts and the quasi-hacking/loopholes, etc., it's no freaking wonder we have so many flare-ups and widespread discontent...

Edit: The 980X and it's Xeon derivatives seem to handle bigadv just fine even though they don't have 8 physical cores. This is another reason I feel that teams should be allowed to experiment, but under supervision and with the intent to of reporting to PG the results so that they can determine yays and nays. Some have commented that certain CPU's with 8 physical cores have difficulty meeting bigadv deadlines. The 8 core restriction seems to be less than ideal for determining what hardware is eligible what what hardware is not eligible."

Back when -bigadv was in limited use (prior public release), the 8 core minimum was established because workstation and server class hardware at that time increasingly employed dual quad-core processors whereas consumer-level systems topped at four cores. It was intended as a rough guideline to distinguish hardware that was capable of completing -bigadv within an appreciable time frame. I say 'rough' guideline because even then dual quads under a 2.3-2.4GHz threshold could not complete -bigadv. Sure, some single processor Penryns clocked @4.5Ghz+ could possibly have completed these WUs within the preferred deadline, but there were risks (as there are presently) and 99% of systems with less than 8 cores back then simply could not meet the minimum performance requirements. It was a good decision - for that time.

Fast forward to the end of 2010, start of 2011 and we are thrust into a new era. The 8 core minimum requirement is now antiquated. What PG needs to do is establish a new minimum guideline for -bigadv without specifying core requirements, because it is quickly becoming meaningless with every new processor that is released by either camp. This can easily be dealt with by simply adjusting the preferred deadline or introduce another level of SMP WUs for the very fastest hardware, which is what I suggested a couple of months earlier, providing there is sufficient work to warrant it. For now, I'd recommend everyone heed project advisories and wait for a resolution on this and other matters.

In regards to comments concerning the new board's 'tardy' dealings with amendments to rules and recommendations, and counseling the general concerns of the folding community: this is holiday season, and patience is the order of the day. We will likely need to wait at least a month if not longer, so let's all relax, have a cool one and see what comes following the new year.

 

[Xilikon]

In regards to comments concerning the new board's 'tardy' dealings with amendments to rules and recommendations, and counseling the general concerns of the folding community: this is holiday season, and patience is the order of the day. We will likely need to wait at least a month if not longer, so let's all relax, have a cool one and see what comes following the new year.

Yes, it's the holidays for most of us with the shopping spree, workplace parties and other related activities. However, be assured that Vijay is currently working hard to setup a place for us to meet and right now, we are getting acquainted together. Since it's not a live meeting, it's not easy to do this quickly.

By january 2011, we should see the first results and a formal announcement by Vijay.

 

[Xilikon]

The DAB now have a sub-section to discuss so I'm bumping this to gather all the comments/suggestions/complaints from the [H]orde that I could bring to the DAB. MaximumPC already have made a thread with a list of items to address so get your voice heard !

 

[may i be worthy]

This is the private DAB forum on FF right? Are you going to summarise the many opinions expressed so far, or are you going to put up a draft for [H] to discuss/vote on? (or is it too early for that?)

A lot of the problems are clear and don't need back and forth, but consensus on points issues might be hard to achieve, given that there will be winners and losers...

It is funny, I have written a lot and I have a million things I want to see discussed, but at this point I am just happy the DAB concept is launching, and I doubt the first few months will be short of material to discuss.

 

[Xilikon]

This is the private DAB forum on FF right? Are you going to summarise the many opinions expressed so far, or are you going to put up a draft for [H] to discuss/vote on? (or is it too early for that?)

A lot of the problems are clear and don't need back and forth, but consensus on points issues might be hard to achieve, given that there will be winners and losers...

It is funny, I have written a lot and I have a million things I want to see discussed, but at this point I am just happy the DAB concept is launching, and I doubt the first few months will be short of material to discuss.

Yes, a private forum on the FF.

I wanted to collect the opinions/comments/suggestions in a single place (here) then bring a summary to the DAB. We can discuss each point to agree with a answer then I come back with answers.

Right now, we are discussing a lot about the points consistency and the benchmarking issues. We are making progress and we acknowledged that there are problems which need to be fixed.

 

[gkanakis]

Right now, we are discussing a lot about the points consistency and the benchmarking issues. We are making progress and we acknowledged that there are problems which need to be fixed.

Just seeing how things are going.

 

[jebo_4jc]

Just seeing how things are going.

the GPU point system has already been improved

 

[may i be worthy]

Today my 3 rigs are choking on nothing but 2684s and SMP, thus at 57% capacity. Which reminds me, how is that DAB process going?

Does it take that long to sit PG down and say, guys, just multiply 2684's points by 1.4x and 6701s by 1.66x.

Put feet up, pour a drink, bask in gratitude of a grateful folding nation.

 

[MrValentine]

Does it take that long to sit PG down and....

This gave me a good laugh!!!!

/poke fun at PG

 

[APOLLO]

the GPU point system has already been improved

In what way? You don't mean anything very recent because nothing is different from my perspective and I've been folding GPU since GPU1.

 

[sirmonkey1985]

In what way? You don't mean anything very recent because nothing is different from my perspective and I've been folding GPU since GPU1.

gpu2 hasn't changed and wont ever change for the foreseeable future since it will eventually be EOL'd. but they fixed the retarded point's in the gpu3 client with the fermi's so there won't be anymore GTS 450's making 18k PPD since they are using the gtx 460 as the benchmark card now. the point system in the gpu2 client has always been flawed due to them using an ATI 3870 as the benchmark card but theres nothing we can do about that now otherwise it would cause to many issues. for the gpu3 non fermi WU's i'm not sure what card they are using now but my guess is its a GTX 260 based on the high point values and the fact that the G92's get retarded high points on them.

 

[APOLLO]

gpu2 hasn't changed and wont ever change for the foreseeable future since it will eventually be EOL'd. but they fixed the retarded point's in the gpu3 client with the fermi's so there won't be anymore GTS 450's making 18k PPD since they are using the gtx 460 as the benchmark card now. the point system in the gpu2 client has always been flawed due to them using an ATI 3870 as the benchmark card but theres nothing we can do about that now otherwise it would cause to many issues. for the gpu3 non fermi WU's i'm not sure what card they are using now but my guess is its a GTX 260 based on the high point values and the fact that the G92's get retarded high points on them.

OK, I remember that was back in late fall when the new WUs for Fermi were released. I thought perhaps something else was being referred to. My mistake.

 

[sirmonkey1985]

no problem. thats about the only thing PG's done so far. theres real time.. then theres Pande Group time... dun dun dun.......

 

[may i be worthy]

Just seeing how things are going.

Yes, BUMP.

 

[may i be worthy]

Yes, it's the holidays for most of us with the shopping spree, workplace parties and other related activities. However, be assured that Vijay is currently working hard to setup a place for us to meet and right now, we are getting acquainted together. Since it's not a live meeting, it's not easy to do this quickly.

By january 2011, we should see the first results and a formal announcement by Vijay.

And feedback on this process would be good feedback....

 

[Xilikon]

Right now, there is no word since december 23th. I'll have to poke them tomorrow and see if people still want to do this.

 

[jebo_4jc]

Thanks Xilikon. Keep us posted.

 

[may i be worthy]

Right now, there is no word since december 23th. I'll have to poke them tomorrow and see if people still want to do this.

The strong impression I get is that this is not a priority for PG. Which is why we as teams need to push them along.

Can we get any idea of what is going on - is there work going on to set up the board, or are they already tackling issues raised? From the outside it is hard to tell if we are a week away from a spate of awesome announcements, DAB up and running, points system revamps, and general spring cleaning... or absolutely nothing has happened. It certainly looks like the latter.

^ this question is exactly the sort of thing setting up the DAB is supposed to fix.

 

[Axdrenalin]

<sigh> I tend to agree with MIBW here. Looks like more of the same from the PG and Stanford. Not exactly what I hoped to see come out of these suggestions that we worked up earlier....

....You know, my electric bill needs to be a liitle more favorable, and I need to start coming up with some ways to save money for a family vacation this Summer. I might have to part with some hardware to fund my trip. I think its about time my priorities shifted more towards my family instead of spending money on stuff like this. I'll always contribute, but I just can't help feeling that my excitement about the F@H project isn't really shared by the people who we're crunching all these WU's for.

Kind of a shame the enthusiasm for the project isn't as widespread as the things we're fighting against...

 

[sirmonkey1985]

well the day they decided to make it a private sub forum i had a feeling it was going to lead to no where. it should be private to post on but the information should be open to the public to read. so i honestly gave up on the whole DAB idea when i saw that. this was PG's attempt to act like they really gave a crap about the community. i'm sure they figured it would get us off their backs for a few months.

 

[jebo_4jc]

The lack of progress on this is really discouraging.

Also, it appears our "best practices guideline" went totally forgotten.

 

[APOLLO]

<sigh> I tend to agree with MIBW here. Looks like more of the same from the PG and Stanford. Not exactly what I hoped to see come out of these suggestions that we worked up earlier....

Kind of a shame the enthusiasm for the project isn't as widespread as the things we're fighting against...

They also have a team that in effect 'purchases' folders for them. The rest of us aren't as important as we once were. When you have a corporate sponsored team that furnishes a continuing supply of folders to the tune of ~19M PPD and growing every year, there is less and less motivation to listen to gripes when a large pool of folders is virtually guaranteed resulting in a diminished concern by proportion for the eruption of community-wide dissent about any one issue.

 

[jebo_4jc]

Is it time to consider more dramatic measures to get their attention?

Like a team wide 48 hour strike or similar?

My power bill could use a break.
Posted via Mobile Device

 

[mashtub651]

or use that time for a WCG project

 

[may i be worthy]

Is it time to consider more dramatic measures to get their attention?

Like a team wide 48 hour strike or similar?

My power bill could use a break.
Posted via Mobile Device

I think that time could come - but right now I would like the elected reps to be actively hassling PG first. Squeaky wheel etc etc...

They also have a team that in effect 'purchases' folders for them. The rest of us aren't as important as we once were. When you have a corporate sponsored team that furnishes a continuing supply of folders to the tune of ~19M PPD and growing every year, there is less and less motivation to listen to gripes when a large pool of folders is virtually guaranteed resulting in a diminished concern by proportion for the eruption of community-wide dissent about any one issue.

I am not so sure about EVGA being the problem, but I totally agree that Moores Law / natural growth in computing in general has lulled PG into thinking Donors are just something that happens without much effort on their part... which is why if this DAB process fails, and a bunch of us shut down they might notice that the hardcore team folders are a lot bigger slice of the pie than they thought.

The sad thing is that we are only asking for less than 1% of their time. Which would pay for itself. But I repeat myself. This is marketing/sales/customer relations 101 to private business. Sigh.

 

[jebo_4jc]

I totally agree that Moores Law / natural growth in computing in general has lulled PG into thinking Donors are just something that happens without much effort on their part

you probably nailed it
Posted via Mobile Device

 

[APOLLO]

I am not so sure about EVGA being the problem, but I totally agree that Moores Law / natural growth in computing in general has lulled PG into thinking Donors are just something that happens without much effort on their part... which is why if this DAB process fails, and a bunch of us shut down they might notice that the hardcore team folders are a lot bigger slice of the pie than they thought..

You're right, EVGA isn't the problem, apathy is. We as a team can disappear off the map and I still don't think it would change things much. I honestly hope to be proven wrong some day. My cynicism is born from years of seeing this sort of thing occur time and again.

To expound on the EVGA part: imagine operating a business where say 25% of your customers were guaranteed no matter what happened to the market, economy or the quality of your product. Would you be as concerned or motivated about catering to every customer's needs and grievances than if your entire client base was contingent on continued timely rapport with them? Although I do not believe EVGA is responsible for the lack of feedback thus far, I do think that a huge surplus in the volume of participation in the past year 'could' lead to greater apathy out of sheer complacency. A common problem with success.

I think you and I have used the business analogy for this DC project in the past, and believe it is the best analogy. I am very skeptical about the whole charity model some propound because it doesn't accurately represent all the kinetics involved in this project (from what I've seen).

 

[sirmonkey1985]

if we were all to quit we alone wouldn't make that big of a dent. but it would create a domino effect, people see us quit and eventually they start trying to figure out why we all quit. then it just trickles down from there and eventually your left with a bunch of people running borg's they no longer have access to. also EVGA/Nvidia will have no reason to support F@H anymore since theres no incentive for them since it just becomes another DC project in the long list of other projects no one cares about anymore. it would take a while but that would be the 6-8 month domino effect death for F@H. its happened before with other projects where the number 1 team quit after getting fed up with the people running the project accusing them of cheating(which they weren't) and the project was completely dead within 2-3 months.

the odd's of it happening? very slim, but it could happen.

 

[Patriot]

I stopped folding gpu2 because it wasn't worth the power usage on my ati cards... for the work they do they get half credit for atom count on top of a terrible client...

f@h gpu another twimtbp program...

now milkyway... they know how to use ati cards...

 

[sirmonkey1985]

I stopped folding gpu2 because it wasn't worth the power usage on my ati cards... for the work they do they get half credit for atom count on top of a terrible client...

f@h gpu another twimtbp program...

now milkyway... they know how to use ati cards...

we can only hope that changes. quite frankly ATI/AMD cards should destroy nvidia if the clients done correctly and they don't manipulate the points so that ATI/AMD cards don't out perform Nvidia.

 

[may i be worthy]

Also posted to the EVGA DAB forum:

DAB had to be suggested by teams, and was not initiated by PG. The whole point is that PG have not been communicating or meeting the needs of the donors who pay their power bills. Naturally it follows that the teams representatives will have to drive this process forward.

Can I respectfully suggest that the elected team reps then consult with each other and light this candle, so to speak.

 

[jebo_4jc]

not going to let this fall off the front page