New BIGADV point system

chaandan

Limp Gawd
Joined
Jun 3, 2011
Messages
256
"Standard points" is what the project would receive for base points if standard SMP. "Old bigadv" is the old bigadv base points (50% bonus). "New bigadv" is the new bigadv base points (20% bonus).

Project Standard points Old bigadv New bigadv Preferred Final k-factor
2684 8529 12790 10235 4 6 26.4
2685 5970 8955 7164 4 6 26.4
2686 5970 8955 7164 4 6 26.4
2689 5970 8955 7164 4 6 26.4
2692 5970 8955 7164 4 6 26.4
6900 5970 8955 7164 4 6 26.4
6901 5970 8955 7164 4 6 26.4
6903 18923 28385 22708 7.2 12 38.05
6904 26284 39426 31541 10.2 17 37.31
 
So, they just poo'd all over the points value? Great, just when we needed it.
 
no they changed the bigadv point system to what it should of been from the very beginning. 50% base value was to high from the start but i don't think they were expecting people to actually dump thousands of dollars into systems just to run bigadv when they originally created it. so this was inevitable.
 
The optimist in me hopes that this is an indication of a serious attempt to correct what has been a long standing complaint and blatant discrepancy in the point system. I do believe however that undercutting our point output for current units is exactly the wrong way to approach this problem and avoid upsetting your most devoted contributors. IMO this should have been handled through a rollout of bonus based payout for all units rather than just a cut of the bigadv.

The pessimist in me wonders how convincing the EVGA business manager's threat to pull out of the project was when they successfully lobbied to remove more than half of the bigadv bonus at the exact time when said bonus was allowing us to pull ahead of them...
 
  • It is a 20% cut in PPD. (on my SR2s)
  • BUT - we still have the new super-bigadv, and more to come. These now have PPD 22% better than when we could get 2 months ago, and now have a 55% premium over the best regular bigadv like 6901. So the average PPD in production might end up the same as before the extra-large bigadv appeared a month ago...
  • It is bad news for [H} as a team - we get hit worse than EVGA. We will see how bad.
  • Is is bad news in terms of showing nothing much has changed at Stanford - consultation? what? We could use some DAB representation at the moment. Calling Tobit.....
  • This is not done yet - they have reduced bigadv, but there are further changes to be made - perhaps GPU might go up more with a QRB system.
 
The optimist in me hopes that this is an indication of a serious attempt to correct what has been a long standing complaint and blatant discrepancy in the point system. I do believe however that undercutting our point output for current units is exactly the wrong way to approach this problem and avoid upsetting your most devoted contributors. IMO this should have been handled through a rollout of bonus based payout for all units rather than just a cut of the bigadv.

The pessimist in me wonders how convincing the EVGA business manager's threat to pull out of the project was when they successfully lobbied to remove more than half of the bigadv bonus at the exact time when said bonus was allowing us to pull ahead of them...


there is no right way to do it. either you go all in and change it or just don't bother doing it at all. either way you piss some one off. its more likely to piss off the people that didn't care about the science from the beginning then it is going to piss off the people that were doing it for the science and point benefit that went along with it.


  • It is a 20% cut in PPD. (on my SR2s)
  • BUT - we still have the new super-bigadv, and more to come. These now have PPD 22% better than when we could get 2 months ago, and now have a 55% premium over the best regular bigadv like 6901. So the average PPD in production might end up the same as before the extra-large bigadv appeared a month ago...
  • It is bad news for [H} as a team - we get hit worse than EVGA. We will see how bad.
  • Is is bad news in terms of showing nothing much has changed at Stanford - consultation? what? We could use some DAB representation at the moment. Calling Tobit.....
  • This is not done yet - they have reduced bigadv, but there are further changes to be made - perhaps GPU might go up more with a QRB system.

i think it will be the exact opposite. i don't see EVGA as a whole that really cares about the science. some one will get pissed and try to boycott F@H and everyone will jump on the bandwagon. we'll know for sure by the end of the week.

oh well back to my silence since i really have nothing invested in any of this nor care. just glad to see it actually happened even if it was 2 years late.
 
Last edited:
The pessimist in me wonders how convincing the EVGA business manager's threat to pull out of the project was when they successfully lobbied to remove more than half of the bigadv bonus at the exact time when said bonus was allowing us to pull ahead of them...

Here's the way I look at it. The timing happened just after we started passing EVGA ppd wise. We know NVIDIA has invested in the GPU client which is one reason why we see NVIDIA's architectures perform so much better than ATI. EVGA is also a very large partner for NVIDIA and also invests in F@H through the EVGA bucks program.

If NVIDIA/ EVGA did exert some sort of influence then there is not much we can do about it. They invest in the project to a far greater extent than even our top 20 does. That is just the way the system works.

Then again, here's something interesting, the 6174 4P is still putting out around 1,000ppd/ watt.The point value change does not really make folding less efficient on Magny Cours, it just biases folding activities away from highly power efficient hardware and towards inefficient hardware. Not a "green" concept, but I can understand why they would want to provide more incentive to new folders. Either way, I have been doing a big consolidation project to keep the ppd/ watt at its highest.
 
It's rather irritating and "interesting" that they would choose now to adjust the point values.

Oh well, I guess I will have to find new ways to get my points up where I want them.

OTH, It hurts me less than some others on our team as half my points come from GPU already.
 
Here is how I'm chosing to look at this.

With the 6903/4s I'm making 30k - 40k ppd more on my SR-2s than I was with the regular -bigadv units be for the adjusment.

We had a gravy train for a month that was quite nice, however it is over.

I'm choosing to stay positive about this.

Currently:
6904: 179,237ppd, total credit of 483,569
6903: 177,355ppd, total credit of 346,910

If you took these points and introduced them a month ago this way, we all would have been happy.
 
Last edited:
I'm choosing to stay positive about this.

I am not - this is the most ridiculous thing I have seen since I have been donating my time and money to this project. I have nothing constructive to say about this, so I will shut up about it...
 
So let me get this strait: They're boosted the PPD for the top guys, left it the same for the bottom lot and just screwed those in the middle.

I have to ask why? It hasn't made the points system more balanced, if anything it's the opposite. To me it just seems like a one-finger salute to those who have upgraded their hardware and spent a bit more to crunch bigAdv (not the mention the extra costs of leaving it going 24/7).
 
Has this already gone into effect? My -bigadv rigs seem to have taken a hit (20K PPD between the 3). Not taking power use in consideration, it makes GPU rigs a little more attractive again.
 
So let me get this strait: They're boosted the PPD for the top guys, left it the same for the bottom lot and just screwed those in the middle.

I have to ask why? It hasn't made the points system more balanced, if anything it's the opposite. To me it just seems like a one-finger salute to those who have upgraded their hardware and spent a bit more to crunch bigAdv (not the mention the extra costs of leaving it going 24/7).

Boosted then dropped for the top guys.

Middle was left alone (SMP and GPU) as was the uni client.

For now that is.
 
[Braveheart] They can take our points, but they can never take our FREEEEDOMMMM [/Braveheart]
 
I am not - this is the most ridiculous thing I have seen since I have been donating my time and money to this project. I have nothing constructive to say about this, so I will shut up about it...

I'm taking this route. People spent big dollars to optimize their production.

Normally I wouldn't care, but I find the timing of this just a tad suspicious. For now, I will just continue to fold but i'm going to mull whether I want to be involved in this in the long term. If this was coming and we know for a fact that some other company didn't use their influence to do this, then i'll admit that this was probably long overdue.
 
I know it's not much compared to what the multi-proc rigs cost, but I spent $300 on a 2600K just so I could fold bigadv. I'm kinda miffed about this.
 
Honestly I think it would have been better if they had done an across the board normalization in a single update (with testing prior to taking effect), but that would probably be more work than they want to do.
 
So here is where its at....

gpus get 62ppd per atom in a wu. (16k gtx 470,avg )
bigadv gets .23 ppd per atom in a wu. (600k my 4p ,6903)

bigadv does more complex calculations with less wu overlap.
Why does its atom count get devalued so.
 
So here is where its at....

gpus get 62ppd per atom in a wu. (16k gtx 470,avg )
bigadv gets .23 ppd per atom in a wu. (600k my 4p ,6903)

bigadv does more complex calculations with less wu overlap.
Why does its atom count get devalued so.

There is also the length of the simulation in simulated time ie. the number of timepoints that have to be calculated.
 
im surprised you guys are bitching, since GPUs get no bonus points and Bigadv was completely lopsided compared to the GPU side of things.
 
im surprised you guys are bitching, since GPUs get no bonus points and Bigadv was completely lopsided compared to the GPU side of things.

Just because they need help in how they calculate points doesn't mean that they are lopsided....

On the gpu the wu are so small the variance in turn in time is small...

With Bigadv the difference can be days...

gpu wu = 260-650 atoms
bigadv wu =1098185-2533797atoms
 
Just because they need help in how they calculate points doesn't mean that they are lopsided....

On the gpu the wu are so small the variance in turn in time is small...

With Bigadv the difference can be days...

gpu wu = 260-650 atoms
bigadv wu =1098185-2533797atoms

yeah, but bigadv has the advantage of multi-cpu setups and bonus points. GPU has none of that, 4 GPUs fold 4 different projects with zero bonus for turning them in early.
 
yeah, but bigadv has the advantage of multi-cpu setups and bonus points. GPU has none of that, 4 GPUs fold 4 different projects with zero bonus for turning them in early.

ok...tell you what... lets give gpu wu an early return bonus and make the overall ppd the same per atom as bigadv...

you can have a normalized 67- 170 ppd but if you turn it in faster you will get more points...

obviously this is not the answer... but the basic point... gpus already get more points than the wu are relatively worth ...
 
Meh, after mulling it over for the day I am more accepting of what they have done.

When Bigadv was launched, 8 thread machines that could fold them were pretty rare, this is no longer the case. As such they have devalued them to something closer to the normal smp bonus.

At the same time, the bleeding edge now has 12 thread units to play with that return more points than the old bigadv did on the same hardware. If we hadn't had the last month of extremely high points, this would seem like a much more reasonable move to raise the bar for getting high points out of bigadv.

What is still extremely wrong is the timing. It coincides way too closely with our rise over EVGA to not have been at least partially influenced by that business relationship.

At the end of the day though, I do think this is a net positive and those of us with 12 thread machines are still earning more ppd than we were a month ago.
 
I know my opinion will be worthless since I fold for EVGA, but I think the conspiracy theories are ridiculous.

This thread is what brought about the change as far as I can see. Vijay posted several times in that thread. His responses appear to be addressing the opinions voiced in that thread, including one where he says that he has scheduled a meeting with the Pande Group to discuss the issue.
 
So from what I can tell, in my 2600k, I'm now going to make about the same ppd regardless of whether I'm folding -smp 8 or -bigadv. At best, bigadv would give a 2k ppd advantage over smp, but at the risk of having a huge WU that takes 2-3 days to chew through that could get lost, whereas multiple SMP units would be completed in a day, making it a much lower risk proposition..
 
So from what I can tell, in my 2600k, I'm now going to make about the same ppd regardless of whether I'm folding -smp 8 or -bigadv. At best, bigadv would give a 2k ppd advantage over smp, but at the risk of having a huge WU that takes 2-3 days to chew through that could get lost, whereas multiple SMP units would be completed in a day, making it a much lower risk proposition..

This sounds about right to me. I don't think bigadv was ever intended to make a huge difference on 4C/8T processors. I wouldn't think the benefit should start becoming drastic until 8C/16T.
 
So from what I can tell, in my 2600k, I'm now going to make about the same ppd regardless of whether I'm folding -smp 8 or -bigadv. At best, bigadv would give a 2k ppd advantage over smp, but at the risk of having a huge WU that takes 2-3 days to chew through that could get lost, whereas multiple SMP units would be completed in a day, making it a much lower risk proposition..

I'm under the same assumption. Going to finish the bigadv I'm on now and switch to regular smp and see what the stats are like.
 
At the end of the day though, I do think this is a net positive and those of us with 12 thread machines are still earning more ppd than we were a month ago.
Hell, us with sr-2 went from estimated up to 100kppd to
120kppd with the release of generous 6900 WUs, to
140kppd with a NUMA fix then
160kppd when optimized linux client was released to
180kppd with tear affinity to
200kppd with 6903 WU release
and then 2684 point structure was fixed..
(rough numbers)

so we do overall get more points ;)
 
Its not the end of the world.... however... the logic behind the point change is lacking in benchmark and logic backing as well as involvement from the donors through DAB and the beta team...THIS is the problem.
 
I read through the first 10 pages or so, but didnt really finish. It is a tough issue, and at first I thought it was terribly convienient, and coincidental with our current battle, but really they have been battling with the points system from the beginning, and will continue to do so as long as the project runs. It will probably just be another bump in the road, and we will see what happens.
 
Well it looks like we laid down 21 million points yesterday! So... dont lose hope. Thats only 12.5% less than our peak at 24mil before the lowered points.
 
Its not the end of the world.... however... the logic behind the point change is lacking in benchmark and logic backing as well as involvement from the donors through DAB and the beta team...THIS is the problem.

Yeah I do feel the project is mismanaged (their communication is worse). It seems that they don't like the influx of 2600k's causing shortages of bigAdv and instead of meeting supply, they do this. Which is fine, if they don't like 2600k's I can put mine to better use elsewhere.

I am going to try some other projects, there's some WCG stuff that I am interested in. This isn't the way to treat donors, so I am voting with my feet.

Sorry guys, but everyday I like PG less and less.
 
The reason for lowering the points on bigadv was pretty simple. People were starting to gripe about the amount of points the bigadv folders were making per WU compared to others. A bigadv folder can make as many points in 1 day as it takes allot of people weeks to months to make.
Some folders were starting to feel left out and not needed any more. If anybody remembers when bigadv started they were designed for machines most people did not have and you had to have laid quite a bit out to have 1. Today 8 core machines are common and really not that big of an investment, you can build a 2600k for $600 to $700 bigadv was never designed for the masses to begin with they were designed for the exceptional hardware.
They have just brought it back into line. The thread that FlipBack pointed to at the FF was open for public comment for quite a while and most of the comments were about bigadv points getting out of control VJ Pandee asked for anybody to come up with a better point system all would be considered he had said that the points were in line with the science but he did not want to keep a point system that everybody hated.

Any way EVGA or any other corporation had nothing to do with it. The folding community was the ones that voiced there opinions.

I do know most of you here at [H] do not know me from Atom but it has affected my PPD just as much as it has allot of yours. I run a farm of 5 - 970"s and 980X rigs But I do believe it was the right thing to do for the science in general.
 
The reason for lowering the points on bigadv was pretty simple. People were starting to gripe about the amount of points the bigadv folders were making per WU compared to others. A bigadv folder can make as many points in 1 day as it takes allot of people weeks to months to make.
Some folders were starting to feel left out and not needed any more. If anybody remembers when bigadv started they were designed for machines most people did not have and you had to have laid quite a bit out to have 1. Today 8 core machines are common and really not that big of an investment, you can build a 2600k for $600 to $700 bigadv was never designed for the masses to begin with they were designed for the exceptional hardware.
They have just brought it back into line. The thread that FlipBack pointed to at the FF was open for public comment for quite a while and most of the comments were about bigadv points getting out of control VJ Pandee asked for anybody to come up with a better point system all would be considered he had said that the points were in line with the science but he did not want to keep a point system that everybody hated.

Any way EVGA or any other corporation had nothing to do with it. The folding community was the ones that voiced there opinions.

I do know most of you here at [H] do not know me from Atom but it has affected my PPD just as much as it has allot of yours. I run a farm of 5 - 970"s and 980X rigs But I do believe it was the right thing to do for the science in general.

your on the beta team I know you...
The gpu folders and smp folders those without much invested voiced opinion...
I am all for shortening the obscenely long return time...
But nuking pt system because of high demand is ludicrous.... a socialist move... should not be that surprised as the decision maker doesn't live or work in the real world.
 
The reason for lowering the points on bigadv was pretty simple. People were starting to gripe about the amount of points the bigadv folders were making per WU compared to others. A bigadv folder can make as many points in 1 day as it takes allot of people weeks to months to make.
Some folders were starting to feel left out and not needed any more. If anybody remembers when bigadv started they were designed for machines most people did not have and you had to have laid quite a bit out to have 1. Today 8 core machines are common and really not that big of an investment, you can build a 2600k for $600 to $700 bigadv was never designed for the masses to begin with they were designed for the exceptional hardware.
They have just brought it back into line. The thread that FlipBack pointed to at the FF was open for public comment for quite a while and most of the comments were about bigadv points getting out of control VJ Pandee asked for anybody to come up with a better point system all would be considered he had said that the points were in line with the science but he did not want to keep a point system that everybody hated.

Any way EVGA or any other corporation had nothing to do with it. The folding community was the ones that voiced there opinions.

I do know most of you here at [H] do not know me from Atom but it has affected my PPD just as much as it has allot of yours. I run a farm of 5 - 970"s and 980X rigs But I do believe it was the right thing to do for the science in general.

To the casual observer (me), this is one of the first moves against the inflating spiral of points for progressively faster hardware. I've been folding since circa 2002-2003 (I don't remember at this point) and for a while had a dozen CPUs running that turned out a whopping thousand points per day in aggregate. Then the PS3s came along that would do 1000 points per day, the same a truckload of boxen that I had running between myself and my friends. Then GPU folding came along, and now SMP and exponential bonus system...

Its somewhat upsetting to me that I spent years accumulating my first million or two points to now there being single systems that can get that many points in a single day. Heck, I did 2.5 million points last month, which represents 25% of the points I've made in total. I have specifically made hardware purchase decisions based on an expected number of points of production and had I known there would have been a change to reduce the point output of particular hardware in the works, I would have chosen differently (i.e. I would have spent an extra couple hundred bucks on a G34 box instead of the 2600k that went into my main rig).

Not sure where my rambling is going, but the point I'm trying to convey is that point accumulation hasn't been "fair" for a long time in my opinion, so the community may want to take the larger point value offerings as a suggestion of where to spend their money to contribute in the most valuable way....
 
Back
Top