Change in BA requirements

[kerryd]

I should also point out that one could equally conclude that BA work units are overvalued rather than that SMP are undervalued. A number of people have made this point, and it is not clear (and not entirely in my hands) which direction the points scheme will move. I would personally be in favor of a larger realignment, but that is of course complicated.

Thank you for responding.

I would say that as far as the value of BA's go it is really up to you and the research that you are doing as to their real value, not other donors. Though the BA donors are not a large segment of F@H would you really want to alienate them more by cutting BA points. They have spent allot of money in hardware and electrical cost to help in your research.

And I would ask the question of the people that have made the point that BA's are overvalued; do they actually run BA capable hardware? Remember that we have also run non-BA hardware in the past and the present.

I was running 2 BA servers but no longer I turned them off now they just set in there rack .

[7im]

The last post I made regarding the number of active machines currently working FAH was 226,224.

Today the number is 224,908. That's another 1316 machines offline.

The attrition rate is something to be worried about.

The crash continues.

No, it doesn't.

There's a problem with core_17 not reporting to the stats correctly for AMD GPUs. viewtopic.php?p=255115#p255115 All those new AMD R2 2xx series cards aren't showing up. The stats are easily a few thousand clients short.

And a lack of core_17 work units: http://folding.stanford.edu/home/more-c ... n-the-way/ So even the NV GPUs are potentially not reporting.

[ChristianVirtual]

PG need to make a decision with BA, Either they allow 2p machines to contribute or they don't. The current plans would indicate that they don't want them.

Share this impression and ask myself: why ?

Each BA-WU done in time on a 2P with some nice new E5s is a WU done in time. Why PG would need to care how a donor makes the WU. Define the deadline and be happy to get the results in time. 2P or 4P, shouldn't matter. And if you really want to exclude those new 8c/16T desktop CPU might come later the year just set the minimum core to 17. Exclude them if you see the risk they don't make the deadline. But keep as many as possible "qualified" hardware and engaged donors in the game.

[Grandpa_01]

We all know that most donors do not run bigadv below is a few reasons why.
#1 It is expensive to own and operate them. The requirements for bigadv folding is the only form of folding that requires frequent upgrades to equipment in order to fold them.
#2 Everybody wants to be completive in the points department and in order to keep up in the bigadv world you have to upgrade continually, (see #1)
#3 Every time there is public outcry big advanced takes a pay cut. (see #2)
#4 Most donors are not willing to build a machine they have no other use for. (see #1)

Bigadv is not over valued IMHO SMP is currently under valued though, if there is a larger demand for equipment than there is equipment to run it or people and not willing to run it then the value is not there. Why would anybody run SMP, on my 12 core rigs I turned smp off because I could drop the clocks down to where I use less than 75 watts for the CPU and run 3 - GPU's and make 270k PPD on under 500 watts why would I run SMP for another 8K points for 100 more Watts of power.
Equal pay for equal science does not work when you have more work to be done in a format that has such a lesser scientific value (SMP). Some day things may change in that department but today that is the way it stands.
There needs to be a road map in some shape or form that lets a person know if they are going to chase bigadv points that this is what is going to be required on on this date and on that date it will most likely be moved. Then you are fully aware your investment is only going to be good for X amount of time.
The world is forever moving forward we either move forward with it or we get left behind. It would be nice to have a map of what lies in front of us.

[Viper97]

7im

You can dispute the numbers... you can in fact rationalize as ye see fit. Truth told less than a month ago there were 240K machines folding. Today... far less. Enjoy.

[Viper97]

The world is forever moving forward we either move forward with it or we get left behind. It would be nice to have a map of what lies in front of us.

We have a map... it does not include us. The next move is to the cloud for PG... and I think that's good.

[Adak]

Since BA points account for things other than a simple benchmark, like more memory, higher core counts, much larger work units, and so higher bandwidth usage, faster internet connections, etc., why should SMP ever line up with BA when SMP only needs a few cores and a GB of memory to run?

The best reason is that it removes the feeling of being pushed off the cliff, by those who will be affected by the increase in the BA threshold. You may say that it's an emotional response, but I would say it's a response that will generate large negative consequences for FAH, if this problem isn't fixed.

@GrandPa:

BA points ARE inflated. Compare an overclocked 2600k PC, folding SMP, with a 4 x 6272 at stock speed:

2600k: 8 cores, x 4.5GHz (36) ~ 25,000 SMP ppd.

4P: 64 cores x 2.1GHz (134.4) ~ 330,000 BA ppd

Let's be generous, and say that an AMD cpu cycle, is equal to the Intel cycle (it is less than that, but let's be generous).
So the relative performance factor is 134.4 / 36 = 3.73. (373%)

And let's give BA work units a 30% boost because they use extra bandwidth, extra memory, more power, etc., and extrapolate it

(SMP ppd * perfmance factor) * 130% boost equals extrapolated points
(25,000 x 3.73) x 1.30 (the boost) = 121,225

And THAT (330,000 / 121,225) shows us that BA is currently 272% higher than SMP folding.

That's why i say BA folder ppd, are currently slightly too high, and SMP points are currently much too low.

[Bill1024]

The 6272 run at 2.400ghz with turbo boost so your formula is off.

The 6272 were 525$ x4 + 650$ for the MB + 200 for memory and another 100 for PSU + 80$ for heatsinks. Over 3000$ for the system.
The 2600K system was 325$ +150 + 60 + 20 +100 is around 660$ or so.

Why invest that much money for a 30% boost.

[Adak]

The 6272 run at 2.400ghz with turbo boost so your formula is off.

I wasn't sure if FAH ran at 2.1 in "high performance" mode in the OS and BIOS, or if it ran in 2.4 Turbo Mode. The conclusion is still valid, however. BA points are a bit too high, and SMP points are very much too low.

The 6272 were 525$ x4 + 650$ for the MB + 200 for memory and another 100 for PSU + 80$ for heatsinks. Over 3000$ for the system.
The 2600K system was 325$ +150 + 60 + 20 +100 is around 660$ or so.

Why invest that much money for a 30% boost.

I'm here for better treatments and cures, and understanding of critical protein structure and movement. I very much like having points, but I'm not here because of the points.

But if you want to talk points, let's do that:

Because the points boost is cumulative - we're talking about 30% PER DAY, EVERY SINGLE DAY YOU FOLD. That quickly amounts to hundreds of thousands and then millions of points.

The 5% reduction in BA points in this example, only amounts to 16,500 points less per day. That's still 313,500 ppd - and that's still a very satisfying amount of points per day.

[orion]

I went through the log on one of my 4p, 32 threads @ 2.47GHz

8566~134,000ppd
8819~130,000ppd
8101~216,000ppd
8103~353,000ppd
8104~337,000ppd
8105~340,000ppd

It looks to me that SMP's and 8101's are undervalued while the rest of the BA's are closer to where they should be due to out of pocket expense, longer trajectories, memory usage and researchers priority.

[Grandpa_01]

Adak

Do you have to upgrade that 2600k every other year to keep folding SMP, Once you have it, it is good for years of folding, I cost me between 17k and 20k every year to build upgrade and run my MP rigs. I can tell you that if all the bigadv WU's paid the same as 8101 I most likely would not upgrade again there are allot of alternatives that look better to me. Bigadv has been devalued many times how many more devaluations can it take before it becomes completely unattractive.
There is allot more that goes into that formula of yours than you have put into it.

[Spazturtle]

Can you just tell me in plain and simply english when you will increase the core count from 32?

My 2P 32 core rig is folding happily, when will you prevent it from folding bigadv?

[Bill1024]

Can you just tell me in plain and simply english when you will increase the core count from 32?

My 2P 32 core rig is folding happily, when will you prevent it from folding bigadv?

Depending on the speed of your 32 core, it may be soon.
They are going to cut the QRB return time, so you may not make the deadline.
16 cores are the min right now, and some 16 core AMD servers will not make deadline. Only the fastest 16 core AMD make deadline.
Keep your fingers crossed!!!!!!

[kerryd]

I went through the log on one of my 4p, 32 threads @ 2.47GHz

8566~134,000ppd
8819~130,000ppd
8101~216,000ppd
8103~353,000ppd
8104~337,000ppd
8105~340,000ppd

It looks to me that SMP's and 8101's are undervalued while the rest of the BA's are closer to where they should be due to out of pocket expense, longer trajectories, memory usage and researchers priority.

389,392 ppd now to get that I had to run 48 cores on smp , 2 gtx 770 hydros and 2 gtx 560 ti's and that is a good days ppd most days its a lot lower. More like 250ppd to 275 ppd
I should point out that 200ppd of that is just the gtx770 hydros + 2 cores and 40kppd on the gtx 560's. So it takes 48 cpu cores just to get 160ppd on smp go figure.

[Adak]

Adak

Do you have to upgrade that 2600k every other year to keep folding SMP, Once you have it, it is good for years of folding, I cost me between 17k and 20k every year to build upgrade and run my MP rigs. I can tell you that if all the bigadv WU's paid the same as 8101 I most likely would not upgrade again there are allot of alternatives that look better to me. Bigadv has been devalued many times how many more devaluations can it take before it becomes completely unattractive.
There is allot more that goes into that formula of yours than you have put into it.

I built my 4P rig three years ago. I've never changed a thing since then, so my upgrade costs have been $0.00. I don't have that perspective of "my contribution is only worthwhile if the points are high enough".

What I know:

1) There is NO BOINC project that will give you nearly as many points for your BA rigs, as FAH does. Check it out!

2) There is a HUGE disparity between SMP points and BA points. That gap needs to be closed, somewhat. Most of it should be closed by an increase in SMP points.

3) If every BA work unit is worth 5% less, than EVERYONE in the top ranking of FAH will be affected, so relatively, there will be very little effect.

4) Everyone knows BA wu's were massively over-inflated, when they first came out. Frankly, I had to double check it, because it was hard to believe. Bringing them back down has helped restore more sanity to the points system inflation spiral. BA folders have not "suffered" from the decreases in the past. We have more BA folders than ever. (Until this change in the threshold was announced, at least).

5) Brilong is going to pass you and become #1 in FAH ranking very soon, in any case. Changing the BA points by 5% won't change that.

Congrats on being #1 in the ranking currently, GrandPa. Quite an accomplishment!

Nevertheless, the above are the facts, and I challenge you to show data or reason, to the contrary.