Posts by marmot
log in
1) Message boards : Number crunching : Collatz server seems to be ignoring my resource settings (Message 20301)
Posted 1066 days ago by marmot
Although this may be slight thread derailing, After reading this thread and some others -- I don't find Collatz to have "inflated" credit. While this is a high credit/hour compared to some other projects, it also seems to have pretty optimized programs, as well as full or nearly full GPU utilization.

On Collatz, my GPUs are near full utilization. On Einstein and others, they are lucky to be 1/2 even if I change it to .5 GPU per wu. And I am not even going to talk about SETI -- it feels there as if they want to penalize you credit-wise for using a GPU.


I had to stop the GPU processing so I could continue watching videos. I switched to Einstein and SETI GPU WU's because SETI and Einstein GPU WU's will peg out at 100% unless you start using the computer for other GPU applications and then they back off to 70-85% after a few seconds. That is some excellent coding. It's great that Collatz GPU WU peg at 100% but they need to adjust downwards if the machine is sharing the GPU with a user otherwise the user will set BOINC to not run GPU WU while the computer is in use and no work gets accomplished.
Collatz WU's need to share the GPU with a user or not be run.


Boincstats credit comparison matrix at http://boincstats.com/en/stats/-1/cpcs
shows that while Collatz is one of the higher credit programs, it is still only .8015 of GPU Grid. Since I run HD 7770s, 7970s, and 7990s, I can't participate on GPUGrid, but on reading their message boards, they don't seem to have the number of complaints about credit, even though they are granting more credit than Collatz.

As with a lot of other posters, I agree that if you don't like the credits, go elsewhere. But if you are going to complain, go to GPU Grid and complain 20% more than you do here.


I'm not involved with GPU-Grid so I wouldn't know.


Personally, I like getting credit for my time and financial investments. It's not like we can spend these credits anywhere, nor like we get any pecuniary reward of any kind. At least for me, it's purely self-satisfaction, and it allows me to at least monetize my expenses into credits.


The credits are a reward. People love getting rewards and payouts but what work actually gets done and how it helps the world is what's important.

Although this may be slight thread derailing,


Yep, this thread was about WU resource allocation and why I was receiving WU while other projects weren't higher priority. I mentioned that this project was low priority to me and seemed the WU were inflated and you and two other users took umbrage at that side-off comment.

I have degrees in mathematics and physics and I'm no stranger to pursuing esoteric problems for the sake of intellectual curiosity but they take lower priority for me to other pressing problems, like climate change, detecting asteroids on life eliminating, collision courses with earth or getting humans into mining asteroids and becoming a space faring species. That is why I tried so hard to keep Collatz and Prime grid at non-0 resource shares. I really wanted my machines to put some effort into calculating Collatz and finding Primes.

The project is set to 0 resource now and is a last resort to keep the machines heating the room in winter.

@Slicker, is it possible to get Collatz GPU WU to backoff slightly when sharing a GPU with Flash video? I could ask in another thread.
2) Message boards : Number crunching : Collatz server seems to be ignoring my resource settings (Message 20300)
Posted 1066 days ago by marmot
If you are not happy with the way Collatz gives credits then why do you stay there must be a reason. or perhaps could it be the so call inflated credits.


It's a backup project because of when, during Christmas, several projects were down.

Still has nothing to do with why Collatz gave out work when four other projects were online and available for work units.


It is because you don't understand how BOINC works, apparently. The projects don't push work to you, your BOINC client requests specific amounts of work from each project. Don't blame Collatz for how this works, blame BOINC.


We are all born ignorant. Ignorance is a state of not-knowing.
The way to alleviate ignorance is to seek knowledge. Slicker has already pointed out that you can only allot resource allocation from 0 to 1000 where I had broken that rule so my ignorance has been lifted. The client was requesting against resource allocation outside it's parameters.

The later message was about BOINC resource allocation bugginess.

Since Collatz is your highest credited project, I'd expect your emotional response to be one of insulting someone who is rejecting your priorities. That's normal human behavior.
3) Message boards : Number crunching : Collatz server seems to be ignoring my resource settings (Message 20261)
Posted 1074 days ago by marmot
why did you set collatz to 1, and not 0?


It still got some GPU WU's after setting it to 1 so it's at 0 now.

It was a test of the priority system. GPU projects SETI and Einstein are set to 134x and 75x Collatz of setting 1 and Collatz was ahead of Einstein and half SETI's total work, way ahead on RAC, yet still received 18 GPU WU's.

That seems to be a bug in the resource sharing system.
4) Message boards : Number crunching : Collatz server seems to be ignoring my resource settings (Message 20255)
Posted 1077 days ago by marmot
A couple items....
Last I checked, the max allowed for the resource share was 1000 and the minimum 0. While one would think that 0 means don't get work, BOINC assumes it means get work if no other projects can get work. Also, when it does get work, it will fill the queue and not just get 1 WU.


OK, reset the resource allocations among all my projects as shares of a total pie of 1000 with Asteroids@Home set to 750 and the others in smaller shares with Collatz set to 1.

I'll see how things go now but winter is coming to an end and will be shutting down all but two machines in a month so not sure if I'll be able to tell if the problem is corrected.

Thankyou for the clue to the problem, Slicker.


I didn't realize how hot a potato credit inflation was and sorry I even mentioned it. Some here don't agree with my value judgement and we are all entitled to our opinions. We're all donating our real world assets, work and energy to these projects so I respect that and that Mikey, willie and Smelix took the time to respond to my question.

Thanks.
5) Message boards : Number crunching : Collatz server seems to be ignoring my resource settings (Message 20254)
Posted 1077 days ago by marmot
A couple items....
Last I checked, the max allowed for the resource share was 1000 and the minimum 0. While one would think that 0 means don't get work, BOINC assumes it means get work if no other projects can get work. Also, when it does get work, it will fill the queue and not just get 1 WU.


OK, that sounds like an answer to my question. I can't set resource share to numbers like 1,638,400 (power of 2) for the top project and 100 for the lowest. I'll have to scale my resources definitions between 0 and 1000. Setting to 0 for work when all other projects are down sounds like a very agreeable definition of 0 for BOINC.
Thankyou.


Secondly, if you have an 8 core processor and set Collatz at 100 and another project at 300 and another at 400, you would expect it to use 1 core for collatz, 3 cores for project B and 4 cores for project C. Nope. It may use all 8 cores for one project at a time. It may run multiple projects. Eventually, it should work out that 12.5% of the time is Collatz. I've never found it to actually work that way for me so I run one project at a time only, especially since it now counts WUs from suspended projects when determining whether to get more work or not.


This again sounds reasonably like how resources would be divided although not exactly explaining how they are divided over weekly periods. Since credit handout per project differs we can't expect credit to break down exactly along resource share.
The problem was that Collatz was getting work when 4 other projects were online and had higher resource allocation.


As far as credits goes, if the GPU app is 70-100 times faster than the CPU app, shouldn't it get 70-100 times the credit of the CPU app?


And yet on the same GPU, Einstein@Home, which is a generous project for credit according to ratings from 2013, is giving out 1/30th the credit of Collatz.


Maybe I should implement a new policy that if someone complains about too much credit, they get switched over to creditNew while everyone else continues status quo. That way, everyone would be happy.


Matters not to me.
These machines are heating the house (bedroom mostly, but the heat seeping to the hall generally keeps it at 50 degrees) and have to continue working non-stop during the winter.
I can't let them idle.
6) Message boards : Number crunching : Collatz server seems to be ignoring my resource settings (Message 20253)
Posted 1077 days ago by marmot
If you are not happy with the way Collatz gives credits then why do you stay there must be a reason. or perhaps could it be the so call inflated credits.


It's a backup project because of when, during Christmas, several projects were down.

Still has nothing to do with why Collatz gave out work when four other projects were online and available for work units.
7) Message boards : Number crunching : Collatz server seems to be ignoring my resource settings (Message 20252)
Posted 1077 days ago by marmot
Collatz is set to 100 resource (the base) and the other 8 projects I'm attached to on this machine go up by multiples of 4x from 100. Collatz is set to run on GPU's only as a last resort if Einstein, SETI, Moo! and Prime (the other 4 projects that will give out GPU work units for this pre-Fermi, NVidia Quadro FX 3700 GPU) have no work or their servers are down.

So here is a chart of the resource shares allotted:
Project Resource Credit
Einstein - 409,600 112,000 SETI - 102,400 494,000 PrimeGrid - 1,600 101,777 Moo! - 400 20,566 Collatz - 100 272,988


Now, Collatz gives out extremely inflated credit for WU and is way ahead of almost all the other projects, even though my machines have only given it approx. 96 total hours of work over the Christmas break. It's at half the level of SETI even though at least one of my older machines worked on it for 2 years and SETI is allotted 1,000 times the resource share.

I made sure the servers of the other 4 projects were not down after seeing Collatz download new work. So, why did Collatz flood my GPU with 4 days (my buffer setting because of the typical, expected holiday server downtimes) worth of new WU's that I had to mostly abort (I left the minis run)?


Run only the selected applications
(Size hints: Large = 16 x Solo;
Solo = 16 x Mini;
Mini = 16 x Micro)

Looking at the above I dont see where the inflation on credits comes in.

I tried looking at the history of all the units that gave me the 272,000 credit that were done in the last 2 months and it's not available so I'll go by individual work units. I know that my machines did under 30 total WU's compared to Asteroids@Home that did 1,000+ in the same time period.

Two units available:
Unit 1: 28,277 seconds for 15,953.76 credit = 2031 credit per hour.
Unit 2: 28,004 seconds for 16,237.59 credit = 2087 credit per hour.


On the same GPU
Einstein@Home CUDA WU took 14.9 hours to receive 1000 credit giving 67 credit per hour.

Asteroids@Home has 2,179,320 over 3800 WU's for 573 per WU that takes average of 6:29 hours for 88.1 credit per hour. Those are all CPU WU's and comparable to credit handed out by Einstein GPU WU's. Asteroids is also a generous project for WU credit.

Einstein@Home is fairly generous in credit yet Collatz is 30x more generous on the same GPU.

That's the inflation I'm referring to.

***Again, it has nothing at all to do with why Collatz gave out WU's when Einstein, SETI, Moo! and Prime servers were all up and running with higher resource shares set. ***
8) Message boards : Number crunching : Collatz server seems to be ignoring my resource settings (Message 20250)
Posted 1077 days ago by marmot
Collatz is set to 100 resource (the base) and the other 8 projects I'm attached to on this machine go up by multiples of 4x from 100. Collatz is set to run on GPU's only as a last resort if Einstein, SETI, Moo! and Prime (the other 4 projects that will give out GPU work units for this pre-Fermi, NVidia Quadro FX 3700 GPU) have no work or their servers are down.

So here is a chart of the resource shares allotted:
Project Resource Credit
Einstein - 409,600 112,000 SETI - 102,400 494,000 PrimeGrid - 1,600 101,777 Moo! - 400 20,566 Collatz - 100 272,988


Now, Collatz gives out extremely inflated credit for WU and is way ahead of almost all the other projects, even though my machines have only given it approx. 96 total hours of work over the Christmas break. It's at half the level of SETI even though at least one of my older machines worked on it for 2 years and SETI is allotted 1,000 times the resource share.

I made sure the servers of the other 4 projects were not down after seeing Collatz download new work. So, why did Collatz flood my GPU with 4 days (my buffer setting because of the typical, expected holiday server downtimes) worth of new WU's that I had to mostly abort (I left the minis run)?


Because Boinc doesn't work like you think it does, Boinc works off the rac of each project and then combined with your percentage setting tries to even them all out. The formula is VERY complicated and has been evolving for a LONG time, and is still being worked on even today.

As for the credits each project gives, they are NOT comparable between projects, each project sets their own amount of granted per workunit credits. They set that number based on a TON of things, more credits means more users hang around but also means a bigger load on the Projects Server. Some projects even get paid for their results, so in some cases that figures into the equation too.

You also cannot compare cpu credits to gpu credits as a gpu can do about 10 times, or more, as much work in the same amount of time a single core of a cpu
can.

The easiest thing for you to do would be to get more machines and try and even out the high paying projects on one machine, or set of machines, and the lower paying projects on other machines. Another option would be to turn the work fetch off for some projects for a given time period, say 30 days, that way the other projects would get exclusivity to your machine for that time period, letting you decide who sends you work and who doesn't. You could also reduce the percentages to zero for some projects, meaning don't get ANY work unless the other projects don't have any. Your setting of 100 only kinda sorta does that, but obviously not to your satisfaction.


OK, you are making some poor assumptions. I have 7 machines running with four of the fastest connected to Collatz and 8 other projects. They have been setup this way since September and yet Collatz still is WAY out of control with credit approaching the second and first highest projects that have 1000+ times the resource share and this was done in under 96 hours. The top two projects were worked on for a couple years with an Intel t7500, among older CPU's.

That is some RIDICULOUS WU inflation for Collatz!!!
GPU efficiencies over CPU on those calculations can not account for that inflation.

Yes, I understand each project sets it's own WU payout and I'm saying Collatz is ridiculously high for a mathematics problem that has no apparent real world usefulness as opposed to say, mapping asteroids and their orbits or folding proteins.

*** That all has nothing to do with the point that Collatz downloaded a lot of work when it's resource share is 1/4000th that of Asteroids@Home WHILE Asteroids@Home was still up and running. ***
9) Message boards : Number crunching : Collatz server seems to be ignoring my resource settings (Message 20235)
Posted 1080 days ago by marmot
Collatz is set to 100 resource (the base) and the other 8 projects I'm attached to on this machine go up by multiples of 4x from 100. Collatz is set to run on GPU's only as a last resort if Einstein, SETI, Moo! and Prime (the other 4 projects that will give out GPU work units for this pre-Fermi, NVidia Quadro FX 3700 GPU) have no work or their servers are down.

So here is a chart of the resource shares allotted:
Project Resource Credit
Einstein - 409,600 112,000 SETI - 102,400 494,000 PrimeGrid - 1,600 101,777 Moo! - 400 20,566 Collatz - 100 272,988


Now, Collatz gives out extremely inflated credit for WU and is way ahead of almost all the other projects, even though my machines have only given it approx. 96 total hours of work over the Christmas break. It's at half the level of SETI even though at least one of my older machines worked on it for 2 years and SETI is allotted 1,000 times the resource share.

I made sure the servers of the other 4 projects were not down after seeing Collatz download new work. So, why did Collatz flood my GPU with 4 days (my buffer setting because of the typical, expected holiday server downtimes) worth of new WU's that I had to mostly abort (I left the minis run)?




Main page · Your account · Message boards


Copyright © 2018 Jon Sonntag; All rights reserved.