How to block WU 7611

[Lloyd]

How can i stop from getting WU 7611's.They are 788 points with a PPD of 785.93.They take over 24hrs with a TPF of 14 minutes 26 sec. I can turn out 10583 points per day with other work units.It appears my AMD 955 Black 4 core is not big enough to run this thing and would rather give them to more powerful systems to complete but they keep getting assigned and dragging me down.Any help would be appreciated.

[Zagen30]

Where are you getting the 785 PPD figure from? If you're using v7 and getting that number from the client's PPD estimate, know that most of the work servers currently don't provide the bonus information to the client, and thus it calculates PPD from base points alone. Assuming you're bonus-qualified, you'll get the bonus anyway; the stats servers have the necessary information and will add the proper amount of bonus points.

That being said, the Pande Group does not let you block certain projects so as to prevent cherry-picking.

The only projects for which your 955 are not capable are those that can only be obtained using the -bigadv flag, but even if you entered that flag you wouldn't get them since your CPU has to have 8 cores for the servers to send them out.

[Lloyd]

788 base points in v7 and the bonus which i get doesn't add enough to make a difference. I am not interested in cherry picking but 7611 is a drag on my system so i guess i will shut down SMP and just run my twin GTX 460 for 21000 points a day instead of dragging all 4 cores for 24+ hours for a couple thousand points.That makes no sense.

[Jesse_V]

Your SMP is important though. Sorry that the ppd fluctuates like that. Regardless of the points you get, know that the CPU cycles you contribute does not have to be done by someone else, therefore you advance science that much further. Points do mean something though, and sorry the fluctation annoys you that much. I trust that PandeGroup does their best to make the credit as even as possible, but some can slip through somehow. Your SMP will crunch other projects besides 7611s, so most of the time it should be fine. Perhaps they'll make adjustments to the credit or take steps to have that project distributed more approriately. So please, fire up your SMP.

[7im]

We don't have a choice in work units, or everyone would try to fold some work units before others.

Bu as mentioned, all the work units need to get folded. I fold what they send me.

[bruce]

I don't see a recent credit for 7611 so I don't know what you're actually getting for them. Project 7611 is eligible for a bonus if your passkey is qualified. Are you using a passkey? The only WUs I can see don't seem to be getting a bonus but that doesn't mean you haven't gotten them in the past.

One such project was 10200 which you completed in 1.54 days. You earned 493 points for it which includes no bonus. Until we get your bonus status cleared up you don't really have a good basis to compare various projects.

[Zagen30]

788 base points in v7 and the bonus which i get doesn't add enough to make a difference. I am not interested in cherry picking but 7611 is a drag on my system so i guess i will shut down SMP and just run my twin GTX 460 for 21000 points a day instead of dragging all 4 cores for 24+ hours for a couple thousand points.That makes no sense.

Why is this an either/or scenario? SMP can fold alongside GPUs, and if you set the GPUs' priority to slightly higher than the SMP slot, they won't be slowed down by anything the SMP slot does.

[Lloyd]

Just finished 7611 (4,29,26) Base points 788 with a bonus of 2219 making a total of 2997. This is listed in the log.(Bruce). I am waiting till September 19th to get my hands on the AMD 8150p 8 core processor as i believe that it will cut down those times and make it worth while to run SMP and will result in much faster times.The way the point system works it would cost wise be better to run 4 gtx460 Sc in quad. However i will spend the 320.00 for an 8 core chip when comes out next month and give it a shot.I dont mind spending 90 dollars more a month in electricity to help out with these projects but to me and maybe only me, the 4 core is just too small for these projects. I am folding for EVGA now @ Lloyd61 and will run both my Gtx 460"s till the new chip comes out and fire that up and see how it does.I believe it will make a big difference. AMD 955 black overclocked 3.74 ghz ,4 core =2297 points for 24 hrs. Gtx 460SC,= 10517 points for 24 hrs x2.Hope this helps with my thought process.

[patonb]

Your issue is smp 4, try smp 3. 2 fermi cards suck the life out of smps, as the gpu slows the cpu calculations on that 1 core, and thus dumps the total speed.

Give smp 3 a try and see whats up.

[Lloyd]

Thanks for the all the info and my question was answered. NO. Only a month or so till new processor Amd 8150p , 8 core.

990FXA-UD3 Gigabyte, AMD 955 Black 4 core Overclocked 3.8 GHZ, 2- GTX 460 SC SLI, 8 Gb Corsair 1600 Duel channel.1000 watt Ultra PS
Ultra overclock push-pull fan cpu cooler. Thermaltake a60 Armour case with 1-220 mm Fan and 4- 180 mm fans filtered.

[bruce]

Your issue is smp 4, try smp 3. 2 fermi cards suck t5he life out of smps, as the gpu slows the cpu calculations on that 1 core, and thus dumps the total speed.

Give smp 3 a try and see whats up.

It's worth a try, but my guess is that setting smp to the actual number of CPUs that he has will be no worse than setting it to a lower number, provided the GPU cores are running at a higher priority than SMP. (This is a hunch, not an established fact. I'm open to actual data being presented, but only from AMD chips where HyperThreading is not a factor.)

The right answer is probably the one Lloyd already suggested. Wait a month and run smp 8 -- or better still, wait a little longer until there are actual reports of FAH running on those new chips. Advertising and hype before new chips hit the market may be accurate or tests may show some inaccuracies, and we won't know which until FAH is used for accurate benchmarking of the performance factors that we care about.

[Lloyd]

Thanks Bruce, I will let you know how the high end 8 core does as i will get it right when it comes available.In addition i will post the results of folding with all 8 core without the GPU's and then add 1 @ a time.So i have heard that i have to have 1/2 a cpu for each GPU that i am running so i will also check with 6 cores running with both GPU<s running. Will post right after i get the chip and run the WU,s.

[tjlane]

Hi Lloyd,

Please do post again if there are technical issues with p7611. I'm the project manager for this project, so want to ensure that everything is going smoothly. You should be getting a QRB that will boost your points above the value in your original post... the mods (posting here) should be able to help with most things, but feel free to post again with comments if p7611s act up.

Thanks for folding! It does make a difference for us here at Stanford!

TJ

[Lloyd]

tjane,Thanks for the post, at the present time i am completing approx. 15 GPU,s WU,s every 24 hrs and will continue to help all i can with what i have to work with. I am looking forward to installing the new 8 core processor and start helping with the CPU workunits. Keep up the good work and lets turn out some answers to some of these horrible afflictions that people all over the world suffer from.

[TheCraiks]

Is it something with AMD quad cores and this WU, I have had at least six 7611 units in the last few weeks on my Athlon X4 640 24/7 folding machine, each taking between 26 and 29 hours to fold with a return of about 2700 points. As a comparison it has folded a 7808 unit in about 35 hours with a return of 7870 points and a 7017 in about 8 hours for 2800 points, I realise all the units need folding but this one does not seem to suit my computer and keeps turning up. I have a standard out of the box V7 running on Server 2003 with 4G memory and not much else to do most of the time, is it just a question of waiting for different units or is there something I can do to either skip this unit or improve its folding rate?