Announcement

Collapse

Welcome Back!

I apologize for not getting this up and done sooner. Needless to say the forums, and the front page got hacked, we since then have locked things down, and things appear to be better over all.

I was able to get the front page done sooner, but the forums with all of its data was MUCH more difficult to deal with. Needless to say, I just finished it! YAHOO!

Just in time with Facebook about to take a nose dive, and doing other crazy things to people. (Censorship, etc etc).
See more
See less

XCPUs.com's FOLDING NEEDS YOU!!!

Collapse
This is a sticky topic.
X
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Having to cut back a bit because of heat and the expected high electricity bill. Turned one folder off so far today and may turn another off later. Kind of wish I could build another SR-2. The PPD isn't near as good as it used to be on one of those computers, but its still a lot of PPD. One SR-2 still produces more PPD that two of my lessor computers for about one third the electricity cost since I don't use a GPU client with it. Reslly wish a SR-X wasn't so expensive to build and also had better performance. Oh well.
    Over 60- Seen it, Done it, Can't remember it, but I miss it.
    The great economist Milton Friedman once observed, "Many people want government to protect the consumer. A much more urgent problem is to protect the consumer from the government.
    Remember, the government cannot give anything to anyone that they have not first taken away from someone else.
    Tu ne cede malis, sed contra audentior ito - Do not give in to evil but proceed ever more boldly against it:

    https://www.fuelly.com/driver/sailer...ang?fu=2315423

    Comment


    • I've wondered why my SR-2 hasn't gotten any -bigadv WUs for a long time. It got a couple project 6900s, and then nothing. Looking through the FAH forums I found an answer:

      It's also worth noting that you won't get any BigAdv work units on a Windows client. Try Linux.
      Because the larger bigadv work units crash on Windows because the fahcore is still 32 bit and bigadv WUs needs more memory than the 32 bit fahcore can reference.

      Your next question should be addressed to Pande Group. I don't know why a 64 bit windows fahcore hasn't been released.
      So its all Pande Groups fault. That leaves the next move by myself being to load a HDD with Linux and try again. More work.

      Seeing the SR-X has become a great idea in concept but a total failure in reality, my next Folding build will either be another SR-2 or and AMD based 4 CPU computer with Magny Cours chips, and of course, a Linux OS. But wow, a 4P AMD machine would get expensive fast. Never built anything like that, and if I bought one, I hate to think of the money spent. Why Magny Cours and not Interlagos? Because the Interlagos chips don't fold good at all even though they have more cores. The blame seems to be that Interlagos chips have less FPUs and that hurts Folding performance a lot. So its looks like for now, both Intel and AMD have screwed up their CPUs when it comes to Folding and leaving high end Folders to old hardware.

      But Fold on, me maties. Its all for science. Every WU submitted helps.
      Over 60- Seen it, Done it, Can't remember it, but I miss it.
      The great economist Milton Friedman once observed, "Many people want government to protect the consumer. A much more urgent problem is to protect the consumer from the government.
      Remember, the government cannot give anything to anyone that they have not first taken away from someone else.
      Tu ne cede malis, sed contra audentior ito - Do not give in to evil but proceed ever more boldly against it:

      https://www.fuelly.com/driver/sailer...ang?fu=2315423

      Comment


      • I've started converting my computers back to using v6 SMP and GPU clients. The v7 client has slowed way down on all machines, dropping my PPD 5k to 10k PPD on each one. Even the SR-2 has dropped to about 60k in recent days. Some of the new projects take a very long time to complete and get very few points. Until Stanford makes some adjustments, its back to v6 until there are no more A3 WUs left.
        Over 60- Seen it, Done it, Can't remember it, but I miss it.
        The great economist Milton Friedman once observed, "Many people want government to protect the consumer. A much more urgent problem is to protect the consumer from the government.
        Remember, the government cannot give anything to anyone that they have not first taken away from someone else.
        Tu ne cede malis, sed contra audentior ito - Do not give in to evil but proceed ever more boldly against it:

        https://www.fuelly.com/driver/sailer...ang?fu=2315423

        Comment


        • Hmmm, no one around for a long time.

          Anyway, the recent heat has been a bit hard on my computers, with two of them acting up a lot. Had to drop the overclocks and voltage a fair amount to keep them going. Also the WUs that have been coming out are taking a lot more time for the points they give. My PPD from my 920s is pitiful and the 970s are getting as much PPD as the 920s used to get. Its close to time that I need to do some upgrades, just not sure which direction though. If Intel released the 8 core Ivy Bridge-e I would be happy, but I haven't seen any hard news when that will happen. That leaves an upgrade to a 6 core 2011 setup or building another SR-2. Perish that thought, but the alternative is to play hardball and go with a 4 processor AMD setup which would be very expensive, besides which I might have to have the house rewired to support the energy draw. SR-X is also very expensive and so far doesn't seem to do any better than a SR-2.

          Oh well, perhaps the time has come that I slowly back away from folding as the hobby is getting rather expensive. Besides, having all these space heaters going when its nearly 100f outside makes the house into a sauna.

          Want to give thanks to Leganfuh for the folding he put in during the past couple months. I remember that it depended a bit on finances and such is tight. I'm feeling the pinch now myself.

          Fold on people. Even if its only a little bit now and then, science is good.
          Over 60- Seen it, Done it, Can't remember it, but I miss it.
          The great economist Milton Friedman once observed, "Many people want government to protect the consumer. A much more urgent problem is to protect the consumer from the government.
          Remember, the government cannot give anything to anyone that they have not first taken away from someone else.
          Tu ne cede malis, sed contra audentior ito - Do not give in to evil but proceed ever more boldly against it:

          https://www.fuelly.com/driver/sailer...ang?fu=2315423

          Comment


          • I'm still going on the 2 machines I have, but I'm focusing more on high-efficiency. I want all my machines to pull 150W from the wall or less.
            1: Main Rig - eVGA Classified SR-2 | 2 x Xeon X5670 @ 4.05GHz | 12GB Mushkin PC3-1600 DDR3 | OCZ IBIS 240GB RAID SSD | ATI Radeon 5770 1GB | Dual Dell 3007WFP-HC Monitors
            2: Server - eVGA X58 SLI | L5640 @ 4GHz | 6GB OCZ PC3-1866 DDR3 | Intl X25-M G2 80GB OS Drive | 6 x 1.5TB Seagate 7200.11 RAID 10


            Comment


            • My SR-2 has a pair of 80wt chips in it and a 8400GS video card which doesn't run a GPU client. I figure that it hardly pulls anything from the wall. If I built a new strictly folding machine, it would be a duplicate of that. I don't have any way to measure the actual draw. I'll be experimenting with Linux soon. From what I've read, Linux is the only way to run Bigadv anymore, which would about double the PPD from the SR-2, perhaps more. If so, two SR-2s by themselves would increase my PPD even if I shut down all the rest of my computers. That sounds good to me, though I would probably keep a couple 970 machines going.

              I plan to sell all my 920s, perhaps the oldest 970 as well. My present computers are all old and don't fold well. Its just getting to be too much money invested, not just the computers, but the electricity to power them. Top that off with the games Stanford is playing and I get discouraged. Oh well. Would be nice if Stanford wrote a 64 bit program for Windows so I could run bigadv on them.

              One other thought. I used to use Virtualbox to run Bigadv. Would that work instead of Linux?
              Over 60- Seen it, Done it, Can't remember it, but I miss it.
              The great economist Milton Friedman once observed, "Many people want government to protect the consumer. A much more urgent problem is to protect the consumer from the government.
              Remember, the government cannot give anything to anyone that they have not first taken away from someone else.
              Tu ne cede malis, sed contra audentior ito - Do not give in to evil but proceed ever more boldly against it:

              https://www.fuelly.com/driver/sailer...ang?fu=2315423

              Comment


              • Originally posted by brentpresley View Post
                I'm still going on the 2 machines I have, but I'm focusing more on high-efficiency. I want all my machines to pull 150W from the wall or less.
                Interesting. I always like to read on power efficiency. In the case of folding there are 2 very different ways to consider efficiency. Either you want A)maximum folding per watt (no matter the watts, even 250+,etc.), or B) low power period, and just whatever folding you can get during low power (and then there are two choices: setting a wattage limit (as you have), or finding the lowest possible wattage setup that provides a certain level of folding).

                Interesting choices!

                For my own aims, which are different than yours, but overlap, I want the lowest possible idle power draw for my system, in combination with a very good system speed when I have a demanding application (and then I don't care much about that maximum power). The reason for this is that my computer when fully on will not do much most of the time, and when I do something demanding that matters, like chess position analysis, I only do it for 1-5 hours in a month. So that high-power draw doesn't matter in my case.

                But the overlap we have in power consideration is everything else: the psu efficiency, the power draw of the monitor(s) (3 in my case, and they draw most of the power in my case!), and any other ongoing power draws, like an old printer power brick (I found one was drawing 6 watts all the time!), etc.....
                A foolish consistency is the hobgoblin of little minds -- Emerson
                On Earth, what goes up tends to come down.

                Here's more at my blog:
                http://findingourdream.blogspot.com/

                Comment


                • Well, I'm on the front page again and rising.

                  Individual Overtakes - EXTREME Overclocking Folding @ Home Stats

                  I'm running 8 computers at the moment; three 920s, three 970s, one SR-2 and one 1090T with an assortment of GTX469 and GTX560 Ti cards. Haven't built or updated anything for several months. Wish Intel would release the 8-core Ivy Bridge-E or AMD would release the 8-core Piledriver and give me an excuse to build. With luck, a 8-core Ivy Bridge-E would put out as much PPD as two of my 920 machines and I could retire them. The other possibility is to build another SR-2. Really mixed feelings on that since the tech is so old. Going with a SR-X would be theoretically better, but the cost is very high. Even so, a SR-2 without a GPU client puts out about the same PPD as two of the 920 machines with GTX560 Ti clients as well, so the SR-2 ends up burning a hole lot less electricity and making less heat. Oh well. Time will tell what I do.

                  In the meantime, fold on everyone. Its all for the good of science, as well as the team.
                  Over 60- Seen it, Done it, Can't remember it, but I miss it.
                  The great economist Milton Friedman once observed, "Many people want government to protect the consumer. A much more urgent problem is to protect the consumer from the government.
                  Remember, the government cannot give anything to anyone that they have not first taken away from someone else.
                  Tu ne cede malis, sed contra audentior ito - Do not give in to evil but proceed ever more boldly against it:

                  https://www.fuelly.com/driver/sailer...ang?fu=2315423

                  Comment


                  • Bought one of the new GTX660 Ti cards for Folding. Its an EVGA SC model. Tried it on one computer yesterday and v7.1.52 wouldn't recognize it as being there. So I tried it on a different computer this morning, using the v6.41 client. It starts, but then quits. Doesn't work with v7.1.52 either. Don't know if the card itself is bad or if Stanford is lagging in allowing new cards. Its the pits. I suspect the problem is that the card is too new and Stanford hasn't written in the programs to allow it yet. Guess its back to using using the old GTX460 card for now.

                    The card seems to work well enough otherwise, so I don't have a good reason to blame the card itself.

                    Edit: Asked about this in the Folding forum and was told that the Kepler cards aren't supported. Worse, it might be another year before they do get support, though no one seems to know for sure. Was told I could become a BETA tester, but that seems like a lot of work for little, if any, return. Guess I should have read the forums before buying the card. Would have bought a GTX570 instead. Wondering about returning it and going for a 570 or just keeping it for games and hopes that Stanford will update to include the Kepler cards.
                    Last edited by Sailer; 09-07-2012, 01:51 PM.
                    Over 60- Seen it, Done it, Can't remember it, but I miss it.
                    The great economist Milton Friedman once observed, "Many people want government to protect the consumer. A much more urgent problem is to protect the consumer from the government.
                    Remember, the government cannot give anything to anyone that they have not first taken away from someone else.
                    Tu ne cede malis, sed contra audentior ito - Do not give in to evil but proceed ever more boldly against it:

                    https://www.fuelly.com/driver/sailer...ang?fu=2315423

                    Comment


                    • I don't know why and I can't explain it, but I hooked up my old QX9650 and its getting almost identical PPD as one of my 920 computers. I really wonder if the v7 clients can tell the difference between real cores and hyper-threaded ones. If so, it would explain why my AMD X6 1090T turns a higher PPD than any of my 920 computers.
                      Over 60- Seen it, Done it, Can't remember it, but I miss it.
                      The great economist Milton Friedman once observed, "Many people want government to protect the consumer. A much more urgent problem is to protect the consumer from the government.
                      Remember, the government cannot give anything to anyone that they have not first taken away from someone else.
                      Tu ne cede malis, sed contra audentior ito - Do not give in to evil but proceed ever more boldly against it:

                      https://www.fuelly.com/driver/sailer...ang?fu=2315423

                      Comment


                      • I never bothered to read up on hyperthreading, but my guess is that it only helps in situations where a core is momentarily idle due to fetching or waiting on a hard drive, etc., so that another thread can execute for a bit of time. But a well-designed computational problem will rarely have that kind of situation, so that kind of thread switching would not really help. I wonder if that is exactly why the hyperthreading doesn't help.
                        A foolish consistency is the hobgoblin of little minds -- Emerson
                        On Earth, what goes up tends to come down.

                        Here's more at my blog:
                        http://findingourdream.blogspot.com/

                        Comment


                        • Quite possibly. I don't know enough about it to say one way or another. The folding clients through version 6 allowed for hyperthreading and that was a major reason that Intel chips had such a good lead over AMD chips. With the version 7 client, it seems like hyperthreading isn't getting counted as real computational threads, which then gives the AMD chips a big comparative boost. It does get a bit confusing with the v7 client, as my SR-2 shows up as 24 cores instead of the physical 12 cores with v7, but the actual computing speed is slow and the resulting PPD is low. As it is, for the highest PPD I'm stuck running v6 and even then only get 75-80K PPD on the SR-2. The top setups currently appear to be AMD with multiple 12 core Magny Cours CPUs that have no hyperthreading. In the meantime, My AMD 1090T with 6 real cores is averaging about 10k PPD more than the 4 real core 920s. Again, its like the real cores are working better than the hypertrheaded cores.
                          Over 60- Seen it, Done it, Can't remember it, but I miss it.
                          The great economist Milton Friedman once observed, "Many people want government to protect the consumer. A much more urgent problem is to protect the consumer from the government.
                          Remember, the government cannot give anything to anyone that they have not first taken away from someone else.
                          Tu ne cede malis, sed contra audentior ito - Do not give in to evil but proceed ever more boldly against it:

                          https://www.fuelly.com/driver/sailer...ang?fu=2315423

                          Comment


                          • Did a new build a few days ago, replacing a 920 with a GTX460 with a 3930K and a GTX660 Ti. Haven't managed to get the GTX660 to fold yet, but just the 3930K at 3.8ghz is turning between 30k and 35K PPD. This is about 5k more without using a GPU client so less electricity used for more PPD sounds good to me.

                            I went to the FAH forum to find out if I could get the GTX660 going, but the instructions there are either incomplete or wrong. Folding@home: Support for new GPUs (such as Kepler) in the v7 FAH client They say to copy and paste the new "White List" into a run/data directory, but I can't find one in the FAH client folder. Don't know if I should make a separate folder named "Data directory" or what. The instructions need to be a bit more explicit for me. In any case, the CPU is folding pretty good all by itself. Might install the GTX460 from the old computer and see how that works when I have the time for it.
                            Over 60- Seen it, Done it, Can't remember it, but I miss it.
                            The great economist Milton Friedman once observed, "Many people want government to protect the consumer. A much more urgent problem is to protect the consumer from the government.
                            Remember, the government cannot give anything to anyone that they have not first taken away from someone else.
                            Tu ne cede malis, sed contra audentior ito - Do not give in to evil but proceed ever more boldly against it:

                            https://www.fuelly.com/driver/sailer...ang?fu=2315423

                            Comment


                            • It appears that the A3 WUs are either running out or have run out. The computers on which I've been using v6 are now only getting A4s, with the PPD dropping anywhere from 5k to 10k on them. The worst problem has been that going to v7.1.52 resulted in my new LGA 2011 computer with a 3930K chip dropping from 35-40k PPD with v6 on just a CPU client to just 35k PPD with v7 after adding a GTX460 to run a GPU client. Some of the new A4 Us are taking over a day and a half to complete while giving relatively low PPD numbers. Even my SR2 has dropped from 80k to 65k with the A4 WUs. Until and unless Stanford starts getting the PPD more in line with the time element, along with the cost to build a new computer, I've built my last folder. Its just way to expensive for me to justify any longer.

                              Edit: From what I am observing with the LGA 2011 3930K 6/12 core running at 3.8ghz, its getting no higher PPD with v7.1.52 than my old AMD 1090T 6 core running at 3.6ghz with v7.1.52. As far as folding goes, Intel no longer has any advantage over AMD when it comes to single 6 core CPUs. Just my observation, but one based on several computers that I currently have running.
                              Last edited by Sailer; 10-26-2012, 11:13 PM.
                              Over 60- Seen it, Done it, Can't remember it, but I miss it.
                              The great economist Milton Friedman once observed, "Many people want government to protect the consumer. A much more urgent problem is to protect the consumer from the government.
                              Remember, the government cannot give anything to anyone that they have not first taken away from someone else.
                              Tu ne cede malis, sed contra audentior ito - Do not give in to evil but proceed ever more boldly against it:

                              https://www.fuelly.com/driver/sailer...ang?fu=2315423

                              Comment


                              • A new folding client has been released, version 7.2.9.x86. Supposedly this will finally give support to Kepler video cards. I'll be giving it a try in a few hours when the present WU finishes and then report if it actually works or not, as well as how efficiently it works. Early reports seem to be mixed, but if it folds better than the last GTX460 that I am using, I'll be satisfied.

                                Edit: The GTX660 Ti Kepler card has been going for a couple hours now and is actually getting less PPD than the GTX570 did. From what little I've read on the Folding forums, while v7.2.9 now allows the use of the Kepler cards, it isn't tuned to use them properly and a number of people have complained about getting less PPD than the Fermi cards. While I was changing cards, I replaced an old GTX460 with the GTX570 and the 570 is only getting a couple thousand more PPD than the 460 did. Probably this is because that 460 had 1gb on memory instead of the standard 760kb for the GTX460 cards. So with the Kepler card loosing a couple thousand PPD and the Fermi card gaining a couple thousand, it ends up as a wash. For now, the GTX570 cards seem to be the best buy for the PPD.
                                Last edited by Sailer; 11-03-2012, 05:27 PM.
                                Over 60- Seen it, Done it, Can't remember it, but I miss it.
                                The great economist Milton Friedman once observed, "Many people want government to protect the consumer. A much more urgent problem is to protect the consumer from the government.
                                Remember, the government cannot give anything to anyone that they have not first taken away from someone else.
                                Tu ne cede malis, sed contra audentior ito - Do not give in to evil but proceed ever more boldly against it:

                                https://www.fuelly.com/driver/sailer...ang?fu=2315423

                                Comment

                                Working...
                                X