Jump to content

Change Mode

uncle fuzzy

Members
  • Content Count

    192
  • Joined

  • Last visited

Everything posted by uncle fuzzy

  1. Thanks for the heads up. I've now done a bunch of these. +1500 for GTS450 and +3K for GTX460. The temps are steady on the 450s, but up 10C on the 460s.
  2. Sorry, but that is wrong. You have to have GPU3 to run Fermi, you do not have to have Fermi to run GPU3. I've been running GPU3 on all NV from 8800 up since it came out. You can also run 8800 and up on V7. I think the only cards that can't use GPU3 are ATI 2xxx-3xxx or 2xxx-4xxx.
  3. Having an entire Distributed Computing forum instead of my single thread, you guys are all set up to support a PitCrew BOINC team right here. It does sound like a good use of hardware for those obsoleted out of FAH.
  4. What is this BPINC of which you speak? jk I'm not advertising, just answering questions. I'm on FAH "Team Elder Scrolls", but I'm not doing BOINC. We started on Planet Elder Scrolls with a bunch of people who played Morrowind. The team thread has since migrated to the Bethesda forums (makers of the Elder Scrolls series). A guy posted in the folding thread about running BOINC. With a Moderator (former folder) we got him his own thread (still floating, not pinned like mine). They also use the name Team Elder Scrolls to stay connected with the forum. Thread- http://forums.bethso...rolls-on-
  5. Well, it kind of makes my i7 purchases a waste of money, especially the 980X. I'm really glad I never gave in to the urge to buy a 2600K. I can still get 15-20K each with SMP (instead of the 50-80K with -bigadv), but in hindsight more GPUs on cheap dual core systems might have been a better buy. I'm not a terribly happy camper, but I'll live. I had played with the idea of building a server over the last few years. I may still go that way in the future. I see AMD came out with a 16 core Opteron last week....
  6. Several people on my "home" forum started a BOINC team about a month ago. A couple of the guys are former folders with older hardware. At least one of them does both BOINC and [email protected]
  7. What does a Kansas boy know about them big city minutes?
  8. Da nada. At least it's folding again instead of sitting idle in my parts closet. The P45 board it came from is just sitting here gathering dust, if you think you might want to try it. Gigabyte EP45-UD3P 2 PCIE slots
  9. Nope, it was my standard brand- Gigabyte. My old P35 has been running at 3.4 for about 4 years, but I could never get the 2 P45 boards above 2.8.
  10. He has v7. You have to run smp 8 to get bigadv with v7. smp 7 will download regular smp WU and repeatedly crash. v7 doesn't like most odd numbers. I can run smp 3 on a quad, but on an i7 I never tried 5, but 7 will always crash. QRB- quick return bonus, the faster you complete it, the more points you get. Requires the passkey and completing 10 a3/a5 smp WUs on time before you qualify for the bonus. bigadv WUs- I'm mostly seeing 6900 and 6901 with times of 28 (980X smp 10) to 37 minutes (930 and 950). Both are a5 WUs. First, we need to get the right WU on your machine, then we ca
  11. Yes, I'll be repeating a few things you already covered. Please don't yell at me. Scrolling up and down while trying to write sucks. Are you running v7 or v6? v7 saying 1300 PPD could be right because it doesn't figure QRB worth a crap. 1300 PPD on Fahmon or HFM running v6 could be a problem, or "figure QRB" is not checked, or something is wrong in config and it doesn't think you get the QRB. TPF is a better indicator. Depending on your clocks, you should see 30-45 minutes. QRB needs a passkey and 10 WUs completed on any SMP client using that passkey, on any machine. If you entered the p
  12. Yup. They even let me come around once in a while.
  13. Any chance of a bios update to handle the Fermi?
  14. Now it sounds like a mobo slot or PSU problem. Bummer, Dude.
  15. Try a different driver. There have been a crap load of reports on folding problems using 27x.xx drivers. I'm still running 260.99 on all of mine. My GPUZ has OpenCL, CUDA, and DirecrCompute 5.0 checked (I manually set PhysX to the cpu on my boxes).
  16. I've had problems with running mixed Fermi/non-Fermi with both v6 and v7. Some peole have done it, so it seems to be a gpu/mobo quirk that you can't really predict. I did manage it briefly on one box with a 260 and 460 over a year ago. Have you tried the most basic setup- the 460 only, no -forcegpu flag, delete work folder, core, queue.dat, and unitinfo.txt (I'm writing on a laptop with no folding files, so I'm working from memory)? A new install might be easier. -remove video drivers (I just uninstall, I don't use driversweeper) -shut down -install 460 only -restart, let Win
  17. Is this the only card in the box? If so, try it without the -forcegpu flag. I run without it on the 2 boxes using v6 clients, running both 450s and 460s. What card is it? If you're forcing Fermi on a non-Fermi, that could be your problem.
  18. I asked about this one on my home forum before coming here for suggestions on the 980X. Everyone said it was loud and no more effecient than a good air cooler. I haven't seen or heard anything about the one you posted.
  19. 13K ain't bad. I can hit 11.5K on a heavily OCd 460, so it might be right for your current clocks.
  20. You don't show the top of the logs. Which version of the client are you running? v6.23 will not work with a Fermi card, no matter what flags you add. Check this thread, section 5 for download, and either 6 (systray) or 7 (console) for installing. http://foldingforum.org/viewtopic.php?f=59&t=15196 The cudart files you need are included in the zip download. If you had an existing GPU2 console client, you can use the zip download to upgrade what you have (section 8).
  21. The Fermis actually need more cpu time than the earlier nVidia gpus. From the little I've seen (450 and 460), the higher the model number, the more cpu they need. The Fermis don't require the systray client. You can use it if you prefer it. Like RAH, I run console clients. To make them play nice together, set the SMP at "idle" priority and the systray GPU at "slightly higher" or the console GPU at "low". The second step is to run the SMP at # of cores minus 1 (a quad would be -smp 3). It slows down the SMP a little, but gives full resources to the GPU client, which is the main point-gett
  22. I'm all over the place with my OCs. It's been far from scientific or systematic. Pretty much, I just change something that makes the cpu operating frequency go up, and if it doesn't crash the system, it's good. Some day I'll have to go through all of them and do it right, but I'm afraid of loosing what I have. [email protected] 216X19 QPI 3888MHz 1.376v temperature limited at 80C, Megahalem w/ 2 Scythe 110cfm in push/pull [email protected] 172X21 QPI 3096MHz 1.36v temp limited at 80C, using the stock Intel cooler that came with the 980X [email protected] 133X29 QPI 3184MHz 1.
  23. I was a little early on the happy thoughts. After I powered up, I started having to add coolant several times a day. It took a while, but I finally found slow drips from both fittings on the radiator. I cranked them down and stopped the leaks, but I now have slight twists in the tubing. I'll take it all apart and re-do it when I take vacation at the end of the year. The initial 3.4 was not stable. The machine crashed very shortly after I turned off the monitor at the end of my last post. I went back in and started over. I updated the BIOS, hoping it would give me the right default setting
  24. I got my cables on Friday evening, but had to work 12 hours on Saturday. I still got most of it wired Sat. night, and finished it up Sunday. Many hours today installing the OS, drivers, updates.... Initial results- For some reason, this defaulted to 2.6GHz on startup instead of the 3.33 it should have been. I bumped it up to 3.4 and ran 3 hours of StressCPU2. Idle was 20C, and full load made it all the way to 40. I figured I'd take the chance and set up my clients. For the moment, this has 2x 460s, spaced out and folding at 45C. I'm not sure what I'll end up putting in here as I
  25. I don't think that's quite what they have in mind when they talk about "silent computing".
×
×
  • Create New...