Jump to content

viming_aint_easy

Member
 View Profile  See their activity

  • Posts

    36

  • Joined

  • Last visited

Awards

Recent Profile Visitors

The recent visitors block is disabled and is not being shown to other users.

viming_aint_easy's Achievements

Newbie

Newbie (1/14)

32

Reputation

  1. viming_aint_easy
    I'm not sure if the cores the card would need to specifically physical, or if hyperthreaded ones are fine, but I presume they are. So, depending on your workload, you may be fine with a r7 or r5.
  2. viming_aint_easy
    I believe that this is a consequence of what GPU you have. Nvidia cards use polling (which works best with dedicated cores), while AMD cards use interrupts (which doesn't care). I'd link to some folding forum posts, but it looks like their SSL cert expired today.
  3. viming_aint_easy
    I have a Linux box running a 660 Ti and FAHControl is reporting 230k PPD. Not sure how much I trust it, but...
  4. viming_aint_easy
    Yeah, I've been fortunately successful with sigterm, rather than sigkill, so I haven't gotten that far. Reminds me of a page presenting the description of "kill -9" as the equivalent of using a combine to harvest a flowerpot. However, what's odd is that the client behaves just fine on my main boxen, which is running the same exact OS version. Agreed. I've seen plenty of adequate FAHClient.system files, but that's up to the FAH team to incorporate, and I do not envy them for the mountain of workload that they have to take on.
  5. viming_aint_easy
    Anyone else running Linux have a problem with systemd/sysV and the FAHClient. It seems that even though I do a $(systemctl stop FAHClient) or $(service FAHClient stop), the process still remains. This has lead to an irritating problem where I would be sending instructions to what I would expect to be FAH, but it would be the wrong client! Systemd says the client is dead, but the process is clearly still running. OS: Ubuntu 20.04 Any suggestions?
  6. viming_aint_easy
    Wait, there are daily updates? Nice! </spoiled>
  7. viming_aint_easy
    I was aware that you can use multiple GPUs in one system and have them all fold, but I was under the impression that you could only map one WU to one GPU, not have multiple cards work on an individual WU. Is that correct?
  8. viming_aint_easy
    Wait, F@H supports SLI? Or does the name of the card, by default, include "SLI"?
  9. viming_aint_easy
    Is there a documented AMD equivalent (specifically Navi) or is this an Nvidia exclusive?
  10. viming_aint_easy
    I've used it on Ubuntu, not Fedora, but I've had good success with getting the control working with this code, which addresses the python2 dependency: https://github.com/cdberkstresser/fah-control Apply standard "use at your own risk" template.
  11. viming_aint_easy
    F@H science question: Are atom counts per project "easily" malleable? Are projects pretty inflexible on how many atoms they involve, or, generally, can they be scaled on-demand? I'm curious if the under-utilization of 2080s and higher is a problem that can be fixed with time, or are we stuck with "low" atom counts because that is how physics works.
  12. viming_aint_easy
    Oh, don't you worry. This puppy will keep chugging along even if I were to get 2 3080's. I have a nice spot on my local NAS (repurposed from my previous desktop) that would fit beautifully.
  13. viming_aint_easy
    Congrats to those that are benefiting from CUDA and congrats to the science! While I weep for my RX5700, I'm still grateful for its service. I'm just hoping that by the time I'll be able to get my hands on a 3080 (probably the beginning of next year), I too can join the good-a CUDA comput-a club.
  14. viming_aint_easy
    Are you on Beta? I believe they are sending out test units that they are benchmarking for CUDA. https://foldingforum.org/viewtopic.php?f=66&t=36129#p343091
  15. viming_aint_easy
    I'm curious as well. Considering that sprint 4 is still slowly progressing at 44%, I'd imagine they'd still have a boatload of WUs to hand out. Maybe it's not lack WUs, but an issue with infrastructure/demand (pure speculation)? Edit: Further investigation on my end indicates that all of the "WUs are not available" errors, on my system, are for CPU slots, not GPU.
×