Folding Forum

My first question:

I see you can set folding power to "Light", "Medium" or "Heavy" and set when to "While I'm working" or "Only when idle".

Is there no way to set it to "Light" when I'm working and "Medium" when idle? Seems like there should be, but I can't find that anywhere on the UI. I'm on Windows BTW.


My second question:

Does "Only when idle" mean only when the F@H screensaver is running? If I disable the screensaver and idle my computer my using the Windows lock screen (no screensaver, no power management), does it still fold?


Last question:

F@H assigned my computer two assignments:

CPU:
Estimated points: 3863
My computer has 32.95 days to complete this work unit.
ETA: 13 hours 14 mins
I'm at about 50% after leaving it overnight.

GPU:
Estimated points: 15965
My computer has 8.75 days to complete this work unit.
ETA: 11.83 days
I'm at about 3% after leaving it overnight.

If the ETA on the one on the GPU is longer than the time limit assigned, does that mean it won't complete on time? Then what? Any why did it give me a shorter time limit on that one when the GPU is so slow on it?

1) No. The client does not adjust it' settings based on whether the OS identifies the machine as in-use or idle. The only such adjustment is an option to suspend processing of one or more slots when the machine is in-use.

2) The idle condition is identified by your OS. That identification process is, in fact, used by screensavers and FAH merely detects the same condition. (It's not necessary to actually have a screensaver active, other than to test whether the OS is identifying the right condition.)

3) If your system is configured for two slots, such as one for GPU processing and one for CPU processing, they will process independently of each other so running 2 WUs is expected.

Deadlines do vary, based on the complexity of the assignment so you're probably OK in both cases. You can look up the project numbers here to confirm the deadlines for each project. That's also given in the information you quoted. desn't look good HOWEVER, the ETA is is often unreliable unless you've been processing continuously for quite a while. FAH needs to establish enough of a history to estimate how much progress you're machine is making.

(not asked) -- (1A/3A) It's possible to enable FAH's "idle" processing independently for each slot. If you're seeing screen-lag where FAH interferes with foreground processing, I recommend suspending GPU processing but leave CPU processing active even when you're using the machine. The OS does an excellent job of running the CPU processing almost continuously by interrupting FAH's processing very briefly every time you move the mouse or type a character and resuming processing immediately until your next action. Unfortunately such rapid interruption/resumption doesn't work as well on GPUs.

The GPU is Geforce GT 610, so it's about a 6 year old card.

CPU is i7-4770. (8 logical cores)

JimboPalmer, if I let it run on any setting that uses the GPU while using the PC, then it lags everything even my mouse movement. I wasn't sure that was due to the GPU until I disabled GPU folding. BUT, if I disable the GPU folding, then it's fine, I can run on "Medium" or even "Full" and the computer is still responsive. My CPU is maxed, but Windows is handling it.

I've had both the CPU and GPU folding over two nights now and the GPU isn't getting anywhere.

I think perhaps I should just remove the GPU slot (slot 1) and only use slot 0 (CPU). Do I lose any points for not finishing the WU?

You are probably best off removing the GPU slot. Your GT 610 is a rebranded GT 520, about the bare minimum GPU for processing currently available WU's would be a GT 630. That can depend on which version of the GT 630, cards based on both Fermi and Kepler GPU's have been released under that designation.

I would pause the GPU, then delete the Slot. That should let Stanford know, so they can re assign the Work Unit.

javaxnerd wrote:I've had both the CPU and GPU folding over two nights now and the GPU isn't getting anywhere.

It will make zero progress if you only run it for a short time. Every time you interrupt it, it will restart from the previous checkpoint which may, in fact, replace small increments of progress with zero progress. Moreover, the PPD cannot be estimated until several frames have been completed.

This neither supports or denies whether your GPU is too slow, only that you don't have enough information to make that determination.

If you do, in fact, remove the GPU slot, screen lag should vanish, which is a good thing. Also, your CPU will be able to use all 8 threads to process the CPU assignment which will improve the throughput for that slot slightly.

javaxnerd wrote: if I let it run on any setting that uses the GPU while using the PC, then it lags everything even my mouse movement. I wasn't sure that was due to the GPU until I disabled GPU folding. BUT, if I disable the GPU folding, then it's fine, I can run on "Medium" or even "Full" and the computer is still responsive. My CPU is maxed, but Windows is handling it.

Computer science has been working on multitasking for CPUs for over 60 years and is getting really good at it. F@H is set to be a lowest priority task and should gracefully get out of the way of your other uses for the PC.

GPUs have almost no multitasking planned or executed, so F@H is competing with all other uses of that hardware and it shows.

I just noticed that my new work unit says "Collection Server: 0.0.0.0", isn't that localhost? What does the collection server mean?

Collection server is a pointer to where the finished results are to be reported. An interesting post about 0.0.0.0 can be found in this thread: viewtopic.php?f=18&t=30590&start=15