dont get points from SMP or GPU

[Rel25917]

Browsers these days do use the gpu for rendering pages faster but have never heard of them interfering with FAH.

[bollix47]

Hardware acceleration is turned on by default in some versions of Chrome.

If you want to experiment by ensuring that hardware acceleration is turned off then try:

Turning Off GPU Acceleration In Google Chrome

[CaqKa]

with chrome now shut down, the gpu folding slot came right now to 1%.
lets see what will happen withou chrome.

[7im]

Failed to read stream is a networking problem. Something on your network is blocking or failing to resolve the name/IP of the Stanford servers.

I get the same thing in a test lab at work that is sectioned off from the corporate metwork by a locked down router.

[P5-133XL]

Games, browsers and specifically flash now commonly use the GPU. There are settings that claim to seable GPU use but I've found no set of settings that actually works. So I now commonly have to pause the client when watching a video to prevent stuttering. Normal browsing hasn't been a problem as long as there isn't a video on the page. If I do see any problems, pausing and unpausing GPU folding hasn't been a problem. Note, you only need to pause the primary GPU (one slot) by Rt. clicking on that one slot and selecting pause.

While theoretically those applications could affect folding in general I haven't found anything other than a slowdown occurring. Folding while gaming is really bad where I have to pause GPU folding else I will get a video reset at some point during the game and those really are to be avoided. The problem is not that they damage the folding results but rather they reset the video clocks to a slow level that is only solvable by rebooting the machine

[CaqKa]

in cca 7hours i will see whats happening. chrome and utorrent not running, so far the gpu slot is on 42%, and the 2 uni processor also doing fine.

i already managed to see that i can pause slots individualy.
if every thing will be fine, ill be testing with gpu pauses and chrome running, and also disabling gpu in chrome.
i dont know if it is placebo, but somehow i see more crashing bars of schockwawe flash in chrome lately. :>

[bollix47]

Flash also has a hardware acceleration setting. If you go to a site with flash you can right click on the flash video and select settings. On the bottom of the window that pops up you will see a number of icons and the one on the bottom left looks like a monitor. Click on that and make sure there isn't a checkmark in the Enable hardware acceleration box.

[P5-133XL]

utorrent should not be a problem except when sending and receiving WU's over the internet. Depending upon how much BW you use it can kill WU's by simply timing out the sending or receiving of them. That may be recoverable but occasionally the v7 client just gets permanently stuck in a sending mode that is only solvable by killing the WU that is stuck. Really you need to supply folding enough BW to do these operations in a timely fashion by reserving BW from your torrent client. The second best solution involves properly configuring a QOS system, typically in router that supports it.

[codysluder]

While theoretically those applications could affect folding in general I haven't found anything other than a slowdown occurring. Folding while gaming is really bad where I have to pause GPU folding else I will get a video reset at some point during the game and those really are to be avoided. The problem is not that they damage the folding results but rather they reset the video clocks to a slow level that is only solvable by rebooting the machine

One of the very early versions of flash player that used gpu acceleration did, in fact, conflict with FAH. They fixed it in a week or two so that's not today's problem, but it does demonstrate that there's a possiblity of a programmer testing something that runs fine by itself and later finding that there's a conflict.

[CaqKa]

so the gpu wu that yesterda i was talking about has finished. http://folding.extremeoverclocking.com/ ... s=&u=35457
it also got thru a pause, so i am assuming chrome gpu process is conflicting with fah gpu.
will do another test with disabling gpu functions in chrome (hope it will not slow down the flash videos, because the atom is not enough for this videos)

As of the network... i dont think that should be a big problem, i saw that the client uploads ~100 KB, thats 1 second of my upload. And my upload is not so heavily used.
how much times it tries before it is falling back on the collection server?

[Rel25917]

Just FYI all uploads will not be the same size, they vary greatly from project to project. I'm not sure about the UNI slots as i don't run any of those but i know even the small smp units are 1-2 MB in size, big ones can be 10 or 25+. BigAdv units are like 100+. My most recent gpu units from my gtx 560 ti do seem be on the smaller side at only about 500k. Just be aware that the size can change. See http://folding.stanford.edu/English/FAQ-Configure#ntoc4 You can limit the slots to smaller(but still up to 5MB) by adding an extra slot option for max-packet-size small normal or big (5mb)(10mb)(500mb).

example of slot option

[7im]

All SMP work units of the a3 fahcore are hardcoded to the BigWU size. Small, normal, and big settings have no affect on SMP clients. I know a5 and a6 are the same way, and I assume a4s are also.

GPU WUs are all small so far so it doesn't really apply there either.

[bruce]

i do not understand why is there a fallback on a server thats not working? shouldnt that be programmed so that it finds it self a working server?

I don't think this ever got answered.

The Collection Servers are only needed when there's a problem with the primary server . . . and problems with the primary Work Servers are generally repaired quite quickly, so not having a working CS has never been on the "critical" list. Also, there have been a number of upgrades to the server code which included fixes for the way the WS and the CS communicate with each other. Those changes REQUIRED changes to the CSs but that introduced two groups of servers, some which worked the old way and others which worked the new way. It's amazingly easy to postpone upgrading a server that's working just fine.

The statements made here by 7im are certainly applicable in relation to the state of the Collection Servers.

[CaqKa]

http://folding.extremeoverclocking.com/ ... s=&u=35457
when pausing fah before loading chrome, and shutting it down before resuming fah.
the 300+ points are GPU ones. around 8. september i changed the slot configuration of one SMP to two UNI. 15. sept, i started with pausing...

where should i report interference between chrome and fah? its chrome "21.0.1180.89 m" that i am currently running.