A little help if you please

[bunimomike]

Hi guys.

Firstly to the whole F@H project. Congratulations on starting something like this. The more I read about it the more it impresses me. To everyone's efforts... wow!

A quick summary so you know what's potentially going wrong. I'm pretty new to F@H and have been at it a month. I have CPU Folding on my old PC and also on my new works machine. I also have GPU Folding installed on the latter and it runs overnight (not at the same time as CPU Folding ). Also, I have my PS3 chipping in daily too (sometimes more). As you can see, I've been turning around quite the number of WUs but something's happening as of recent. The last 40 hours or so has seen my total WUs only go up by one or two. My PS3 should be cranking out a WU every eight hours. My quad-core work PC was happy doing a CPU WU every 24 hours (quarter load across the cores) but the GPU was motoring. I remember seeing it crank out 6 WUs overnight.

So what's happening? Am I being given massive WUs as I'm giving a lot of resources over or is something wrong at F@M's end? I say this as last night I checked my PS3 and even that said "60% completed, 10 hours left" which is bonkers seeing as each WU is eight hours.

Let me apologise if any or all of the above is confusing/smacks of a noobie. I just want to make sure my WUs are getting done, being recorded and ultimately helping F@H.

Thanks in advance,

Mike.

[P5-133XL]

How about adding your FAHLog.txt files for everything you would like people to comment on.

[Christopher N. Lewis]

Hi Mike,

Welcome to the fold. I hope this helps you understand what is happening.

There are different projects for both the CPU and GPU clients. On the same hardware these can take different amounts of time, for example I can complete some GPU2 units within a couple of hours, but others can take up 18 hours. The different units give different points on completion. I'm not sure if the same is true of the PS3, suggest you check the FAQ and/or Wiki.

Are you monitoring your progress with FahMon (see 3rd party software)? Recomended to do so, as it makes comparisons and explanations much easier. This will show you the time remaining, and the project number, also run, clone and generation. You can then lookup the project to learn more about it. FahMon also gives an estimate of the time remaining and points per day (PPD; caution, can be calculated in different ways).

It is up to you to choose the clients that best fit your hardware and the way you use it. Once that is done, you can not do anything to change the WUs that you are assigned. Stanford does not change the downloads for different hardware (there is no feedback), their priorities are with the science and they just queue up the WU on their servers. Because of this the WUs sometimes come and go, at the moment there are lots of 787 point GPU units, but sometimes there are lots of 1888 pointers which take significantly longer to complete, or the dreaded 511 pointers which are slow and hot.

Advice for the future - It helps other members to usefully reply if you give your hardware, OS, client and Fahlog.txt (within the

Code: Select all

tags) when you post.

CPU and GPU folding can be run together with a little care taken during configuration

[bunimomike]

Hi guys.

Ah, right. First time I've had to have a nose at the logfile. One sec... *goes to get it*. This is from the GPU client. I'm on XP Pro with a quad-core Intel of suitable speed (close to 3GHz I think). More importantly the graphics card is an ATI 4870 with 1GB onboard. Like I said, it ate up the first handful of WUs. However, the FAQ on the PS3 mentions 8 hour units as it's intended for people to start up when they go to bed and be done when they're awake the next morning. Lovely idea and it's what's been happening all month until now. I'll go hand-type the details of my current PS3 WU in a moment.

Apologies if I'm being extra-daft but still very wet behind the ears on all this. Also, I'll go have a look at that 3rd party software right now. *tips hat* Thank you.

Code: Select all

--- Opening Log file [September 29 06:50:42 UTC] 


# Windows GPU Console Edition #################################################
###############################################################################

                       Folding@Home Client Version 6.23

                          http://folding.stanford.edu

###############################################################################
###############################################################################

Launch directory: C:\Documents and Settings\MikeW\Application Data\Folding@home-gpu


[06:50:42] - Ask before connecting: No
[06:50:42] - User name: bunimomike (Team 78731)
[06:50:42] - User ID: 732DB3266557C123
[06:50:42] - Machine ID: 4
[06:50:42] 
[06:50:42] Loaded queue successfully.
[06:50:43] Initialization complete
[06:50:43] 
[06:50:43] + Processing work unit
[06:50:43] Core required: FahCore_11.exe
[06:50:43] Core found.
[06:50:43] Working on queue slot 05 [September 29 06:50:43 UTC]
[06:50:43] + Working ...
[06:50:51] 
[06:50:51] *------------------------------*
[06:50:51] Folding@Home GPU Core - Beta
[06:50:51] Version 1.24 (Mon Feb 9 11:00:12 PST 2009)
[06:50:51] 
[06:50:51] Compiler  : Microsoft (R) 32-bit C/C++ Optimizing Compiler Version 14.00.50727.762 for 80x86 
[06:50:51] Build host: amoeba
[06:50:51] Board Type: AMD
[06:50:51] Core      : 
[06:50:51] Preparing to commence simulation
[06:50:51] - Looking at optimizations...
[06:50:51] - Files status OK
[06:50:51] - Expanded 96577 -> 489152 (decompressed 506.4 percent)
[06:50:51] Called DecompressByteArray: compressed_data_size=96577 data_size=489152, decompressed_data_size=489152 diff=0
[06:50:51] - Digital signature verified
[06:50:52] 
[06:50:52] Project: 5737 (Run 2, Clone 41, Gen 378)
[06:50:52] 
[06:50:52] Assembly optimizations on if available.
[06:50:52] Entering M.D.
[06:50:59] Will resume from checkpoint file
[06:50:59] Tpr hash work/wudata_05.tpr:  3750964994 2657599421 1849165809 2761105001 1363395191
[06:51:09] Working on Protein
[06:51:09] Client config found, loading data.
[06:51:09] Starting GUI Server
[06:51:14] Resuming from checkpoint
[06:51:14] fcCheckPointResume: retreived and current tpr file hash:
[06:51:14]    0   3750964994   3750964994
[06:51:14]    1   2657599421   2657599421
[06:51:14]    2   1849165809   1849165809
[06:51:14]    3   2761105001   2761105001
[06:51:14]    4   1363395191   1363395191
[06:51:14] Verified work/wudata_05.log
[06:51:15] Verified work/wudata_05.edr
[06:51:15] Verified work/wudata_05.xtc
[06:51:15] Completed 6%

Folding@Home Client Shutdown.
[06:51:22] Got kill signal -- issuing INTERRUPTED core shutdown

--- Opening Log file [September 29 12:58:52 UTC] 


# Windows GPU Console Edition #################################################
###############################################################################

                       Folding@Home Client Version 6.23

                          http://folding.stanford.edu

###############################################################################
###############################################################################

Folding@home-gpu


[12:58:52] - Ask before connecting: No
[12:58:52] - User name: 
[12:58:52] - User ID: 
[12:58:52] - Machine ID: 4
[12:58:52] 
[12:58:52] Loaded queue successfully.
[12:58:52] Initialization complete
[12:58:52] 
[12:58:52] + Processing work unit
[12:58:52] Core required: FahCore_11.exe
[12:58:52] Core found.
[12:58:52] Working on queue slot 05 [September 29 12:58:52 UTC]
[12:58:52] + Working ...
[12:58:52] 
[12:58:52] *------------------------------*
[12:58:52] Folding@Home GPU Core - Beta
[12:58:52] Version 1.24 (Mon Feb 9 11:00:12 PST 2009)
[12:58:52] 
[12:58:52] Compiler  : Microsoft (R) 32-bit C/C++ Optimizing Compiler Version 14.00.50727.762 for 80x86 
[12:58:52] Build host: amoeba
[12:58:52] Board Type: AMD
[12:58:52] Core      : 
[12:58:52] Preparing to commence simulation
[12:58:52] - Looking at optimizations...
[12:58:52] - Files status OK
[12:58:52] - Expanded 96577 -> 489152 (decompressed 506.4 percent)
[12:58:52] Called DecompressByteArray: compressed_data_size=96577 data_size=489152, decompressed_data_size=489152 diff=0
[12:58:52] - Digital signature verified
[12:58:52] 
[12:58:52] Project: 5737 (Run 2, Clone 41, Gen 378)
[12:58:52] 
[12:58:52] Assembly optimizations on if available.
[12:58:52] Entering M.D.
[12:58:58] Will resume from checkpoint file
[12:58:58] Tpr hash work/wudata_05.tpr:  3750964994 2657599421 1849165809 2761105001 1363395191
[12:58:59] Working on Protein
[12:58:59] Client config found, loading data.
[12:58:59] Starting GUI Server
[12:59:03] Resuming from checkpoint
[12:59:03] fcCheckPointResume: retreived and current tpr file hash:
[12:59:03]    0   3750964994   3750964994
[12:59:03]    1   2657599421   2657599421
[12:59:03]    2   1849165809   1849165809
[12:59:03]    3   2761105001   2761105001
[12:59:03]    4   1363395191   1363395191
[12:59:03] Verified work/wudata_05.log
[12:59:03] Verified work/wudata_05.edr
[12:59:03] Verified work/wudata_05.xtc
[12:59:03] Completed 6%
[12:59:21] Opening http://fah-web.stanford.edu/cgi-bin/main.py?qtype=userpage&username=bunimomike...
[13:03:20] Completed 7%
[13:07:37] Completed 8%
[13:11:52] Completed 9%
[13:16:09] Completed 10%
[13:20:26] Completed 11%
[13:24:42] Completed 12%
[13:28:59] Completed 13%
[13:33:15] Completed 14%
[13:37:31] Completed 15%
[13:41:48] Completed 16%
[13:46:04] Completed 17%
[13:50:21] Completed 18%
[13:54:38] Completed 19%
[13:58:53] Completed 20%
[14:03:10] Completed 21%
[14:07:27] Completed 22%
[14:11:43] Completed 23%
[14:16:00] Completed 24%
[14:20:20] Completed 25%
[14:24:40] Completed 26%
[14:28:59] Completed 27%
[14:33:16] Completed 28%
[14:37:45] Completed 29%
[14:42:08] Completed 30%
[14:43:09] Opening http://fah-web.stanford.edu/cgi-bin/main.py?qtype=userpage&username=bunimomike...
[14:43:25] Opening http://fah-web.stanford.edu/cgi-bin/main.py?qtype=userpage&username=bunimomike...

[bunimomike]

PS3 work unit
CURRENT WORK UNIT
Name: p4044_supervillin_StillGBSA_AM03
Core: SCEARD2 1.40.93520
Progress 91.76%
0.1458s/frame 118.50ns/day
Time to Completion: 3hours and 20 minutes left
Estimated End: My current time plus the above.

Truly, a massive unit. Weird.

Edit: Have FahMon installed. If you want me to grab some information from it, feel free to ask.

[kiore]

http://fah-web.stanford.edu/cgi-bin/fahproject?p=5737
511 points, smallish work unit but a notoriously slow series.
Units do take varying times depending on hardware. Fahmon will give you a better idea of how they are going.

PS3 unit: http://fah-web.stanford.edu/cgi-bin/fahproject?p=4044
Yes this one is an extra big one 1250 points should take 1.5 days you have 20 days till preferred deadline.
kiore.

[jrweiss]

For the GPU client, have you looked at the Environment Variables discussions and set them in your machine? If not, you may be using excessive CPU time (20-50% of a core) and not enough GPU time (60% instead of 98%). Check your Catalyst Control Center and/or use GPUz to verify GPU load and temperature.

[bunimomike]

Thanks guys. Much appreciated on the clueing-up. So I get the "points" but is each one a single Work Unit as well? If not, no probs... like I said... Just trying to suss out what's what. As always, thank you in advance for information.

Oh, by the way, I've tweaked my ATI settings so they're running under-spec. Not a huge amount but enough to keep the fan pretty-damned quiet. Will pop over into the Environment Variables discussions to see how I can alter things that bit further (remove the load from the quad-core itself when running GPU).

[Zagen30]

All the clients work by downloading one WU at a time and processing it. There are a certain number of points assigned to each WU based on how long it takes to complete on a benchmark machine; all of the WUs within a certain project get the same number of points, as they should theoretically complete in the same time as all the others in that project (minor variances do occur and are normal). When you finish a WU and upload it, Stanford credits the appropriate number of points to your username/team#.

[kiore]

Different workunits are from different projects and have different 'point' value which is an aproximation of 'science done'. Look up at the top of the page and go to project summary where they are listed by number. Yes they can be very different in size, just a few days ago I did a cpu workunit worth 47 points (very fast) now am working on one worth 2137 points, also fast but should me take 3 days as is so big.
The PS3 ones are normally around 250 points you got one that was like 5 joined together.
Have fun.
kiore.

[bunimomike]

Ah, right. Got it. Final question (for the night, over here in the UK)... is there a dedicated thread/area to Environmental Variables discussion? I'm still finding my feet with the forum and Folding@Home, which is making things a bit tricky for finding stuff.

The PS3 ones are normally around 250 points you got one that was like 5 joined together.

Crikey guv'nor!

[kiore]

Ah, right. Got it. Final question (for the night, over here in the UK)... is there a dedicated thread/area to Environmental Variables discussion? I'm still finding my feet with the forum and Folding@Home, which is making things a bit tricky for finding stuff.

That would be in the ATI forum area Not somewhere I venture very often.
http://foldingforum.org/viewforum.php?f=51
kiore.

[bunimomike]

Evening guys (technically good morning as it's just gone midnight here). Thanks to everyone for the help. I think (although the chances must be slim) that I was stuck with multiple mahoosive Work Units. Everything's ticking over nicely, now. 2 x PC, 1 x PS3.

You're polite explanations have been most helpful. *tips hat*

[codysluder]

You'll find many discussions mentioning PPD (Points Per Day). In the case of the PS3 you should earn a very, very uniform number of PPD. If you get a WU that earns 5x as many points, it's also one that run 5x as long. The numbers are uniform because PS3's all fold at pretty much the exact same speed as the benchmark machine.

In the case of either the gpu or the cpu client, there will be more variation between projects. They'll all still be in the same general range, but different projects often use the various components in your pc in differnt proportions. Your pc might have a slightly faster cpu and slightly slower RAM so while one project mght match the speed of the benchmark machine, the next project might depend more on RAM speed and less on cpu speed and run (slightly) slower or vice-versa. (That's not necessary a good example, because floating point speed seems to always be the most important factor, but it's just an example.)