noob user, never connects to work server?

[mushu999]

Windows XP/32-bit running ATI card with latest drivers, nod32 *disabled completely*, no firewall, router set to port-forward TCP 8080 to my machine. Had it run overnight and log file just repeats last section over and over about no appropriate work server available. So I decided to let the Display run all night and day and this is what I ended up with:
(open image into new browser tab to see entire thing.) Notice how the progress bar has run clear over and out of its containing outline?! Plus it used to say Time Left was only a few hours but now it has that wild number of days.

Systray icon tooltip always says "Attempting to get work packet", here is log file:

Code: Select all

--- Opening Log file [July 25 04:48:19 UTC] 
# Windows GPU Systray Edition #################################################
###############################################################################
                       Folding@Home Client Version 6.41r2
                          http://folding.stanford.edu
###############################################################################
###############################################################################
Launch directory: C:\Documents and Settings\zzz\Application Data\Folding@home-gpu
Arguments: -verbosity 9 

[04:48:19] - Ask before connecting: No
[04:48:19] - User name: Mushu999 (Team 69411)
[04:48:19] - User ID: zzz
[04:48:19] - Machine ID: 2
[04:48:19] 
[04:48:19] Gpu type=1 species=4.
[04:48:19] Loaded queue successfully.
[04:48:19] Initialization complete
[04:48:19] - Preparing to get new work unit...
[04:48:19] Cleaning up work directory
[04:48:19] - Autosending finished units... [July 25 04:48:19 UTC]
[04:48:19] Trying to send all finished work units
[04:48:19] + No unsent completed units remaining.
[04:48:19] - Autosend completed
[04:48:19] + Attempting to get work packet
[04:48:19] - Will indicate memory of 3062 MB
[04:48:19] Gpu type=1 species=4.
[04:48:19] - Detect CPU. Vendor: GenuineIntel, Family: 6, Model: 10, Stepping: 7
[04:48:19] - Connecting to assignment server
[04:48:19] Connecting to http://assign-GPU.stanford.edu:8080/
[04:48:19] Posted data.
[04:48:19] Initial: 0000; + No appropriate work server was available; will try again in a bit.
[04:48:19] + Couldn't get work instructions.
[04:48:19] - Attempt #1  to get work failed, and no other work to do.
Waiting before retry.
[04:48:28] + Attempting to get work packet
[04:48:28] - Will indicate memory of 3062 MB
[04:48:28] Gpu type=1 species=4.
[04:48:28] - Connecting to assignment server
[04:48:28] Connecting to http://assign-GPU.stanford.edu:8080/
[04:48:28] Posted data.
[04:48:28] Initial: 0000; + No appropriate work server was available; will try again in a bit.
[04:48:28] + Couldn't get work instructions.
[04:48:28] - Attempt #2  to get work failed, and no other work to do.
Waiting before retry.

So, any ideas on how I can get this software to work properly? Do I need to open or forward any other ports or protocols on the router? Help!

[7im]

Hello mushu999, welcome to the forum.

You didn't list the model of your ATI card, but if it is less than a 5xxx series GPU (4xxx or below), the End of Life for that card in regards to FAH was announced a long time ago. The supply of work units for these older ATI cards has slowly been drying up.

[Zagen30]

What GPU is this, specifically? I believe at this point that all work for pre-5000 series AMD/ATI cards has ended, so if you're running a 4000/3000/2000 card you can't contribute with it. If you have a 5000 series or newer card, AMD removed OpenCL support for XP starting with driver version 11.11; you'd need to roll back to 11.10 to fold on it, assuming that your GPU was supported by 11.10. If it wasn't supported when 11.10 came out, I think you're out of luck and would need to upgrade to Vista/7 to fold on it.

[P5-133XL]

Do not use the viewer supplied in v6 clients. It is quite buggy and uses enough resources that it will significantly impact your folding productivity. If you need a viewer, try v7 of the client.

[mushu999]

Oops sorry! It's a Radeon 5770HD and yeah, Catalyst version is 12.6 but how would that affect the network connection issue I'm having? I'd assume it would affect the display and perhaps any calculations being done on the GPU itself, but network...?

I'll see about finding the older driver, but that sucks for playing LOTR and WoW and games in general...

[bruce]

Oops sorry! It's a Radeon 5770HD and yeah, Catalyst version is 12.6 but how would that affect the network connection issue I'm having? I'd assume it would affect the display and perhaps any calculations being done on the GPU itself, but network...?

I'll see about finding the older driver, but that sucks for playing LOTR and WoW and games in general...

Read the messages carefully.

Code: Select all

[04:48:19] - Connecting to assignment server
[04:48:19] Connecting to http://assign-GPU.stanford.edu:8080/
[04:48:19] Posted data.
[04:48:19] Initial: 0000; + No appropriate work server was available; will try again in a bit.

You have successfully connected to the assignment server (so you don't have a network problem). The Assignment Server sees that you do not have OpenCL so it looks for a server that has WUs that don't require it -- and no Work Server has any of that type of WUs. Once your device is recognized as an OpenCL capable device, it will be able to find a work server that has appropriate work to give you.

On another subject, are you also running a CPU or SMP client? The AMD GPUs put an appreciable load on one of your CPU cores and this may require special adjustments.

[Zagen30]

One thing to note is that you need to upgrade to v7 to run the OpenCL AMD core, as it's unsupported in v6. v6 can only get the non-OpenCL work (it ran on OpenCL-capable GPUs just fine), but it appears there's none left.

[Joe_H]

One thing to note is that you need to upgrade to v7 to run the OpenCL AMD core, as it's unsupported in v6. v6 can only get the non-OpenCL work (it ran on OpenCL-capable GPUs just fine), but it appears there's none left.

That is correct, the one remaining server with non-OpenCL work for ATI cards is reading the following:

09:30:00 PDT 2012 171.64.65.103 vspg3v2 densign GPU full Accepting 1.30 0 0 338 17883 5414 0 99 99 1

And has been that way for over a week or two. The 99/99 counts imply there might be just a few WU's left to process, but not enough to meet the threshold before the assignment and work servers will send them out.

[mushu999]

On another subject, are you also running a CPU or SMP client? The AMD GPUs put an appreciable load on one of your CPU cores and this may require special adjustments.

Nope, only the GPU client as shown by the tagline in my logfile.
. . .installs new client. . .
Wow, installing the v7 client made everything work, it's like magic! :) It's now doing the CPU/SMP thing, and it maxes out all four of my processor cores so I think the lack of OpenCL in my vid driver means it won't use the GPU, and after staring at the log file a while it seems that is the case. It grabs and runs using SMP but the GPU gives an error each attempt. So now I assume I can either set it to run SMP (CPU) only or try to locate an older vid driver to add OpenCL capability. Since I leave my system on 24/7 and I'm at work all day and sleeping at night there is only a short window of time when I will be playing games so perhaps the SMP option is good enough...
Thanks everyone for the help, it was actually helpful.

*edit: backspace over CUDA and add OpenCL.

[bruce]

CUDA is for NVidia GPUs. OpenCL is for AMD/ATI GPUs.

[mushu999]

(edited previous message.)
New question, why does the v7 viewer now say Protein: Demo
and doesn't list the actual name of the protein the control window says it is working on? Is the image a "fake" one also, or is it the real thing I'm working on? Here is the image:

Briefly, when I hit the [fold] button to switch to the other process, in the viewer I see Protein: Loading but it never gives the full name and goes back to saying Demo. :( No biggie I guess, and might be a bug also, I was just curious is all.

[DoctorsSon]

It is just a demo screen.
The FAHControl panel will give you better information.

[7im]

It is not just a demo screen. It's a fully functional protein viewer when the particular work unit uses a fahcore that outputs the correct image format. Fahcore_78 works great with the viewer. However, the recent crop of SMP and GPU fahcores need an enhancement to be able to output to the viewer. And we have bug tickets open to add those enhancements.

Demo is simply the name of the default protein shown when the actual protein cannot be shown. As we've seen with the PS3's viewer, when the viewer works, it's pretty spectacular.

[bruce]

In other words, don't bother with running the viewer in its present form. It's just windowdressing anyway -- not needed for either science or points.