GPU not being assigned WU, no appropriate WS available

[Joe_H]

(These posts have been split from a topic that was about problems getting uniprocessor assignments.)

Getting the same message for my GPU3 client for the last day and a half.

Which version of the F@H client are you using, and what model GPU is in your system? Could you post which WS's or projects your GPU client usually got assignments from in the recent past.

[Fadamor]

Sure.
O/S: Win 7 SP1 x64
FAH (GPU): Windows Client ver. 6.41r2 (Oct 21, 2010)
GFX Card: Nvidia GTX 580
GFX Drivers: Ver 310.90 (1/5/2013)

WU's worked on based on the log (opened Feb 7) are multiple Project 8072 WU's (50000 steps). A few weeks back were more intense projects that took twice as long to complete, but were only 40000 steps. I'm not sure where I could get a full list of my completed WU's - or at least a list of the ones completed in the last few months.

[Joe_H]

Shouldn't need to go back too far, at most the previous log file. The Project 8072 info at least gives a WS to look at, 171.67.108.36. It is up and appears to be sending and receiving WU's. May have to get some more reports to see if there is a pattern, your 580 should be getting assignment from different servers than warbeast's 8400GS.

[Fadamor]

Also, 8073 WU's were run a few times. The majority seem to be 8072 WU's from the previous log.

Here's my queue dump (slot 6 is the one trying to get filled now, I guess). As you can see, I haven't received any new work since early Saturday morning.

EDIT: Actually, if the queues are supposed to be continually filled, I haven't been getting new work since Thursday the 7th.

Code: Select all

[20:30:32] Printing Queue Information
Current Queue: 
Slot 06  Fetching 

Slot 07  Empty/Deleted
Project: 8072 (Run 0, Clone 1053, Gen 64), Core: 15
Work server: 171.67.108.36:8080
Collection server: 171.65.103.160
Download date: February 6 23:24:24
Finished date: February 7 02:35:18

Slot 08  Empty/Deleted
Project: 8072 (Run 0, Clone 1053, Gen 65), Core: 15
Work server: 171.67.108.36:8080
Collection server: 171.65.103.160
Download date: February 7 02:35:25
Finished date: February 7 14:50:43

Slot 09  Empty/Deleted
Project: 8072 (Run 0, Clone 1053, Gen 66), Core: 15
Work server: 171.67.108.36:8080
Collection server: 171.65.103.160
Download date: February 7 14:50:50
Finished date: February 8 06:35:44

Slot 00  Empty/Deleted
Project: 8073 (Run 0, Clone 2937, Gen 37), Core: 15
Work server: 171.67.108.36:8080
Collection server: 171.65.103.160
Download date: February 8 06:35:51
Finished date: February 8 09:47:58

Slot 01  Empty/Deleted
Project: 8073 (Run 0, Clone 2673, Gen 42), Core: 15
Work server: 171.67.108.36:8080
Collection server: 171.65.103.160
Download date: February 8 09:48:05
Finished date: February 8 12:58:19

Slot 02  Empty/Deleted
Project: 8073 (Run 0, Clone 1642, Gen 79), Core: 15
Work server: 171.67.108.36:8080
Collection server: 171.65.103.160
Download date: February 8 12:58:26
Finished date: February 8 16:08:43

Slot 03  Empty/Deleted
Project: 8073 (Run 0, Clone 1642, Gen 80), Core: 15
Work server: 171.67.108.36:8080
Collection server: 171.65.103.160
Download date: February 8 16:08:50
Finished date: February 8 19:19:12

Slot 04  Empty/Deleted
Project: 8072 (Run 0, Clone 2267, Gen 53), Core: 15
Work server: 171.67.108.36:8080
Collection server: 171.65.103.160
Download date: February 8 19:19:19
Finished date: February 9 04:33:03

Slot 05 *Empty/Deleted
Project: 8072 (Run 0, Clone 131, Gen 83), Core: 15
Work server: 171.67.108.36:8080
Collection server: 171.65.103.160
Download date: February 9 04:33:10
Finished date: February 9 07:44:27

PF: 0.983673 based on last 4 slot(s)

[bruce]

There are 3 types of GPU assignments. There's the OpenCL assignments for AMD, there's the Fermi/Kepler assignments for NV, and there's the NV assignments for older GPUs such as G80/Tesla (aka GPU2). There were a couple of somewhat more primitive NV GPUs that were still named as part of the 8000 series but didn't have all of the necessary features of the 8800+ Your 8400GS might be one of those but I can't lay hands on the exact list of the GPUs that had to be excluded due to high WU failure rates.

I notice that the 8400 GS is whitelisted for V7, but it's not really clear if updating your client would be helpful.

[stickman]

I'm having the same error on my Ubuntu 12.10 machine since I set it up on Feb 8th. I have a pair of 8800GTS GPUs in that machine. Is there just a dearth of G80 WU's available?

I just tweaked up my verbosity, will see if there's anything in it when my system checks for new WU's.

18:22:38: Options: -std=gnu++98 -O3 -funroll-loops -mfpmath=sse -ffast-math
18:22:38: -fno-unsafe-math-optimizations -msse2
18:22:38: Platform: linux2 3.2.0-1-amd64
18:22:38: Bits: 64
18:22:38: Mode: Release
18:22:38:******************************* System ********************************
18:22:38: CPU: AMD Athlon(tm) X2 Dual Core Processor BE-2400
18:22:38: CPU ID: AuthenticAMD Family 15 Model 107 Stepping 2
18:22:38: CPUs: 2
18:22:38: Memory: 3.86GiB
18:22:38:Free Memory: 2.06GiB
18:22:38: Threads: POSIX_THREADS
18:22:38: On Battery: false
18:22:38: UTC offset: -5
18:22:38: PID: 2979
18:22:38: CWD: /var/lib/fahclient
18:22:38: OS: Linux 3.5.0-17-generic x86_64
18:22:38: OS Arch: AMD64
18:22:38: GPUs: 2
18:22:38: GPU 0: NVIDIA:1 G80 [GeForce 8800 GTS]
18:22:38: GPU 1: NVIDIA:1 G80 [GeForce 8800 GTS]
18:22:38: CUDA: Not detected
18:22:38:***********************************************************************

[PantherX]

I have the same problem to GPU:0:G86 [GeForce 8400GS]

20:02:55:WARNING:WU03:FS02:Failed to get assignment from 'assign-GPU.stanford.edu:8080': Empty work server assignment
20:02:55:ERROR:WU03:FS02:Exception: Could not get an assignment

Please respond.

Welcome to the F@H Forum warbeast,

Can you please post the initial section of your log so we can see what version of V7 you are using and how it is configured.

[PantherX]

Sure.
O/S: Win 7 SP1 x64
FAH (GPU): Windows Client ver. 6.41r2 (Oct 21, 2010)
GFX Card: Nvidia GTX 580
GFX Drivers: Ver 310.90 (1/5/2013)...

Is there a reason that you are still using the GPU3 Client? If not, I would suggest that you try out V7.2.9 or the latest beta, V7.3.4, depending on your preference.

[PantherX]

I'm having the same error on my Ubuntu 12.10 machine since I set it up on Feb 8th. I have a pair of 8800GTS GPUs in that machine. Is there just a dearth of G80 WU's available?

I just tweaked up my verbosity, will see if there's anything in it when my system checks for new WU's.

Welcome to the F@H Forum stickman,

Can you please post a section of your log? Please note that since you are running an unofficial configuration, support might be limited. Do note that my GTX 260 is receiving plenty of FahCore_11 WUs on V7.2.9 so a server issue is less likely.

[Fadamor]

Is there a reason that you are still using the GPU3 Client? If not, I would suggest that you try out V7.2.9 or the latest beta, V7.3.4, depending on your preference.

Didn't know there WAS a 7.x GPU client. What's the difference between the two versions?

[PantherX]

Key points are as follows:
Integration of SMP, GPU clients into a single installer
Better GPU detection method
A GUI interface to control all Slots (each slot can be considered as a separate v6 Client)
Access to future FahCore and WUs

For a more detailed change log, you may want to check out the documentation and links here

[P5-133XL]

Is there a reason that you are still using the GPU3 Client? If not, I would suggest that you try out V7.2.9 or the latest beta, V7.3.4, depending on your preference.

Didn't know there WAS a 7.x GPU client. What's the difference between the two versions?

7.2.9 is an officially released version while 7.3.4 is a new beta. It defaults to fully using the machine which may produce some lag for the user.
7.3.4 adds a web control that uses a single slider to simplify folding for novices. It defaults to a lower folding performance level that shouldn't be noticeable to the user. It adds a folding screensaver.

[Fadamor]

I was running 7.2.9 for my SMP client. I just added a GPU client (after stopping the GPU3 client) and it brought down an 8072 WU. Looks like that solved the issue.

EDIT: Yep. already 2.7% complete on the WU.

[bruce]

I was running 7.2.9 for my SMP client. I just added a GPU client (after stopping the GPU3 client) and it brought down an 8072 WU. Looks like that solved the issue.

EDIT: Yep. already 2.7% complete on the WU.

The default installation of V7.2.9 (and earlier) should have installed both a SMP and a GPU slot if your hardware permitted such installations. Did you manually disable/remove the GPU slot at some point?

[7im]

I'm having the same error on my Ubuntu 12.10 machine since I set it up on Feb 8th. I have a pair of 8800GTS GPUs in that machine. Is there just a dearth of G80 WU's available?

I just tweaked up my verbosity, will see if there's anything in it when my system checks for new WU's.

Welcome to the F@H Forum stickman,

Can you please post a section of your log? Please note that since you are running an unofficial configuration, support might be limited. Do note that my GTX 260 is receiving plenty of FahCore_11 WUs on V7.2.9 so a server issue is less likely.

No current support for running GPU slots in Linux, but a recent blog post said they are working on it.