To test a little theory of mine - what OS is having these problems? Linux, Windows, OS X? Or all three? I'm Windows, and I see the downloads stalling sometimes - but the work is usually fully downloaded by the time I need it.

I am on Win7 64-bit. It is not an actual operational problem for me at the moment. For the past four days, even with one or two backoffs, I get the downloads in less than 20 minutes, and usually about 10 minutes. With a little overlap (buffer setting of 0.01 + 0.01 days), it is working OK for my GTX 960, though the problem to some degree is still there.

To test a little theory of mine - what OS is having these problems? Linux, Windows, OS X? Or all three? I'm Windows, and I see the downloads stalling sometimes - but the work is usually fully downloaded by the time I need it.

Both my XP and Win7 boxes are having the issue. Win7 is the most problematic because I'm running dual cards with 2 tasks each. Because of the 2 task per GPU limit I have no buffer to cover the stuck downloads. A 3 tasks per card limit would probably eliminate the issue but I doubt that will happen.

To test a little theory of mine - what OS is having these problems?

Windows 7 Pro here. It's a minor problem, usually, however I am not crunching at the scale some are.

---

Team USA forum | Team USA page
Join us and #crunchforcures. We are now also folding:join team ID 236370!

Hi nanoprobe and the gpugrid community,

I have forwarded this issue already several times to our local IT service but apparently they also have to forward it to some other higher IT service and I'm not sure to which extent we have power to prioritize our problem. I kindly ask you to be patient. I particularly share your frustration and I will try to keep insisting them to fix the problem.

I just noticed another phantom task that was assigned to me, but my BOINC client never got the scheduler's response:

02-Mar-2016 14:35:22 [GPUGRID] Sending scheduler request: Requested by project.
02-Mar-2016 14:35:22 [GPUGRID] Requesting new tasks for NVIDIA GPU
02-Mar-2016 14:40:27 [GPUGRID] Scheduler request failed: Timeout was reached
02-Mar-2016 14:40:27 [GPUGRID] Sending scheduler request: Requested by project.
02-Mar-2016 14:40:27 [GPUGRID] Requesting new tasks for NVIDIA GPU
02-Mar-2016 14:40:29 [GPUGRID] Scheduler request completed: got 1 new tasks

After the timeout, my BOINC client merrily requested once more for new tasks...

I wish there was a way to cancel tasks using the project's site, e.g. a Cancel button on the task list.

---

Intrigued by this post by Bjarke I decided to do some trace-routing for GPUGRID and my other projects (WCG and POEM). Here's the output from tracert, having appended the geographic location of each hop using http://www.ipligence.com/geolocation:

C:\Users\vagelis>tracert www.gpugrid.org

Tracing route to www.gpugrid.org [84.89.134.145]
over a maximum of 30 hops:

## Skipping trace of my own ISP ##

  8    77 ms    75 ms    76 ms  nl-sar.nordu.net [80.249.209.203] -- NETHERLANDS
  9    80 ms    80 ms    98 ms  uk-hex.nordu.net [109.105.102.97] -- SWEDEN
 10    96 ms   105 ms    95 ms  ndn-gw.mx1.lon.uk.geant.net [109.105.102.98] -- SWEDEN
 11    85 ms    98 ms    86 ms  ae0.mx1.par.fr.geant.net [62.40.98.77] -- UK
 12    81 ms    81 ms    81 ms  83.97.88.129 -- UK
 13   104 ms     *      105 ms  83.97.88.130 -- UK
 14   139 ms   139 ms   120 ms  TELMAD.AE4.uv.rt1.val.red.rediris.es [130.206.245.89] -- SPAIN - MADRID
 15   118 ms   120 ms   121 ms  anella-val1-router.red.rediris.es [130.206.211.70] -- SPAIN - MADRID
 16     *        *        *     Request timed out.
 17   126 ms   126 ms   126 ms  grosso.upf.edu [84.89.134.145] -- SPAIN - BARCELONA
 18   216 ms   126 ms   126 ms  grosso.upf.edu [84.89.134.145]
 19   118 ms   117 ms   120 ms  grosso.upf.edu [84.89.134.145]

Trace complete.

Note that I took traces from two locations using different ISPs to determine the entry point to GPUGRID's ISP network. The trace above is the common part.

Comparing GPUGRID's route trace to my other projects, it is evident that there's a lot of hopping around across Europe: Netherlands to Sweden to the UK to finally reach Spain. In contrast, WCG's trace shows a hop in the UK and then it goes to the USA. POEM's again has a hop in the UK and then goes to Germany.

Now, I'm not saying that hopping across Europe is a bad thing, even for an IP packet :D, but more hops does mean more points that can cause network problems.

It would be interesting to have a route trace from before the GPUGRID ISP switch to compare...

---

A tracert from eastern Pennsylvania seems simple enough. My guess is that it is a local problem near UPF.
Or maybe there is more than one problem? It seems to be worse for some people than others.

3 9 ms 13 ms 11 ms 207.172.196.203
4 11 ms 11 ms 11 ms xe-7-0-2.bar2.Philadelphia1.Level3.net [4.30.46.33]
5 107 ms 105 ms 107 ms ae-1-3101.bar1.Madrid2.Level3.net [4.69.210.222]
6 107 ms 111 ms 107 ms ae-1-3101.bar1.Madrid2.Level3.net [4.69.210.222]
7 106 ms 108 ms 113 ms 213.242.113.78
8 113 ms 118 ms 113 ms TELMAD.AE4.uv.rt1.val.red.rediris.es [130.206.245.89]
9 121 ms 119 ms 119 ms anella-val1-router.red.rediris.es [130.206.211.70]
10 * * * Request timed out.
11 119 ms 118 ms 119 ms grosso.upf.edu [84.89.134.145]

I wonder if it is possible for BOINC, or some ancillary program, to do a tracert on a file as it is being downloaded? That would be more useful in finding the sticking points than doing a tracert after the fact, when conditions have changed.

Did I miss something?

I installed my old GTX970 GPU on a Linux Mint Computer a few days ago:
Operating system: Linux Mint 17.3 Cinnamon 32-bit
Cinnamon Version: 2.8.6
Linux Kernel: 3.19.0-32-generic
AMD FX™-6100 Six-Core Priocessorx3
Memory: 7.8GiB
Hard Drive: 114.7 GB
Graphics Card: Nvidia Corporation GF110 [GeForce GTX570 Rev.2]

I have not received any work until today, although the Server Status indicates that there are “Tasks ready to send: 181”. However, I do receive PRIMEGRID.net Wus for my GTX570 without any problem! And all my Windows machines are loaded with work. So I do not understand what is wrong.

Please find the relevant log below:

Mon 14 Mar 2016 09:25:38 AM PET | | Starting BOINC client version 7.2.42 for i686-pc-linux-gnu
Mon 14 Mar 2016 09:25:38 AM PET | | log flags: file_xfer, sched_ops, task
Mon 14 Mar 2016 09:25:38 AM PET | | Libraries: libcurl/7.35.0 OpenSSL/1.0.1f zlib/1.2.8 libidn/1.28 librtmp/2.3
Mon 14 Mar 2016 09:25:38 AM PET | | Data directory: /var/lib/boinc-client
Mon 14 Mar 2016 09:25:38 AM PET | | CUDA: NVIDIA GPU 0: GeForce GTX 570 (driver version unknown, CUDA version 7.5, compute capability 2.0, 1279MB, 1144MB available, 1530 GFLOPS peak)
Mon 14 Mar 2016 09:25:38 AM PET | | OpenCL: NVIDIA GPU 0: GeForce GTX 570 (driver version 352.63, device version OpenCL 1.1 CUDA, 1279MB, 1144MB available, 1530 GFLOPS peak)
Mon 14 Mar 2016 09:25:38 AM PET | | Host name: kle1boinc-GA-970A-D3
Mon 14 Mar 2016 09:25:38 AM PET | | Processor: 6 AuthenticAMD AMD FX(tm)-6100 Six-Core Processor [Family 21 Model 1 Stepping 2]
Mon 14 Mar 2016 09:25:38 AM PET | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good nonstop_tsc extd_apicid aperfmperf pni pclmulqdq monitor ssse3 cx16 sse4_1 sse4_2 popcnt aes xsave avx lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs xop skinit wdt lwp fma4 nodeid_msr topoext perfctr_core perfctr_nb arat cpb hw_pstate npt lbrv svm_lock nrip_save tsc_scale vmcb_clean flushbyasid decodeassists pausefilter pfthreshold vmmcall
Mon 14 Mar 2016 09:25:38 AM PET | | OS: Linux: 3.19.0-32-generic
Mon 14 Mar 2016 09:25:38 AM PET | | Memory: 7.89 GB physical, 7.98 GB virtual
Mon 14 Mar 2016 09:25:38 AM PET | | Disk: 109.15 GB total, 98.89 GB free
Mon 14 Mar 2016 09:25:38 AM PET | | Local time is UTC -5 hours
Mon 14 Mar 2016 09:25:38 AM PET | | Config: GUI RPCs allowed from:
Mon 14 Mar 2016 09:25:38 AM PET | SETI@home | URL http://setiathome.berkeley.edu/; Computer ID 7953465; resource share 100
Mon 14 Mar 2016 09:25:38 AM PET | PrimeGrid | URL http://www.primegrid.com/; Computer ID 511403; resource share 0
Mon 14 Mar 2016 09:25:38 AM PET | | General prefs: from http://www.malariacontrol.net/ (last modified 08-May-2012 21:54:17)
Mon 14 Mar 2016 09:25:38 AM PET | | Host location: none
Mon 14 Mar 2016 09:25:38 AM PET | | General prefs: using your defaults
Mon 14 Mar 2016 09:25:38 AM PET | | Reading preferences override file
Mon 14 Mar 2016 09:25:38 AM PET | | Preferences:
Mon 14 Mar 2016 09:25:38 AM PET | | max memory usage when active: 4037.73MB
Mon 14 Mar 2016 09:25:38 AM PET | | max memory usage when idle: 7267.92MB
Mon 14 Mar 2016 09:25:38 AM PET | | max disk usage: 54.57GB
Mon 14 Mar 2016 09:25:38 AM PET | | max CPUs used: 5
Mon 14 Mar 2016 09:25:38 AM PET | | suspend work if non-BOINC CPU load exceeds 50%
Mon 14 Mar 2016 09:25:38 AM PET | | max download rate: 409600 bytes/sec
Mon 14 Mar 2016 09:25:38 AM PET | | max upload rate: 204800 bytes/sec
Mon 14 Mar 2016 09:25:38 AM PET | | (to change preferences, visit a project web site or select Preferences in the Manager)
Mon 14 Mar 2016 09:25:38 AM PET | | gui_rpc_auth.cfg is empty - no GUI RPC password protection
Mon 14 Mar 2016 09:25:38 AM PET | | Not using a proxy
Mon 14 Mar 2016 09:25:39 AM PET | SETI@home | Sending scheduler request: To fetch work.
Mon 14 Mar 2016 09:25:39 AM PET | SETI@home | Requesting new tasks for NVIDIA
Mon 14 Mar 2016 09:25:43 AM PET | SETI@home | Scheduler request completed: got 0 new tasks
Mon 14 Mar 2016 09:29:48 AM PET | | Fetching configuration file from http://www.gpugrid.net/get_project_config.php
Mon 14 Mar 2016 09:30:35 AM PET | GPUGRID | Master file download succeeded
Mon 14 Mar 2016 09:30:40 AM PET | GPUGRID | Sending scheduler request: Project initialization.
Mon 14 Mar 2016 09:30:40 AM PET | GPUGRID | Requesting new tasks for CPU and NVIDIA
Mon 14 Mar 2016 09:30:45 AM PET | GPUGRID | Scheduler request completed: got 0 new tasks
Mon 14 Mar 2016 09:30:45 AM PET | GPUGRID | No tasks sent
Mon 14 Mar 2016 09:30:45 AM PET | GPUGRID | No tasks are available for Long runs (8-12 hours on fastest card)
Mon 14 Mar 2016 09:30:47 AM PET | GPUGRID | Started download of logogpugrid.png
Mon 14 Mar 2016 09:30:47 AM PET | GPUGRID | Started download of project_1.png
Mon 14 Mar 2016 09:30:49 AM PET | GPUGRID | Finished download of logogpugrid.png
Mon 14 Mar 2016 09:30:49 AM PET | GPUGRID | Started download of project_2.png
Mon 14 Mar 2016 09:30:52 AM PET | GPUGRID | Finished download of project_1.png
Mon 14 Mar 2016 09:30:52 AM PET | GPUGRID | Finished download of project_2.png
Mon 14 Mar 2016 09:30:52 AM PET | GPUGRID | Started download of project_3.png
Mon 14 Mar 2016 09:31:20 AM PET | GPUGRID | Sending scheduler request: To fetch work.
Mon 14 Mar 2016 09:31:20 AM PET | GPUGRID | Requesting new tasks for NVIDIA
Mon 14 Mar 2016 09:31:25 AM PET | GPUGRID | Scheduler request completed: got 0 new tasks
Mon 14 Mar 2016 09:31:25 AM PET | GPUGRID | No tasks sent
Mon 14 Mar 2016 09:31:25 AM PET | GPUGRID | No tasks are available for Long runs (8-12 hours on fastest card)
Mon 14 Mar 2016 09:44:11 AM PET | GPUGRID | Sending scheduler request: To fetch work.
Mon 14 Mar 2016 09:44:11 AM PET | GPUGRID | Requesting new tasks for NVIDIA
Mon 14 Mar 2016 09:44:14 AM PET | GPUGRID | Scheduler request completed: got 0 new tasks
Mon 14 Mar 2016 09:44:14 AM PET | GPUGRID | No tasks sent
Mon 14 Mar 2016 09:44:14 AM PET | GPUGRID | No tasks are available for Long runs (8-12 hours on fastest card)

Dear all,

As I have not received an answer to my post above. I would like to repeat my request. Does anybody have an idea, why I can't get WUs on this Linux System? Is it because of the type of the card? Shall I try with a newer generation GTX670? Or is it because of the 32 Bit Linux?

Any comment would be highly appreciated! Thanks.

Your 32 bit linux is a problem. There are no 32 bit linux apps listed on https://www.gpugrid.net/apps.php

I also don't know if compute capability 2.0 is enough any more. There was a thread listing the minimum but I can't find it any more.

And, if that's not enough, your driver version is not listed.

Your log says

Mon 14 Mar 2016 09:25:38 AM PET | | CUDA: NVIDIA GPU 0: GeForce GTX 570 (driver version unknown, CUDA version 7.5, compute capability 2.0, 1279MB, 1144MB available, 1530 GFLOPS peak)

but mine says

14-Mar-2016 22:04:40 [---] CUDA: NVIDIA GPU 0: GeForce GTX 750 Ti (driver version 355.11, CUDA version 7.5, compute capability 5.0, 2047MB, 2010MB available, 1388 GFLOPS peak)


and it is useful to know what driver version you are running.

fractal spotted it; 3.19.0-32-generic

x86 Linux is a non-starter and the GPU needs to be CC3.0 or above:

https://www.gpugrid.net/join.php

Supported OS


Linux 64-bit
Windows 32/64-bit



GPU: NVIDIA Kepler GPU (CC3.0) (Geforce 600 series and later)

FAQ - Recommended GPUs for GPUGrid crunching

Your system might still be useful at one of the other Boinc GPU projects.

---

FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help