CPU info has changed

[jgillett]

Looking at 'Your computers' from BOINC I'm seeing a change in the info for CPU on this iMac. It's barely 2 months old and one of the first things I did was set up seti. Looked then and it was showing 8 processors, which is correct. Now it shows only 2 processors.

Nprmally it runs AP 6.02 and Setiathomecuda units. It was a long process getting the cuda stuff to run. Turned out there is a problem in the latest NVIDIA SDK drivers (libraries could not be found) and we had to drop back a level to fix it. Cuda started right up, but it stopped running the 'normal' Setiathome_v7 stuff as it was prior to getting cuda started. Now the AP 6.02 is all 'waiting to run' in favor of the pogs which is running 'high priority' (some units, but not all). Confused yet?

Think there has been a bug in app_info or something all along, but don't know where. Just confused as to why I should lose 6 processors - at least according to BOINC.

3.5 GHz Intel Core i7
16 GB 1600 MHz DDR3
NVIDIA GeForce GTX 780M 4096 MB

Shot is what currently running looks like, including a stalled AP 6.02. Behind this is a huge list of Ready to start stuff.

Thoughts appreciated.

1657

[myshortpencil]

Zombie67 is our resident iMac expert, among others. If he doesn't reply to this entry in a day, try sending a private message (PM) to him. :)

[Mumps]

Well, for starters, I see you're running 7.2.39. And only first joined pogs with that host on Saturday. Could it be that you upgraded to 7.2.39 then? Looks like 7.2.39 for Mac has only been available since Jan 28th, much more recent than your connecting the system to Seti. Some things to check would be starting with the local settings.

Under Tools->Computing Preferences, make sure you have "On multiprocessor systems, use at most" to something near 100%. To "reserve" a core to help feed the GPU better, you may want that set down to something like 90%. On your 8-thread system, that should reserve a full Core simply to meet the needs of feeding data for your 3 GPU WU's you're running, and only try running at most 7 WU's on the CPU threads at a time. If that's set to something like 25%, that could explain the missing 6 cores.

Another thing you could provide that would be helpful in trouble shooting this is the start of the logs from right after starting/restarting BOINC. Included in that should be a few lines identifying your CPU, and another configuration file that could be restricting you is named "cc_config.xml" You might see a line in the startup indicating that the number of processors has been set to 2 by an "ncpus" option in a cc_config.xml file. That would show up in the early parts of the log as a line saying:

Config: simulate 2 CPUs

[myshortpencil]

This looks like your list of computers at SETI (http://setiathome.berkeley.edu/hosts_user.php?userid=13575). It looks like you are running Darwin for the computers showing only 2 cores for your i7-4771. Is that right?

[Mumps]

Yup. Darwin is MacOS. :)

[myshortpencil]

Yup. Darwin is MacOS. :)

Ok. I'll shut up. :p

[jgillett]

Zombie67 is our resident iMac expert, among others. If he doesn't reply to this entry in a day, try sending a private message (PM) to him. :)

Tank U! :)

[jgillett]

Well, for starters, I see you're running 7.2.39. And only first joined pogs with that host on Saturday. Could it be that you upgraded to 7.2.39 then? Looks like 7.2.39 for Mac has only been available since Jan 28th, much more recent than your connecting the system to Seti. Some things to check would be starting with the local settings.

If I remember correctly (I'm old - therefore allowed to forget!), 7.2.39 went up a few days at most before I joined the pogs challenge.


Under Tools->Computing Preferences, make sure you have "On multiprocessor systems, use at most" to something near 100%. To "reserve" a core to help feed the GPU better, you may want that set down to something like 90%. On your 8-thread system, that should reserve a full Core simply to meet the needs of feeding data for your 3 GPU WU's you're running, and only try running at most 7 WU's on the CPU threads at a time. If that's set to something like 25%, that could explain the missing 6 cores.

Right now that is set to 0 (zero) - i.e., ignore. I just set it to 90% and one of the two running pogs immediately dropped to 'waiting to run'. So now both an AP and a pogs are stalled. All 3 GPUs continue to run (actually it's the CPUs that are funky). Set it back to 0 and the stalled pogs started again.


Another thing you could provide that would be helpful in trouble shooting this is the start of the logs from right after starting/restarting BOINC. Included in that should be a few lines identifying your CPU, and another configuration file that could be restricting you is named "cc_config.xml" You might see a line in the startup indicating that the number of processors has been set to 2 by an "ncpus" option in a cc_config.xml file. That would show up in the early parts of the log as a line saying:

Config: simulate 2 CPUs

cc_config never had existed on this system. Found one online somewhere and put it up. Such a deal... It had <ncpus>2</ncpus>. Just changed that to 8 with the running results shown below.

That's more than just a 5-Star difference!! Good catch. Should I change it back to 7 just to give me some elbow room on the rest of the machine?

Current cc-config is simply...

<cc_config>
<log_flags>
</log_flags>
<options>
<use_all_gpus>1</use_all_gpus>
<ncpus>8</ncpus>
</options>
</cc_config>


Suggestions for something 'correct'?

And one last thing - how do I get this thing to start running the Setiathome_v7 CPU units again? They were running fine prior to changes made to get the GPU stuff running.

Thank you!! :)

[jgillett]

This looks like your list of computers at SETI (http://setiathome.berkeley.edu/hosts_user.php?userid=13575). It looks like you are running Darwin for the computers showing only 2 cores for your i7-4771. Is that right?

Yes, those are mine (except for one you probably recognize), but now fixed, thanks to Mumps!

[John P. Myers]

Personally i'd remove the NCPUS line completely, unless you're trying to trick the server into thinking you have more cores than you actually do. To run the number of cores you actually have, or less, NCPUS isn't needed. It can be set in your BOINC manager.

[jgillett]

Personally i'd remove the NCPUS line completely, unless you're trying to trick the server into thinking you have more cores than you actually do. To run the number of cores you actually have, or less, NCPUS isn't needed. It can be set in your BOINC manager.

Done. Thank you.

[jgillett]

Mumps, you asked to see the log yesterday and I forgot to post it...

Tue Feb 18 09:55:13 2014 | | Starting BOINC client version 7.2.39 for x86_64-apple-darwin
Tue Feb 18 09:55:13 2014 | | log flags: file_xfer, sched_ops, task
Tue Feb 18 09:55:13 2014 | | Libraries: libcurl/7.26.0 OpenSSL/1.0.1e zlib/1.2.5 c-ares/1.9.1
Tue Feb 18 09:55:13 2014 | | Data directory: /Library/Application Support/BOINC Data
Tue Feb 18 09:55:13 2014 | | CUDA: NVIDIA GPU 0: GeForce GTX 780M (driver version 5.5.20, CUDA version 5.5, compute capability 3.0, 4096MB, 3530MB available, 2408 GFLOPS peak)
Tue Feb 18 09:55:13 2014 | | OpenCL: NVIDIA GPU 0: GeForce GTX 780M (driver version 8.18.22 310.40.05f01, device version OpenCL 1.2, 4096MB, 3530MB available, 2408 GFLOPS peak)
Tue Feb 18 09:55:13 2014 | | OpenCL CPU: Intel(R) Core(TM) i7-4771 CPU @ 3.50GHz (OpenCL driver vendor: Apple, driver version 1.1, device version OpenCL 1.2)
Tue Feb 18 09:55:13 2014 | SETI@home | Found app_info.xml; using anonymous platform
Tue Feb 18 09:55:13 2014 | SETI@home | [error] State file error: missing application file setiathome_7.00_i686-apple-darwin
Tue Feb 18 09:55:13 2014 | | Host name: John.local
Tue Feb 18 09:55:13 2014 | | Processor: 8 GenuineIntel Intel(R) Core(TM) i7-4771 CPU @ 3.50GHz [x86 Family 6 Model 60 Stepping 3]
Tue Feb 18 09:55:13 2014 | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clfsh ds acpi mmx fxsr sse sse2 ss htt tm pbe pni pclmulqdq dtes64 mon dscpl vmx smx est tm2 ssse3 fma cx16 tpr pdcm sse4_1 sse4_2 x2apic movbe popcnt aes pcid xsave osxsave seglim64 tsctmr avx rdrand f16c
Tue Feb 18 09:55:13 2014 | | OS: Mac OS X 10.9.1 (Darwin 13.0.0)
Tue Feb 18 09:55:13 2014 | | Memory: 16.00 GB physical, 811.50 GB virtual
Tue Feb 18 09:55:13 2014 | | Disk: 930.71 GB total, 811.25 GB free
Tue Feb 18 09:55:13 2014 | | Local time is UTC -7 hours
Tue Feb 18 09:55:13 2014 | | Config: use all coprocessors
Tue Feb 18 09:55:13 2014 | pogs | URL http://pogs.theskynet.org/pogs/; Computer ID 28565; resource share 100
Tue Feb 18 09:55:13 2014 | SETI@home | URL http://setiathome.berkeley.edu/; Computer ID 7166906; resource share 100
Tue Feb 18 09:55:13 2014 | SETI@home | General prefs: from SETI@home (last modified 19-Dec-2013 13:25:36)
Tue Feb 18 09:55:13 2014 | SETI@home | Computer location: home
Tue Feb 18 09:55:13 2014 | | General prefs: using separate prefs for home
Tue Feb 18 09:55:13 2014 | | Reading preferences override file
Tue Feb 18 09:55:13 2014 | | Preferences:
Tue Feb 18 09:55:13 2014 | | max memory usage when active: 9830.40MB
Tue Feb 18 09:55:13 2014 | | max memory usage when idle: 14745.60MB
Tue Feb 18 09:55:13 2014 | | max disk usage: 698.03GB
Tue Feb 18 09:55:13 2014 | | (to change preferences, visit a project web site or select Preferences in the Manager)
Tue Feb 18 09:55:13 2014 | | Not using a proxy
Tue Feb 18 09:55:14 2014 | SETI@home | Sending scheduler request: To report completed tasks.
Tue Feb 18 09:55:14 2014 | SETI@home | Reporting 16 completed tasks
Tue Feb 18 09:55:14 2014 | SETI@home | Requesting new tasks for NVIDIA
Tue Feb 18 09:55:15 2014 | SETI@home | Scheduler request completed: got 0 new tasks
Tue Feb 18 09:55:15 2014 | SETI@home | Project is temporarily shut down for maintenance
Tue Feb 18 09:55:20 2014 | pogs | Sending scheduler request: To fetch work.
Tue Feb 18 09:55:20 2014 | pogs | Requesting new tasks for CPU
Tue Feb 18 09:55:21 2014 | pogs | Scheduler request completed: got 0 new tasks
Tue Feb 18 09:55:21 2014 | pogs | No tasks sent
Tue Feb 18 09:55:21 2014 | pogs | This computer has reached a limit on tasks in progress
Tue Feb 18 09:55:21 2014 | pogs | Project has no tasks available

One line in particular bothers me...

Tue Feb 18 09:55:13 2014 | SETI@home | [error] State file error: missing application file setiathome_7.00_i686-apple-darwin

That app IS present in the proper location and pref and owners are set correctly.

Thoughts appreciated.

[Mumps]

Mumps, you asked to see the log yesterday and I forgot to post it...

Tue Feb 18 09:55:13 2014 | | CUDA: NVIDIA GPU 0: GeForce GTX 780M (driver version 5.5.20, CUDA version 5.5, compute capability 3.0, 4096MB, 3530MB available, 2408 GFLOPS peak)
Tue Feb 18 09:55:13 2014 | | OpenCL: NVIDIA GPU 0: GeForce GTX 780M (driver version 8.18.22 310.40.05f01, device version OpenCL 1.2, 4096MB, 3530MB available, 2408 GFLOPS peak)
Well, you noticed what I wanted to look at the logs for, so they became irrelevant, seeing as you didn't need help interpreting them. Nice GPU. :)

Tue Feb 18 09:55:13 2014 | | OpenCL CPU: Intel(R) Core(TM) i7-4771 CPU @ 3.50GHz (OpenCL driver vendor: Apple, driver version 1.1, device version OpenCL 1.2)
Tue Feb 18 09:55:13 2014 | SETI@home | Found app_info.xml; using anonymous platform
Tue Feb 18 09:55:13 2014 | SETI@home | [error] State file error: missing application file setiathome_7.00_i686-apple-darwin
Don't think I'd worry about that one. If zombie67 or cineon_lut have recently run Seti, they may be able to confirm if this is even a normal messge when using the Opti App.

Tue Feb 18 09:55:13 2014 | | Processor: 8 GenuineIntel Intel(R) Core(TM) i7-4771 CPU @ 3.50GHz [x86 Family 6 Model 60 Stepping 3]
See that "8" there? That helps confirm BOINC is indeed seeing all your cores.

Tue Feb 18 09:55:13 2014 | | Config: use all coprocessors
Tue Feb 18 09:55:13 2014 | pogs | URL http://pogs.theskynet.org/pogs/; Computer ID 28565; resource share 100
Tue Feb 18 09:55:13 2014 | SETI@home | URL http://setiathome.berkeley.edu/; Computer ID 7166906; resource share 100
With an even resource share, BOINC will try to get your recent hours to match that, so it should run more POGs until it gets closer.

Tue Feb 18 09:55:14 2014 | SETI@home | Sending scheduler request: To report completed tasks.
Tue Feb 18 09:55:14 2014 | SETI@home | Reporting 16 completed tasks
Tue Feb 18 09:55:14 2014 | SETI@home | Requesting new tasks for NVIDIA
Tue Feb 18 09:55:15 2014 | SETI@home | Scheduler request completed: got 0 new tasksTue Feb 18 09:55:15 2014 | SETI@home | Project is temporarily shut down for maintenance
And seeing as SETI is down for their standard weekly maintenance, that's even more reason BOINC will download more POGs to make sure your cache stays occupied.

Tue Feb 18 09:55:20 2014 | pogs | Sending scheduler request: To fetch work.
Tue Feb 18 09:55:20 2014 | pogs | Requesting new tasks for CPU
Tue Feb 18 09:55:21 2014 | pogs | Scheduler request completed: got 0 new tasks
Tue Feb 18 09:55:21 2014 | pogs | No tasks sent
Tue Feb 18 09:55:21 2014 | pogs | This computer has reached a limit on tasks in progress
So, at this point, BOINC is still looking for more work to meet your cache size settings, but even POGs says you've got too much POGs work in cache. So you won't get any more until you've finished some, and by that time, maybe SETI will be back up so BOINC can get more AP work to balance things out again. :)

Tue Feb 18 09:55:21 2014 | pogs | Project has no tasks available

One line in particular bothers me...

Tue Feb 18 09:55:13 2014 | SETI@home | [error] State file error: missing application file setiathome_7.00_i686-apple-darwin

That app IS present in the proper location and pref and owners are set correctly.

Thoughts appreciated.

[jgillett]

Mumps, your detailed answer on the log is most appreciated (and no sense repeating the whole thing again here).

One item - SETI@home | [error] State file error: missing application file setiathome_7.00_i686-apple-darwin - has been there ever since the app_info was 'adjusted' to get cuda apps working. Has nothing to do with pogs, the new BOINC, or seti being off-line right now. Before cuda working that app did run properly - after cuda, it just went and hid.

Thanks!

[jgillett]

Well, you noticed what I wanted to look at the logs for, so they became irrelevant, seeing as you didn't need help interpreting them. Nice GPU. :)

Understood - just wanted you to see it.

As to the GPU - yeah, it is nice. Biggest available for the iMac. Unfortunately can't put in anything else.