Here's the dealie.....

I'm running an E8500 on Vista with a CUDA compatible evga9800 graphics card.

I have my Boinc set to run on all processors.
I get the small SETI files and they run CUDA just fine, BUT...my cpu will (usually) only run one core on Astropulse (or other work) when the 9800 is running the CUDA work.

When the 9800 is idle, the cpu runs on both cores as normal.

Am I missing something or is this a bug?.

I have on rare occasion seen the cpu running both cores AND the 9800 running CUDA work all at the same time, but this is VERY rare.

I have a Q9650 in for delivery tomorrow and I want all cores running, in addition I have 2 more 9800 graphics cards and I want them all running also.

I should be able to see work being done on 7 processors 24/7 !!

Yo-

---

Here's the dealie.....

I'm running an E8500 on Vista with a CUDA compatible evga9800 graphics card.

I have my Boinc set to run on all processors.
I get the small SETI files and they run CUDA just fine, BUT...my cpu will (usually) only run one core on Astropulse (or other work) when the 9800 is running the CUDA work.

When the 9800 is idle, the cpu runs on both cores as normal.

Am I missing something or is this a bug?.

I have on rare occasion seen the cpu running both cores AND the 9800 running CUDA work all at the same time, but this is VERY rare.

I have a Q9650 in for delivery tomorrow and I want all cores running, in addition I have 2 more 9800 graphics cards and I want them all running also.

I should be able to see work being done on 7 processors 24/7 !!

Yo-

Hi,

Boinc not running all the cores is a known issue and when Boinc 6.6x is officially released, it should fix it. There's more info in this thread that may also assist you. with that.

As far as getting more than one gpu running, you could check out the numbers crunching threads, that have a few active threads on the subject.

---

Well, now after the release of the latest Boinc, I see a very noticeable mark of improvement over the older version I was running. I'm quite pleased with what's been done on this front and to this point I'm really tickled pink about it. Somewhat sadly however, I must report that my motherboard will not allow me run three of the graphics cards I have now (9800gt), but does just fine with two and all my cores are steadily working away!

After eagle eying everything that happens and how things are working, I see rise for another improvement that I believe many others would find quite welcome.

Might there be the possibility of making the CUDA work be suspended for certain applications that are running and otherwise run all the time?

For instance, CUDA is running my SETI work just fine. I'm able to do the small tasks of paying my bills online, check my email, do some forum work, all of which CUDA doesn't bother or slow down in any fashion what so ever. BUT THEN... I'd like to use something that's graphic intensive like watch a movie or a recording from my Media Center, play an online MMORPG, work with Windows Movie Maker or something of that nature and now I have to manually suspend my CUDA work so that these types of applications can run smoothly and without the "jitters" (haha).

So my thinking here, is to add a list of some sort, (which is user editable), that will automatically suspend only CUDA work during the instances I need the full capability of my graphics card(s) while running specific programs that I choose and otherwise, let CUDA run normally.

What do you guys think?

Yo-

---

Might there be the possibility of making the CUDA work be suspended for certain applications that are running and otherwise run all the time?

Asked plenty of times already and just turned down again by the developers. They say to either use the throttle function in BOINC (which will throttle both CPUs and GPUs) or to use the <exclusive_app> option in cc_config.xml to completely suspend BOINC. (See my CUDA FAQ on that, link in my signature)

It's not that we haven't tried. Perhaps some day in the future when this CUDA is more mature.

---

he bro your cpu will only do Astropulse units and seti that take appox 5 hrs units under this time are done by cuda if you look at server stats page you will see there has not been many Astropulse units these last couple of days so unless you are getting units above 5 hrs completed time then your system will probbly only be doing cuda seti or in other words 2 units 1 by cpu and 1 by cuda we all have to wait till there are Astropluse units for the cpu"s and cuda to run all at the same time

---

/me takes a deep breath after reading all that.

---

he bro your cpu will only do Astropulse units and seti that take appox 5 hrs units under this time are done by cuda if you look at server stats page you will see there has not been many Astropulse units these last couple of days so unless you are getting units above 5 hrs completed time then your system will probbly only be doing cuda seti or in other words 2 units 1 by cpu and 1 by cuda we all have to wait till there are Astropluse units for the cpu"s and cuda to run all at the same time

Currently, my seti CUDA (GPU) work gets done pretty quickly. Usually, they get knocked out in about 25 mins per WU, so they go fairly quickly (I try to keep LOTS of these on hand at all times, like a few hundred of them). Astropulses, in comparison take a considerable time to do, usually something in the order of 48 hours to complete. Since the 6.6.2 release, I have seen my CPU cores run the normal CUDA seti files also and they generally get completed in the same time frame that the GPUs will complete them in, around 20 to 30 mins, on average.

In general, I usually have at least one Astropulse running almost all the time (until the system switches to a different project like Rosetta or Einstein) with at least one more waiting to be run. At this moment, I have 2 waiting to be run and another is running as I type.

Normally, all cpu cores will be running large WUs, but I have seen on several occasions, when I've run out of large WUs, my cpu will start running the small seti WUs along with my GPUs at the same time. I normally have at least 400 WUs waiting to run (typically 600+ of them) and the vast majority of them are the seti CUDA work that my hardware crunches through and sends back, bam bam bam.

All my cores run virtually 24/7, with exception of things like power outages, reboots when necessary, or needing my GPUs for something that's a little fun!

Yo-

---

Running AP unit on 1 CPU yet the other CPU is running S@H; 1 CUDA running on 1 GPU. So Why not 2 CPUs with 2 AP units?

Hi Ageless ;)


Forgive me I'm tired and after reading threads, I only see big technical words (my new job is over-clocking my brain).

I haven't had time to tweak anything yet with Ramster's files, I only have BOINC installed. I noticed a new version of BOINC Client is out so I have updated to 6.6.31. But before I go the next step (Ramster), I also noticed earlier today that I'm only running 1 Astropulse on my CPU and the other CPU is running SETI@home; 3rd is of course the GPU. This must have started afer I upgraded from version 6.4.7 and deleted my configs (Old message at bottom of this message). I know this is an easy one where I probably place a config file somewhere in my prefs.

I have installed the Nvidia 185.35 drivers and "cudadriver_2.2_winxp_32_185.85_general.exe"

Applications are set to:
SETI@home Enhanced: yes
Astropulse: yes
Astropulse v5: yes


My current messages:

6/4/2009 9:19:02 PM Starting BOINC client version 6.6.31 for windows_intelx86
6/4/2009 9:19:02 PM log flags: task, file_xfer, sched_ops
6/4/2009 9:19:02 PM Libraries: libcurl/7.19.4 OpenSSL/0.9.8j zlib/1.2.3
6/4/2009 9:19:02 PM Data directory: C:\Documents and Settings\All Users\Application Data\BOINC
6/4/2009 9:19:02 PM Running under account User
6/4/2009 9:19:02 PM Processor: 2 GenuineIntel Intel(R) Pentium(R) 4 CPU 3.20GHz [x86 Family 15 Model 4 Stepping 1]
6/4/2009 9:19:02 PM Processor features: fpu tsc sse sse2 mmx
6/4/2009 9:19:02 PM OS: Microsoft Windows XP: Professional x86 Edition, Service Pack 3, (05.01.2600.00)
6/4/2009 9:19:02 PM Memory: 1.50 GB physical, 2.85 GB virtual
6/4/2009 9:19:02 PM Disk: 179.31 GB total, 111.34 GB free
6/4/2009 9:19:02 PM Local time is UTC -7 hours
6/4/2009 9:19:03 PM CUDA device: GeForce GTX 260 (driver version 18585, compute capability 1.3, 896MB, est. 111GFLOPS)
6/4/2009 9:19:03 PM Not using a proxy
6/4/2009 9:19:04 PM SETI@home URL: http://setiathome.berkeley.edu/; Computer ID: 2378854; location: home; project prefs: default
6/4/2009 9:19:04 PM SETI@home General prefs: from SETI@home (last modified 04-Mar-2009 23:03:00)
6/4/2009 9:19:04 PM SETI@home Computer location: home
6/4/2009 9:19:04 PM SETI@home General prefs: no separate prefs for home; using your defaults
6/4/2009 9:19:04 PM Preferences limit memory usage when active to 767.36MB
6/4/2009 9:19:04 PM Preferences limit memory usage when idle to 1381.25MB
6/4/2009 9:19:04 PM Preferences limit disk usage to 89.65GB
6/4/2009 9:19:04 PM SETI@home Restarting task ap_21mr09aa_B0_P1_00219_20090501_06557.wu_2 using astropulse_v5 version 503
6/4/2009 9:19:05 PM SETI@home Restarting task 19mr09aa.676.73399.8.8.248_0 using setiathome_enhanced version 603
6/4/2009 9:19:05 PM SETI@home Restarting task 19mr09aa.14564.83624.7.8.97_0 using setiathome_enhanced version 608




My previous version messages:

5/16/2009 11:59:29 AM||Starting BOINC client version 6.4.7 for windows_intelx86
5/16/2009 11:59:29 AM||log flags: task, file_xfer, sched_ops
5/16/2009 11:59:29 AM||Libraries: libcurl/7.19.4 OpenSSL/0.9.8j zlib/1.2.3
5/16/2009 11:59:29 AM||Data directory: C:\Documents and Settings\All Users\Application Data\BOINC
5/16/2009 11:59:29 AM||Running under account User
5/16/2009 11:59:29 AM||Processor: 2 GenuineIntel Intel(R) Pentium(R) 4 CPU 3.20GHz [x86 Family 15 Model 4 Stepping 1]
5/16/2009 11:59:29 AM||Processor features: fpu tsc sse sse2 mmx
5/16/2009 11:59:29 AM||OS: Microsoft Windows XP: Professional x86 Editon, Service Pack 3, (05.01.2600.00)
5/16/2009 11:59:29 AM||Memory: 1.50 GB physical, 2.85 GB virtual
5/16/2009 11:59:29 AM||Disk: 179.31 GB total, 104.15 GB free
5/16/2009 11:59:29 AM||Local time is UTC -7 hours
5/16/2009 11:59:29 AM||Not using a proxy
5/16/2009 11:59:29 AM||CUDA devices found
5/16/2009 11:59:29 AM|SETI@home|URL: http://setiathome.berkeley.edu/; Computer ID: 2378854; location: home; project prefs: default
5/16/2009 11:59:29 AM||General prefs: from SETI@home (last modified 04-Mar-2009 23:03:00)
5/16/2009 11:59:29 AM||Computer location: home
5/16/2009 11:59:29 AM||General prefs: no separate prefs for home; using your defaults
5/16/2009 11:59:29 AM||Preferences limit memory usage when active to 767.36MB
5/16/2009 11:59:29 AM||Preferences limit memory usage when idle to 1381.25MB
5/16/2009 11:59:29 AM||Preferences limit disk usage to 89.65GB
5/16/2009 11:59:29 AM||Preferences limit # CPUs to 3
5/16/2009 11:59:29 AM|SETI@home|Restarting task ap_22fe09af_B1_P0_00035_20090513_19035.wu_0 using astropulse_v5 version 503
5/16/2009 11:59:29 AM|SETI@home|Restarting task ap_22fe09af_B1_P1_00250_20090513_21196.wu_2 using astropulse_v5 version 503
5/16/2009 11:59:40 AM|SETI@home|Fetching scheduler list
5/16/2009 11:59:45 AM|SETI@home|Master file download succeeded
5/16/2009 11:59:50 AM|SETI@home|Sending scheduler request: Requested by user. Requesting 30240 seconds of work, reporting 45 completed tasks


Thanks as always!

Dave

---

Hi Dave,

Seti has been low on AP units lately and have been sending out a lot of 6.03 MB work units to keep us busy. This is totally normal. Once the AP work builds back up you will see it pick two of them up again after it finishes the 6.03s.

---

PROUD MEMBER OF Team Starfire World BOINC

Yup soon as I woke up I saw the 2 AP units running.

I thought I was going crazy for a bit there!


Dave

---