It is just interesting that to get maximum performance, you need to use older software revisions.

Only somewhat. I deal more with business apps personally but in that arena you are always working with a collection of sometimes competing requirements.

Speed on a driver is paramount but I can see some code jockey doing a trade-off between uber speed and fixing some other thing that needs gobs of cycles. In a sense he sees it as "his" cycles to be applied to the greater "problem" since it looms bigger in his radar for fixing.

We on the other hand look at the "problem" and shrug (probably something for some board manufacturer or bios that we couldn't give a rat's ass about) and go with the older driver because it doesn't reflect the compromise.

Regards

Well said.

---

Warning, addicted to SETI crunching!
Crunching as a member of GPU Users Group.
GPUUG Website

So I wouldn't take everything Device Manager says for truth. :-)

Sage advice. Once you are on Windows, everything is relative.

Regards,

GPU Z sees both without complaint as well - happily reports each one's various measures.

What am I missing?!!!

Regards

---

What am I missing?!!!

Did the cc_config.xml get lost somehow (or the <use_all_GPUs> therein)?

Gruß,
Gundolf

What am I missing?!!!

Did the cc_config.xml get lost somehow (or the <use_all_GPUs> therein)?

Gruß,
Gundolf

It has:

<cc_config>
<options>
<ncpus>2</ncpus>
</options>
</cc_config>

Now that tip also got me looking back as I recalled seeing that one previously once you mentioned it.

I saw the suggestion for: <use_all_GPUs>1</use_all_GPUs>

and gave it a whirl. Nada.

This is the startup messaging Boinc gives me in fact:

24/03/2010 11:31:15 PM Unrecognized tag in cc_config.xml: <use_all_GPUs>
24/03/2010 11:31:15 PM Starting BOINC client version 6.10.43 for windows_intelx86
24/03/2010 11:31:15 PM Config: use at most 2 CPUs
24/03/2010 11:31:15 PM log flags: file_xfer, sched_ops, task
24/03/2010 11:31:15 PM Libraries: libcurl/7.19.7 OpenSSL/0.9.8l zlib/1.2.3
24/03/2010 11:31:15 PM Data directory: C:\ProgramData\BOINC
24/03/2010 11:31:15 PM Running under account David
24/03/2010 11:31:15 PM Processor: 2 GenuineIntel Intel(R) Core(TM)2 Duo CPU T9300 @ 2.50GHz [Family 6 Model 23 Stepping 6]
24/03/2010 11:31:15 PM Processor: 6.00 MB cache
24/03/2010 11:31:15 PM Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 cx16 sse4_1 nx lm vmx tm2 pbe
24/03/2010 11:31:15 PM OS: Microsoft Windows Vista: Ultimate x86 Edition, Service Pack 2, (06.00.6002.00)
24/03/2010 11:31:15 PM Memory: 3.00 GB physical, 6.19 GB virtual
24/03/2010 11:31:15 PM Disk: 360.04 GB total, 78.04 GB free
24/03/2010 11:31:15 PM Local time is UTC -4 hours
24/03/2010 11:31:15 PM NVIDIA GPU 0: GeForce 8700M GT (driver version 18681, CUDA version 2020, compute capability 1.1, 256MB, 80 GFLOPS peak)
24/03/2010 11:31:15 PM SETI@home URL http://setiathome.berkeley.edu/; Computer ID 4597934; resource share 100
24/03/2010 11:31:15 PM SETI@home General prefs: from SETI@home (last modified 20-Mar-2010 10:54:44)
24/03/2010 11:31:15 PM SETI@home Computer location: home
24/03/2010 11:31:15 PM SETI@home General prefs: no separate prefs for home; using your defaults
24/03/2010 11:31:15 PM Reading preferences override file
24/03/2010 11:31:15 PM Preferences:
24/03/2010 11:31:15 PM max memory usage when active: 2455.32MB
24/03/2010 11:31:15 PM max memory usage when idle: 2915.69MB
24/03/2010 11:31:15 PM max disk usage: 75.00GB
24/03/2010 11:31:15 PM suspend work if non-BOINC CPU load exceeds 25 %
24/03/2010 11:31:15 PM (to change, visit the web site of an attached project,
24/03/2010 11:31:15 PM or click on Preferences)
24/03/2010 11:31:16 PM Using proxy info from GUI
24/03/2010 11:31:16 PM Not using a proxy
24/03/2010 11:31:16 PM SETI@home Restarting task 08mr07ad.15989.41064.7.10.166_2 using setiathome_enhanced version 608


Uniquely frustrating as this laptop is otherwise dead steady. The other two crunchers also have two cards each and don't hesitate to use them.

Regards,

---

You have it slightly off, but BOINC does not seem to see the other card either.

<use_all_gpus>1</use_all_gpus> 

---

OK. So try all lower case: <use_all_gpus>1</use_all_gpus>

F.

---

Try to remove
<ncpus>2</ncpus>

In the past BOINC counted CPUs+GPUs = ncpus
I don't know how it is now


Here:
http://boinc.berkeley.edu/wiki/Client_configuration

it is:
<use_all_gpus>0|1</use_all_gpus>

(lowercase)

---

 


- ALF - "Find out what you don't do well ..... then don't do it!" :)
 

Nada. Nada. And nada.

Tried the gpus parm all lowercase and got rid of the error message in the start up messages. Didn't change anything.

Removed the ncpus. Didn't change anything.

Removed bot the gpus and ncpus parms and that didn't work.

Renamed the cc_config file completely so BOINC wouldn't even see one. Didn't work.

Curiouser and curiouser.

Regards,

---

Nada. Nada. And nada.

Tried the gpus parm all lowercase and got rid of the error message in the start up messages. Didn't change anything.

Removed the ncpus. Didn't change anything.

Removed bot the gpus and ncpus parms and that didn't work.

Renamed the cc_config file completely so BOINC wouldn't even see one. Didn't work.

Curiouser and curiouser.

Regards,

The only shot left in my locker:
- Stop all BOINC processing.
- Delete all Nvidia drivers (from Control Panel).
- Re-boot into SAFE mode.
- Run cc_cleaner to remove all traces of Nvidia drivers from Registry.
- Re-load chosen version of Nvidia drivers (I'm still with 191.07 and very happy there).

Hopefully, since you are on BOINC 6.10.43, this should not dump existing CUDA cache.

If it were mine and that didn't work, the box would be in severe danger of suffering GBH.

F.

---

The only shot left in my locker:
- Stop all BOINC processing.
- Delete all Nvidia drivers (from Control Panel).
- Re-boot into SAFE mode.
- Run cc_cleaner to remove all traces of Nvidia drivers from Registry.
- Re-load chosen version of Nvidia drivers (I'm still with 191.07 and very happy there).

Hopefully, since you are on BOINC 6.10.43, this should not dump existing CUDA cache.

If it were mine and that didn't work, the box would be in severe danger of suffering GBH.

F.

It is CCleaner:
http://www.ccleaner.com/


Try also Enable/Disable of SLI & PhysX

.

---

 


- ALF - "Find out what you don't do well ..... then don't do it!" :)
 

The only shot left in my locker:
- Stop all BOINC processing.
- Delete all Nvidia drivers (from Control Panel).
- Re-boot into SAFE mode.
- Run cc_cleaner to remove all traces of Nvidia drivers from Registry.
- Re-load chosen version of Nvidia drivers (I'm still with 191.07 and very happy there).

Hopefully, since you are on BOINC 6.10.43, this should not dump existing CUDA cache.

If it were mine and that didn't work, the box would be in severe danger of suffering GBH.

F.

It is CCleaner:
http://www.ccleaner.com/

Ahhh - you're right, it is indeed. But now that I re-checked, I really meant Driver Sweeper anyway ;P. Sorry for any confusion that I may have caused.

F.

---

And put a new cc_config.xml file in BOINC, too.

<cc_config>
<log_flags>
<coproc_debug>1</coproc_debug>
</log_flags>
<options>
<use_all_gpus>1</use_all_gpus>
</options>
</cc_config>

Btw. all lower-case, file name, too.

---

Very much appreciate all the excellent advice.

I will give it a go in the next 24 hours and report any progress.

I am certain there's something interesting lurking within all of this. As I say, the old laptop has been a steady performer up until this point. I just had to go meddling with it and install new drivers and a new BOINC and upset the apple cart.

Thanks again.

Regards,

---

Do you keep both GPUs with monitors attached until the issue is solved?
Did the re-installed (older) NVIDIA driver had chance to "see" 2 monitors attached?

Can you post GPU-Z screenshots of both GPUs?


My integrated (no-CUDA) GPU:




I use IrfanView & Ploader (very easy to use)
http://ploader.net/

"Ploader is a quick image hosting service that allows you to upload, link and share your images with your friends, family, and clients. No account required."

I found that the name of the file given by Ploader is its MD5 checksum + .Ext

http://ploader.net/files/a7015ed144b9a792061af912d90e33bd.png
___ MD5 of this image is a7015ed144b9a792061af912d90e33bd

So if you forget the link you can generate MD5 checksum (I use Total Commander) and make the link "by hand"

.

---

 


- ALF - "Find out what you don't do well ..... then don't do it!" :)
 

Do you keep both GPUs with monitors attached until the issue is solved?
Did the re-installed (older) NVIDIA driver had chance to "see" 2 monitors attached?

I did. The laptop has its own screen of course and the extra monitor remains attached at the moment. I opened and closed Outlook on it a few times throughout just to keep the electrons moving.

Can you post GPU-Z screenshots of both GPUs?

Shall do! I will have that sometime tomorrow (today is work and tonight is a show so I'm AFK on that machine for a bit).

I use IrfanView & Ploader (very easy to use)
http://ploader.net/

Great tip - thanks.

This is the misbehaving machine by the way, http://setiathome.berkeley.edu/show_host_detail.php?hostid=4597934. I looked back to the top of this thread and the original bad boy is now behaving again admirably.

Regards and thanks again folks for your generous help,

---

And put a new cc_config.xml file in BOINC, too.

<cc_config>
<log_flags>
<coproc_debug>1</coproc_debug>
</log_flags>
<options>
<use_all_gpus>1</use_all_gpus>
</options>
</cc_config>

Btw. all lower-case, file name, too.

Nada...

This is what I get at starting up again after exiting and then editing the cc_config file.

Still running a monitor on the second GPU along with the laptop's one.

28/03/2010 10:17:32 PM Starting BOINC client version 6.10.43 for windows_intelx86
28/03/2010 10:17:32 PM Config: use all coprocessors
28/03/2010 10:17:32 PM log flags: file_xfer, sched_ops, task, coproc_debug
28/03/2010 10:17:32 PM Libraries: libcurl/7.19.7 OpenSSL/0.9.8l zlib/1.2.3
28/03/2010 10:17:32 PM Data directory: C:\ProgramData\BOINC
28/03/2010 10:17:32 PM Running under account David
28/03/2010 10:17:32 PM Processor: 2 GenuineIntel Intel(R) Core(TM)2 Duo CPU T9300 @ 2.50GHz [Family 6 Model 23 Stepping 6]
28/03/2010 10:17:32 PM Processor: 6.00 MB cache
28/03/2010 10:17:32 PM Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 cx16 sse4_1 nx lm vmx tm2 pbe
28/03/2010 10:17:32 PM OS: Microsoft Windows Vista: Ultimate x86 Edition, Service Pack 2, (06.00.6002.00)
28/03/2010 10:17:32 PM Memory: 3.00 GB physical, 6.19 GB virtual
28/03/2010 10:17:32 PM Disk: 360.04 GB total, 81.50 GB free
28/03/2010 10:17:32 PM Local time is UTC -4 hours
28/03/2010 10:17:32 PM NVIDIA GPU 0: GeForce 8700M GT (driver version 18681, CUDA version 2020, compute capability 1.1, 256MB, 80 GFLOPS peak)
28/03/2010 10:17:32 PM NVIDIA library reports 1 GPU
28/03/2010 10:17:32 PM No ATI library found.
28/03/2010 10:17:32 PM SETI@home URL http://setiathome.berkeley.edu/; Computer ID 4597934; resource share 100
28/03/2010 10:17:32 PM SETI@home General prefs: from SETI@home (last modified 20-Mar-2010 10:54:44)
28/03/2010 10:17:32 PM SETI@home Computer location: home
28/03/2010 10:17:32 PM SETI@home General prefs: no separate prefs for home; using your defaults
28/03/2010 10:17:32 PM Reading preferences override file
28/03/2010 10:17:32 PM Preferences:
28/03/2010 10:17:32 PM max memory usage when active: 1534.57MB
28/03/2010 10:17:32 PM max memory usage when idle: 2762.23MB
28/03/2010 10:17:32 PM max disk usage: 20.00GB
28/03/2010 10:17:32 PM don't use GPU while active
28/03/2010 10:17:32 PM suspend work if non-BOINC CPU load exceeds 20 %
28/03/2010 10:17:32 PM (to change, visit the web site of an attached project,
28/03/2010 10:17:32 PM or click on Preferences)
28/03/2010 10:17:32 PM Using proxy info from GUI
28/03/2010 10:17:32 PM Not using a proxy
28/03/2010 10:17:32 PM SETI@home Restarting task 27mr07ag.25228.11524.14.10.25_1 using setiathome_enhanced version 603
28/03/2010 10:17:32 PM SETI@home Restarting task 27mr07ag.26409.9070.15.10.22_1 using setiathome_enhanced version 603


Getting to the end of the ingenuity on this one.

Regards,

---

Have you tried extending the Desktop onto the second Monitor? then restarting Boinc?

Claggy

Are you sure this laptop has 2 GPUs?
Or just one GPU with outputs for 2 monitors?
Is the second external monitor digital (DVI) or analog CRT (VGA, D-sub)

Digital Visual Interface
http://en.wikipedia.org/wiki/Digital_Visual_Interface

VGA connector
http://en.wikipedia.org/wiki/VGA_connector

---

 


- ALF - "Find out what you don't do well ..... then don't do it!" :)
 

Are you sure this laptop has 2 GPUs?

Good question (always worth checking the basics) but "unfortunately": yes. They were in fact running quite well until I did a driver update and one disappeared, in BOINC. When I examine it in the device manager its there and confirmed to be running fine.

I also bought the laptop with two (and two raided hard drives, etc - it was meant to be a desktop replacement as well as portable).

It's been a rock solid cruncher even and never waivered, always running two cuda tasks happily, until I decided to meddle with it.

Regards,

---

Do you have driver disk (CD, DVD) supplied with the laptop?
Try the video driver from that disk.

Can you find Video driver on the web site of laptop manufacturer?
Sometimes there are proprietary (non-standard) chips or hardware solutions on the laptops which prevent the generic drivers to work correctly.

So try to get the driver from the Video Card (or laptop) manufacturer and not from NVIDIA.


Programs you can use to find the exact manufacturer/model of the Video Card (and much more):

EVEREST
http://www.lavalys.com/

SIV 'System Information Viewer'
http://www.rh-software.com/

.

---

 


- ALF - "Find out what you don't do well ..... then don't do it!" :)