Enabling second GPU on MB with embedded GeForce6100 - any ideas?

Message boards : Number crunching : Enabling second GPU on MB with embedded GeForce6100 - any ideas?
Message board moderation

To post messages, you must log in.

1 · 2 · Next

AuthorMessage
Profile Raistmer
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 16 Jun 01
Posts: 6325
Credit: 106,370,077
RAC: 121
Russia
Message 954515 - Posted: 13 Dec 2009, 20:26:45 UTC

What I did till the moment:
1) in BIOS setted "Always enable" for embedded GPU
2) in BIOS setted "Initialize onboard GPU first".

Monitor attached to embedded GPU. nVidia driver sees only first, embedded GPU and doesn't see secondary GPU (8500GT). So it seems I can't "enable PhysX" on secondary GPU.

Any suggestions? Should I attach another monitor to secondary GPU or some software-only solutions?

Both GPUs are nVidia's ones so it should be possible to enable both, isn't ?
ID: 954515 · Report as offensive
Profile arkayn
Volunteer tester
Avatar

Send message
Joined: 14 May 99
Posts: 4438
Credit: 55,006,323
RAC: 0
United States
Message 954528 - Posted: 13 Dec 2009, 20:50:43 UTC - in response to Message 954515.  

Second monitor should enable the computer to find the GPU, then you might want to make a dummy plug for it if you want to use the monitor elsewhere.

ID: 954528 · Report as offensive
Profile Raistmer
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 16 Jun 01
Posts: 6325
Credit: 106,370,077
RAC: 121
Russia
Message 954543 - Posted: 13 Dec 2009, 21:18:58 UTC - in response to Message 954528.  

Second monitor should enable the computer to find the GPU, then you might want to make a dummy plug for it if you want to use the monitor elsewhere.

Thanks.
How do you think, is it enough to plug single monitor to secondary GPU leaving first one w/o monitor until OS (Windows XP btw) booted and then re-plug it to first GPU? Or I should plug to both GPUs simultaneously ?

ID: 954543 · Report as offensive
Wembley
Volunteer tester
Avatar

Send message
Joined: 16 Sep 09
Posts: 429
Credit: 1,844,293
RAC: 0
United States
Message 954548 - Posted: 13 Dec 2009, 21:29:34 UTC - in response to Message 954543.  

Lots of monitors have both an analog and digital input nowadays and let you switch, so you might be able to hook just one monitor up to both gpus.
ID: 954548 · Report as offensive
Profile Raistmer
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 16 Jun 01
Posts: 6325
Credit: 106,370,077
RAC: 121
Russia
Message 954557 - Posted: 13 Dec 2009, 21:52:39 UTC - in response to Message 954548.  

Lots of monitors have both an analog and digital input nowadays and let you switch, so you might be able to hook just one monitor up to both gpus.

Unfortunately that monitor has only single input (that's why my question worth time to type it, in other case direct experiment would be much easier ;) )
ID: 954557 · Report as offensive
Profile Sutaru Tsureku
Volunteer tester

Send message
Joined: 6 Apr 07
Posts: 7105
Credit: 147,663,825
RAC: 5
Germany
Message 954566 - Posted: 13 Dec 2009, 22:31:12 UTC - in response to Message 954557.  


I have a GTX260-216 at 1st PCIe (1.0 x16) and a 6200 LE at 3rd PCIe (1.0 x4 - only here display connected).

I needed to extend the WindowsDesktop to both, that CUDA on the GTX260-216.

WinXP 32bit, nVIDIA_driver_190.38 .


Maybe, extend the WindowsDesktop and after reboot 'new hardware found' for to install the nVIDIA_driver?




ID: 954566 · Report as offensive
Profile Raistmer
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 16 Jun 01
Posts: 6325
Credit: 106,370,077
RAC: 121
Russia
Message 954568 - Posted: 13 Dec 2009, 22:36:00 UTC - in response to Message 954566.  

Maybe, extend the WindowsDesktop and after reboot 'new hardware found' for to install the nVIDIA_driver?

Could you describe how to extend desktop in my case?
Windows "sees" only embedded GPU, I see no options to involve second GPU into desktop drawing...
ID: 954568 · Report as offensive
Profile Sutaru Tsureku
Volunteer tester

Send message
Joined: 6 Apr 07
Posts: 7105
Credit: 147,663,825
RAC: 5
Germany
Message 954569 - Posted: 13 Dec 2009, 22:41:04 UTC - in response to Message 954515.  

What I did till the moment:
1) in BIOS setted "Always enable" for embedded GPU
2) in BIOS setted "Initialize onboard GPU first".
[...]


Hmm.. then I would disable the 2nd option.
Then maybe both GPUs will be found.




ID: 954569 · Report as offensive
Crun-chi
Volunteer tester
Avatar

Send message
Joined: 3 Apr 99
Posts: 174
Credit: 3,037,232
RAC: 0
Croatia
Message 954579 - Posted: 13 Dec 2009, 23:41:39 UTC

Use "dummy" monitor for GPU (8500GT).
Attach it and then boot up machine.
OS in that case will find another GPU, and BOINC will use it.


I am cruncher :)
I LOVE SETI BOINC :)
ID: 954579 · Report as offensive
Profile mimo
Volunteer tester
Avatar

Send message
Joined: 7 Feb 03
Posts: 92
Credit: 14,957,404
RAC: 0
Slovakia
Message 954730 - Posted: 14 Dec 2009, 15:35:04 UTC
Last modified: 14 Dec 2009, 15:54:32 UTC

ill try disable this "onboard initialization first" thing ....

(i have three HW cuda devices under W7.
on first, 9600GT i have attached monitor, on second 8500 i have set psychx and third, nv750 on board i dont see as cuda device under drivers .... so i think its a driver "problem" - maybe onboard devices are excluded from cuda list in drivers)

ID: 954730 · Report as offensive
Wembley
Volunteer tester
Avatar

Send message
Joined: 16 Sep 09
Posts: 429
Credit: 1,844,293
RAC: 0
United States
Message 954753 - Posted: 14 Dec 2009, 16:57:34 UTC - in response to Message 954730.  

ill try disable this "onboard initialization first" thing ....

(i have three HW cuda devices under W7.
on first, 9600GT i have attached monitor, on second 8500 i have set psychx and third, nv750 on board i dont see as cuda device under drivers .... so i think its a driver "problem" - maybe onboard devices are excluded from cuda list in drivers)

I'm not sure what an nv750 is, but it isn't listed here on the cuda capable list: http://en.wikipedia.org/wiki/CUDA#Supported_GPUs
ID: 954753 · Report as offensive
kittyman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Jul 00
Posts: 51468
Credit: 1,018,363,574
RAC: 1,004
United States
Message 954754 - Posted: 14 Dec 2009, 17:02:05 UTC - in response to Message 954753.  

ill try disable this "onboard initialization first" thing ....

(i have three HW cuda devices under W7.
on first, 9600GT i have attached monitor, on second 8500 i have set psychx and third, nv750 on board i dont see as cuda device under drivers .... so i think its a driver "problem" - maybe onboard devices are excluded from cuda list in drivers)

I'm not sure what an nv750 is, but it isn't listed here on the cuda capable list: http://en.wikipedia.org/wiki/CUDA#Supported_GPUs

I would be rather surprised if any mobo manufacturer would spend the money to put a Cuda capable video chipset onboard.
If there are any, I am not aware of them. Onboard video is more apt to be a baseline solution for those who don't want to spend the money on a higher end video card.
"Freedom is just Chaos, with better lighting." Alan Dean Foster

ID: 954754 · Report as offensive
Profile Raistmer
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 16 Jun 01
Posts: 6325
Credit: 106,370,077
RAC: 121
Russia
Message 954799 - Posted: 14 Dec 2009, 20:40:23 UTC

Thanks all!
I will try your suggestions when will be near that PC next time. Starting with BIOS proposals :)
ID: 954799 · Report as offensive
Profile perryjay
Volunteer tester
Avatar

Send message
Joined: 20 Aug 02
Posts: 3377
Credit: 20,676,751
RAC: 0
United States
Message 954819 - Posted: 14 Dec 2009, 21:58:10 UTC - in response to Message 954799.  

Let me know what you find out. I've got the same thing on this little EMachine I'm running and would like to run the monitor on the onboard and CUDA on my 9500GT card. Been afraid to try it myself as everything is running good as is now.


PROUD MEMBER OF Team Starfire World BOINC
ID: 954819 · Report as offensive
Profile rkillian
Avatar

Send message
Joined: 3 Aug 05
Posts: 2
Credit: 2,339,486
RAC: 0
United States
Message 954880 - Posted: 15 Dec 2009, 2:24:50 UTC

The 6100 is not cuda compatable so you will most likely waste your time trying to make it work. I had one of the onboard GPU's and had the same idea. You are better off to disable it.
ID: 954880 · Report as offensive
Profile Pappa
Volunteer tester
Avatar

Send message
Joined: 9 Jan 00
Posts: 2562
Credit: 12,301,681
RAC: 0
United States
Message 954888 - Posted: 15 Dec 2009, 3:48:48 UTC - in response to Message 954515.  

What I did till the moment:
1) in BIOS setted "Always enable" for embedded GPU
2) in BIOS setted "Initialize onboard GPU first".

Monitor attached to embedded GPU. nVidia driver sees only first, embedded GPU and doesn't see secondary GPU (8500GT). So it seems I can't "enable PhysX" on secondary GPU.

Any suggestions? Should I attach another monitor to secondary GPU or some software-only solutions?

Both GPUs are nVidia's ones so it should be possible to enable both, isn't ?


I have a couple of MotherBoards that have embedded GPU's. The BIOS and attached monitors control what is happening. So if the settings are wrong, and there is no monitor attached you get a black screen. Windows may load and sit waiting at the Login Prompt. Or depending Fail under stop on All Errors.
My most notable would be this machine which has basically has an embedded nvidia 9100 chip (cheaper than a full card). But not cuda compatable. Boinc Sees it as a result of the installed drivers.

So if you leave the BIOS at the Factory Defaults and add a card and connect monitors to both video ports (with drivers are installed for both) they both should be active depending on the OS installed. The embedded should be first and that would be here you get the login prompt. A key term in the BIOS is "PEG" if that is activated then it would indicate that the BIOS is supposed to look for an added GPU adapter in a Slot other than the embedded GPU.

Regards

Please consider a Donation to the Seti Project.

ID: 954888 · Report as offensive
Profile mimo
Volunteer tester
Avatar

Send message
Joined: 7 Feb 03
Posts: 92
Credit: 14,957,404
RAC: 0
Slovakia
Message 954928 - Posted: 15 Dec 2009, 6:21:06 UTC - in response to Message 954753.  

ill try disable this "onboard initialization first" thing ....

(i have three HW cuda devices under W7.
on first, 9600GT i have attached monitor, on second 8500 i have set psychx and third, nv750 on board i dont see as cuda device under drivers .... so i think its a driver "problem" - maybe onboard devices are excluded from cuda list in drivers)

I'm not sure what an nv750 is, but it isn't listed here on the cuda capable list: http://en.wikipedia.org/wiki/CUDA#Supported_GPUs


nforce 750i is chipset with pre-ION chips - CUDA device with 8 streams and is using a system memory

ID: 954928 · Report as offensive
Profile zoom3+1=4
Volunteer tester
Avatar

Send message
Joined: 30 Nov 03
Posts: 65746
Credit: 55,293,173
RAC: 49
United States
Message 954933 - Posted: 15 Dec 2009, 6:47:25 UTC

I guess I'm unlucky or something, As pc3 has an ATi chipset and ATi video(m7470n), While I set the bios for built in, PCI-E or PCI, The 3.47 HP(ms-7184) Bios can not disable the ATi Video, The MSI 3.9(ms-7093) bios can, But until I hook up an internal fdd flashing the bios there is impossible.
The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's
ID: 954933 · Report as offensive
Profile Raistmer
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 16 Jun 01
Posts: 6325
Credit: 106,370,077
RAC: 121
Russia
Message 954941 - Posted: 15 Dec 2009, 8:01:38 UTC - in response to Message 954880.  
Last modified: 15 Dec 2009, 8:04:46 UTC

The 6100 is not cuda compatable so you will most likely waste your time trying to make it work. I had one of the onboard GPU's and had the same idea. You are better off to disable it.

I'm not trying to run CUDA MB on 6100. I want to use 6100 as video-card attached to monitor and secondary, CUDA-enabled GPU - as co-processor.
Why it needed in my case:
secondary GPU is low-end CUDA GPU. Running alone it encounters kernel crashes often. I think it's because of 3 seconds limit (for WinXP) for kernel length. Being used as secondary GPU it will be free from this limitation and could run even longest kernels w/o errors or unacceptable slow GUI response.
That's why I try to make such dual-GPU config.
ID: 954941 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14650
Credit: 200,643,578
RAC: 874
United Kingdom
Message 954951 - Posted: 15 Dec 2009, 10:15:30 UTC

For test purposes only, set the BIOS to use the 8500GT and attach a second monitor to that (you'll probably need two monitors available while you do the setting up). You should get Windows running, using the 8500GT as the display card.

Now, is the embedded 6100 visible to Windows? Some BIOSs disable the on-board video when a plug-in card is used, in which case you're stuck.

But if Windows can still see the embedded adapter: extend the desktop to the 6100, set the 6100 as the primary display, and then de-extend it from the 8500GT. Test it running like that for a while - see if it avoids the kernal timeouts. If it passes that test, try rebooting a few times to see if the configuration sticks. And finally, see if it boots OK without a monitor attached to the 8500GT. If you can make all that work (you'll proably be booting blind, not able to see POST or the Windows logo screen), you can give the second monitor back to the colleague you pinched it from!

[All this is guesswork - I've been through the 'add a second monitor' stage often enough, but that's because the user wants to run two monitors. I've never actually tried taking the primary one off]
ID: 954951 · Report as offensive
1 · 2 · Next

Message boards : Number crunching : Enabling second GPU on MB with embedded GeForce6100 - any ideas?


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.