anyone writing CUDA code using Quadro mobile cards (1000M or higher)?

Hi,

finally I’ll probably be given a laptop using a Quadro mobile card.

Anyone is using CUDA on these cards? Is it difficult to configure?

I did not find any literature about using CUDA on this GPUs family.

Thank you for any help you would give me.

Bests,

Martina

I’m using Quadro 2000M, on a ThinkPad W520 laptop with 64-bit Slackware installed, for most of my work. The only issue you may find with this type of machines is with Optimus, as there is no native support on Linux for this switching between integrated Intel GPU and discrete NVIDIA GPU, that is Optimus supposed to handle. On my laptop, and number of others, there is BIOS support for practically turning Optimus off, and then just using NVIDIA GPU all the time, like that integrated GPU doesn’t exist at all - having this switch in BIOS was main deciding factor for me when selecting which laptop to buy. Once Optimus turned off, installing NVIDIA driver and then CUDA toolkit, is all that you need to have machine ready for GPU development work. In my case, 2000M worked great for the development of various types of CUDA codes, but you’d probably want access to some sort of more powerful setup for final testing and performance evaluation - either provided by your school/employer, or by renting time on on-demand clusters like Amazon EC2 or Penguin POD.

An alternative for turning Optimus off under Linux is Bumblebee, but I have no experience with it (if you work under Linux only, and then if most of your work is in GPU development, then there is no good reason, in my opinion, to use Optimus at all) so I cannot comment.

If the laptop has optimus you will have the same problems with quadro as you would have with the gtx cards.

Also you should consider something else about optimus. there are 2 types. In one type of optimus you can disable from BIOS the integrated and always use nvidia. this case is the easiest to handle since it will be just like without optimus at all.

The other possibility is the one with more problems on linux. In this case you can only disable from BIOS the nvidia card, because all the ouput is handle by the integrate card.

Thank you both for your suggestions! I’m checking which HP EliteBook (among 8560w,8740w,8760w) does have this switch in the BIOS, how did you retrieve this info for your laptops? I did a search on the web but I was not lucky… maybe just ask to HP!

I learn most of stuff because of CUDA. The one in which optimus can be disabled are the ones in which there is switch which changes between the integrated card and nvidia. There are few laptops with this, because it is more expensive to do this. I know Lenovo had one or two, maybe also hp. The other optimus uses the integrated card even when nvidia is enabled and in this case you can only disable the nvidia card. This is cheaper to produce.

I read something here http://ubuntuforums.org/showthread.php?t=1657660 but not nos much information there.

The second type of optimus is the one with problems on lionuxT.There are no problems on windows. On linux it works in many cases with bumblebee, but better to check for a specific model.