Confusion on how to get CUDA

ok i have this laptop - [url=“”][/url]
Asus M50Vm-B1

it includes the graphic card Nvidia 9600M GS 1gb Memory

according to the nvidia site This product is also enabled for CUDA applications and is NVIDIA PhysX-ready.

so i tried to look for the newest driver which i used that dled 179.48 beta but i dont see cuda anywhere… as well as the PhysX-ready

The new 2.1 drivers came out recently, perhaps those will work for you.


well what im asking is which driver version should i get? and match it with which cuda ? or just the newest cuda is fine?

can someone help? lol this toooo confusing for me… cuz im tryin to use the coreavc cuda support but i dont know which version to install? like does cuda come with the display driver? or is it seperate? and what version should i be installin if they are seperate? or if they are together?

You’re supposed to ensure your display drivers are properly installed. Then, when people talk of “Get CUDA”, usually it refers to getting the SDK, which is for people who want to write CUDA programs to run on Nvidia GPUs. If you just want to install and run CoreAVC, then you need to follow their instructions, although if it is smart enough to automatically detect CUDA, then by rights, all you need to to is ensure that you have probably the latest drivers for your GPU. Then the application should detect your GPU and use it. You do not need to get the SDK. A quick look at the CoreAVC website specifically says that “You will also need drivers 182.05 or higher from NVIDIA” in order to make use of NVIDIA CUDA.

I’m a professional software developer of many years, and yet I had ‘exactly’ the same question as posted. (And this ‘is’ the “NVIDIA Forums > CUDA GPU Computing > CUDA on Vista” forum isn’t it? No explicit ‘developer’ designation in that, so one might consider that nvidia end users might also have questions about CUDA too, no?)

In any case, thank you much for an excellent answer (with regards to the CoreAVC requirements), if rather round-about delivered. Sadly, and oddly, it looks like the latest nvidia ‘supported’ driver for the 9600M GS is still back at 179.48, not up to CoreAVC’s requirements (according to the search page on nvidia for 9600M GS driver downloads). Hopefully, some day that will be brought up a more current driver version.

It is ‘not’ easy to determine cross-compatibilities of nvidia product features to particular systems and software products, and that’s not nvidia’s fault, but rather the complexity of the beast. So, with all the heavy nvidia commercial advertising of CUDA, such questions seem pretty natural.

There is one place though where nvidia could provide a bit of an assist though, and that is to at least identify somewhere in the nvidia control panel (or somewhere) whether the system’s GPU is CUDA enabled (and perhaps some more end user info on ‘CUDA to driver release’ info). I think that (or those) alone would substantially reduce a lot of such questions. (As it is, from an end users’ perspective CUDA appears more like a catch-phrase that could mean anything from an add-on, to a future-capability, to ???.) Some of that mystery of CUDA for end users definitely needs to be swept away.

Thanks again.


P.S. And please be a bit more accepting of us less technical end users.