I’m shopping for a laptop, and I keep coming across machines with NVIDIA cards with model names that don’t even exist!
I really want to run CUDA on this laptop, but when I match up the model on the specs with the list of models that work with CUDA, they are always missing some letter or the G* is always different.
I built something cheap and powerful with HP, but before I buy, I want to make sure that “NVIDIA GeForce 9200M GE” will work with CUDA. “NVIDIA GeForce 9200M GS” is on the list, but when I try to go to:
Sometimes our OEMs get a bit creative with the product naming. I’m not sure what what the GE indicates, it may have slightly different clock speeds, or maybe it’s just a typo! You should check with HP.
it will run cuda if it is 8 series or above, but the 9200 is not a powerful card… at all. get a 9600 if you can. the 9200 is a low end 9 series. the first number denotes series, the second number denotes how good it is in that series.
not all 9 series are better than 8 series, for example an 8800 is way betetr than a 9200 and a 9400. the difference between 8 and 9 series isnt much, its the second number that denotes performance.
if you are just going to be playing around a bit with cuda then go for a low end 9 series, but if you dont want to be waiting around for hours, get a higher end one.
As a Mac user and developer, I have to say that EVERY actual mac laptop is CUDA enabled (and OpenCL enabled too!) with GeForce 9400M IGP (MCP79) alone or coupled with a GeForce 9600M GT.
Moreover, from a CUDA developer point-of-view, the MacBook Pro equipped with GeForce 9600M GT have the 9400M active simultaneously, enabling to test multi-gpu code and even asymmteric multi-GPU (as performance are really different on these 2 chips), it’s the only platform that enable you to test all that on a train or a plane :-)
The GeForce 9600M GT is 32SP at 1.25Ghz, 9400M is 16SP at 1.1Ghz.
Beware of Snow Leopard, it doesn’t have CUDA driver at this point, so if you buy one, find it with Leopard, and upgrade to Snow Leopard (FREE, DVD included in the box), when the CUDA drivers will be released!
Except that isn’t even close to correct. There are still Intel GMA X3100 based Macbooks on sale in some markets, and it was only in February this year that every tier of the Apple laptop lineup officially transitioned to MCP79 chipsets. There are literally millions of Intel Apple laptops in the wild, many as little as 6 months old, which cannot run CUDA.
Depend if you buy an used one, or a brand new one on AppleStore. Specifications are clear on the Apple Store, for brand new products and refurbisehd ones too. This a the same sense as if you have said “there’s still PC laptop sold with Pentium 3” : you could find them!
Anyway the best thing with MacBook Pro are they have 2 GPUS to test multi-GPU code, and of different performance-level (9400M & 9600M GT) so that you could develop for asymetric configurations and test your code :-)
Better than that, if you plan to go to OpenCL, you could TODAY code for OpenCL with bundled XCode tool, and have it running on your Mac CPU, on the first GPU, on the second GPU, and even on all CPU+2 GPU at once :-)
That is an advanced platform for GPGPU development, and for me probably the best if you don’t focus only on raw performance but on feature implementation/ability of testing/ability to code today for tomorrow!
yes yes, it has two mid range gpu’s… there are laptops avaliable that have two HIGH END gpu’s, im talking two gtx 280’s and the like. that wee’s all over any macbook. i appreciate the usefulness of two different GPU’s but tbh i would want two high end cards rather than two mid range ones, especially for SETI or F@H or any other cuda based processing task that is resource intensive.