I am a student carrying on a summer job started last year where I developed a simple CUDA code (fluid solver, LB D3Q19 MRT), with the aim this summer of implementing the code on multi GPU. I am still quite new to CUDA, and also Linux Redhat.
The server I was working on has two Tesla C2070 cards, but on different nodes (or so, not sure on terminology) which will not get much performance gain when scaling up to 2 GPUs according to my server admin. Hence I was thinking of just getting laptop with 2 GPUs to test how the code scales up, and also to develope the code more easily.
I looked online for CUDA enabled multi GPU laptops and really struggled to find anything. I have never searched laptops online and am pretty ‘noob’ at reviewing all the tech specs. The best by far I have found is the Lenovo IdeaPad Y500 with an ‘UltraBay’, allowing for 2 NVIDIA GeForce GT650M GPUs. However, I do not know whether the Ultrabay is designed for CUDA coding (the Lenovo tech support team also doesn’t apparently) or whether there is adequate driver support (not 100% sure what that means in practice yet) for the two when I try to set up a dual boot system with windows 8 (for general use) and Linux redhat (for working with to facilitate easier testing on the servers later).
- Does the Lenovo Y500 have adequate driver support for use with Linux for multi GPU coding? Particularly, I hear there are problems with Optimus, are these fixed/fixable?
- Will what I am trying to do work? Asin, how difficult will it be to actually set up the dual boot system on the Lenovo Y500, I don’t have too much time since I recently became a father.
- (bonus!) Are there better options for a multi GPU laptop than the Lenovo Y500 that will easily facilitate multi GPU CUDA on Linux Redhat for my purposes?
Many many thanks for any advice and assistance! Also, if I have overlooked any similar topics on the forum, my apologies. I am happy to provide any further information requested.