Difference between Fermi, Tesla and GeForce Architecture?


Can anybody explain, What is difference between these architecures?

Within the same generation of products, architecturally, nothing. But different products are targeted at different end uses, and the hardware features and driver support enabled in each product is different.

Fermi is the code-name of our latest GPU architecture.

Tesla and GeForce are different product lines, compared here:

Actually, I want to use a MAGMA library which gives source code of BLAS routine for Fermi Architecture and I have GeForce 8800 GT. I want to confirm that can I use this library code for GeForce architecture?

Code designed for the Fermi architecture will not run (or at a minimum, not run well) on your GeForce 8800 GT because it is too old. A GeForce GTX 400 or 500 series card, however, would work. Those newer GeForce cards have GPUs based on the Fermi architecture. In the same way, code for the Fermi architecture will work on the Tesla C2050 or C2070, but not the Tesla C1060.

You won’t be able to run Magma on an 8800GT. Magma includes double precision floating point code and requires a compute capability 1.3,2.0 or 2.1 card.

Thanks for your reply. Can I use Single precision floating point functions for the time being or taking source of single precision function and making my own library.

You would have to ask the developers of the library about that, but I suspect that the answer is no.

Thanks for your reply.