nvidia cuda gpu calculation

At the end of past weeks NVIDIA declared about the beginning of public beta- testing of CUDA, pack for developers, earlier than the accessible to the small circle to the company.

This means that in the near future it is worthwhile to expect the appearance of different IN, the achieving calculation not on CPU and on GPU. Graphic processor video card geForce 8800 GTX can reach peak performance in 520 gigaflop (GFLOPS, billion of operations with floating point in second) thus SLI- system can cross mark in teraflop.

Because of technology CUDA parallel computational possibilities GPU are now accessible to practically all developers through the standard language programming C. the developers they can download the beta- version CUDA SDK and C compilator for Windows XP and Linux (RedHat release 4 Update 3) from NVIDIA site .

In the user medium with the aid of CUDA will be able to be calculated physics in games. Among the more serious tasks - the calculations in seismology, finances, hydrodynamics, medicine and the like also graphic card can simulate the work of neurons brain, the behavior of electromagnetic wavesâ€¦

It is necessary to note that in the comparison with traditional narrowly specialized coprocessors, intended for fulfilling operations with floating point (Clearspeed CSX600 and similar), the accelerators NVIDIA have essential deficiencies. They are capable of working only in 32- bit, but not 64- bit regime, giving the result of calculations with the single accuracy, instead of required dual. Company says that this deficiency will be removed in accelerators, which will leave more lately in the present year.

By the way, the same limitation is characteristic of the processor tera-Scale, developed by intel.

nvidia CUDA GPU calculation

NVIDIA CUDA GPU CALCULATION