I have some simple questions. As far as I understand tesla cards have some equal counterparts in Geforce family which are way cheaper than tesla's. For example
9800 GTX and tesla c870 both have 128 Streaming processors.But tesla is nearly 800 dollars where 9800 GTX is around 200$.
And the new card gtx280(around $350) has 240 cores which looks like equal to tesla c1060.
So what could be the reason to buy tesla instead of cheaper equal capability geforces?
I know one difference that tesla dont have direct video output. So does this mean it wont be limited by (I think it was 5 secs?!) running time? This leads to another question.
Is running time related to direct video output?Or is it only specific for tesla?(that would make sense that they are expensive)
if I dont use my geforce for direct video output(i can put another crappy video card for display) , can my kernels pass time-limit also??
The level of tech support needed far surpasses anything you would find on forums and message boards on the internet. The people providing the support are very educated and specially trained staff. When a business is relying on their hardware providing the services that they have paid for, they need top level support to keep their investments providing dividends.
Tesla Cards are best used in conjunction with workstation motherboard/cpu combos, the point of them is, they run together, like SLI, but also make use of their cuda cores to assist the CPU (Think of taking a brain, then adding WAYYYYY bigger brains onto it) Long story short, Video Cards/CPU/Motherboard features all work TOGETHER on the task at hand, as opposed to being limited to either or.
Intel 1156 CPU = 4 cores, 8 hyperthreaded
Tesla 2075 GPU = essentially 448 Cores (not to mention the 6GB DDR5 Ram which speeds things up)
The difference would be in a supercomputer MOBO (http://ca.asus.com/en/Motherboards/Intel_Socket_1156/P7...) the 448 cores on your GPU is used hand-in-hand with CPU