Tesla K80 global bandwidth specs

http://www.anandtech.com/show/8729/nvidia-launches-tesla-k80-gk210-gpu

As I understand it, this is two GPUs in one? How does this differ from the Titan-Z, and can I utilize directly the stated 480 GBs without having to deal with spliting the problem between two units?

Also not sure what is meant by the ‘memory pool’. I really need more GBs…

http://www.nvidia.com/object/tesla-servers.html
Memory bandwidth (ECC off)² 480 GB/sec (240 GB/sec per GPU)
Memory size (GDDR5) 24 GB (12GB per GPU)

One difference to previous high-end solutions that were based on GK110 is that K80 uses a new GPU called GK210. I am unable to find specifics on the NVIDIA site, but one third-party website notes:

http://www.techpowerup.com/207265/nvidia-breathes-life-into-kepler-with-the-gk210-silicon.html
“While both chips are based on the “Kepler” architecture, GK210 features double the shader cache amount. Each of the 15 streaming multiprocessors (SMXs) features 128 KB of shader cache, compared to 64 KB per SMX on the GK110. The GK210 also has a 512 KB register file per SMX, double the size of the 256 KB register file size, of the GK110.”

Really exciting stuff, particularly in conjunction with the developments of NV-Link which will go great lengths towards creating a seamless memory pool between many gpus :D

You can check out the relative performance of K80 vs Titan Z for one particular application, the molecular simulation application AMBER, here:

http://ambermd.org/gpus/benchmarks.htm#Benchmarks