The NVIDIA Accelerators family improves computational time which can dramatically reduce compute time for parallel applications, offering quicker time to solutions.
By co-locating the NVIDIA Kepler™-based GRID GPUs and computational servers in the data center so that they share rather than transfer large datasets, display refresh-rates can be greatly improved.
Peak double precision floating point performance
190 GFlops (95 GFlops/GPU)
Peak single precision floating point performance
4577 GFlops (2288 GFlops/GPU)
Number of GPUs
Memory size per board (GDDR5)
Memory bandwidth for board (ECC off)
320 GB/sec (160 GB/sec/GPU)
GPU Computing Applications
Single precision computational applications
Kepler generation, passive cooling
Compatible with ProLiant SL250s/SL270s servers with Intel Xeon E5-2600 v2 processors