NVIDIA’s Kepler architecture offers 1,536 cores which is an increase of over 1000 cores compared to the previous Fermi architecture core count. This GPU takes advantage of 2nd Generation PCIe x16 connection interface with data transfer speeds of 8GB/s which is double the 1st Generation PCIe transfer speeds. Data Science and AI Modeling has been improved with FP32 2.491 TFLOPS and FP64 103.8 GFLOPS.
The Tesla K8 supports 8GB of GDDR5 memory with Error Correcting Code (ECC) which can be enabled or disabled. It has a memory bandwidth of 160GB/s and a clock speed of 2.5GHz. This GPU has a memory interface of 256-bit.
Maximum power consumption on this card is rated at 100W and uses a single 6-pin power connector. Active cooling on the NVIDIA K8 GPU is provided by an active fan sink.
Offering impressive performance and reliability, the NVIDIA Tesla K8 GPU is ideal for servers and embedded devices that have limited power or space constraints. It is a fantastic choice for general purposes GPU applications at a low power budget of 100W. 8GB of GDDR5 memory and 1,536 cores are available on this card.