Amazon has started offering servers with Nvidia Volta V100 GPUs
Amazon has started offering servers with Nvidia Volta V100 GPUs
What is most interesting about Nvidia’s Volta V100 GPU is its ability to offer 14TFLOPs of single-precision performance and offer 1:2:4 FP64-FP32-FP16 performance scaling, which can be very handy depending on your desired workloads. One of the most interesting additions to Nvidia’s design is the addition of what Nvidia calls Tensor cores, which are purpose-built to calculate FP16 4×4 matrix operations, giving this GPU a performance advantage when computing neural workloads, which is a handy feature for those in the AI market.Â
 Each of the NVIDIA GPUs is packed with 5,120 CUDA cores and another 640 Tensor cores and can deliver up to 125 TFLOPS of mixed-precision floating point, 15.7 TFLOPS of single-precision floating point, and 7.8 TFLOPS of double-precision floating point. On the two larger sizes, the GPUs are connected together via NVIDIA NVLink 2.0 running at a total data rate of up to 300 GBps. This allows the GPUs to exchange intermediate results and other data at high speed, without having to move it through the CPU or the PCI-Express fabric.
Â
 Â
 Adding their new Volta-based V100 series GPUs into Amazon’s AWS (Amazon Web Services) ecosystem is a great design win for Nvidia, bringing more money into their coffers to fund the development of future hardware/chip designs.Â
At the time of writing, it is unknown when Nvidia’s Volta architecture will come into the hands of consumers, though to say the least most of the advanced features offered by the V100 are almost worthless to PC gamers, as Tensor cores and FP32 compute are not utilised by gaming applications.Â
You can learn more about Amazon’s new P3 servers here.Â
You can join the discussion on Nvidia’s Volta-based V100 GPU coming to Amazon’s Web Services on the OC3D Forums.Â