NVIDIA Tesla V100, powered by NVIDIA Volta™ architecture, is the world’s most advanced data center GPU ever built to accelerate AI, HPC, and graphics applications. NVIDIA Volta, Tesla V100 offers the performance of 100 CPUs in a single GPU—enabling data scientists, researchers, and engineers to tackle challenges that once were impossible.
NVIDIA Volta features:
- Tensor Core: New compute core built for AI that performs mixed precision tensor arithmetic. Equipped with 640 Tensor cores, Volta delivers 120 TeraFLOPS of deep learning performance.
- New Architecture: With over 21B transistors, Volta is the most powerful GPU architecture ever built. It couples CUDA cores and Tensor Cores to deliver 7.5 TFLOPS of double precision, 15 TFLOPS of single precision and 120 TFLOPS of mixed precision deep learning performance.
- CoWoS with HBM2: 900 GB/s of memory bandwidth, HBM2 in Volta delivers data faster to compute engines, netting significant performance gains across diverse workloads.
- NVIDIA NVLink: The revolutionary high-speed interconnect designed to scale applications across multiple GPUs now delivers 2x higher throughput compared to previous architectures. Up to 8 Tesla V100 accelerators can be interconnected at up to 300 GB/s with NVLink.
- Volta-optimized Software: With Volta-optimized CUDA and NVIDIA Deep Learning SDK libraries such as cuDNN, NCCL, and TensorRT, industry’s most important frameworks, and applications can easily tap into the power of Volta. New Tensor Cores are incorporated into cuDNN and TensorRT allowing developers and users of the AI frameworks to use it without rewriting their applications.
The maximum efficiency mode allows data centers to achieve up to 40% higher compute capacity in each rack while maintaining the existing power budget. In this mode V100 runs at peak processing efficiency, providing up to 80% of the per-node deep learning performance at half the power consumption.
But NVIDIA isn’t abandoning Pascal just yet. The NVIDIA GPU computing platform still provides Nor-Tech’s HPC and AI clients with significant value over alternatives such as CPU-only systems. With NVIDIA Pascal and Volta GPU-based systems, clients can work faster, get results faster, and quickly realize ROI.
