NVIDIA has officially launched the world’s most powerful GPU computing architecture named Volta. It has been created to drive the next wave of advancement in AI and high-performance computing. Besides, the company has also unveiled its first Volta-based processor- the NVIDIA Tesla V100 data center GPU. It delivers extraordinary speed and scalability for AI inferencing and training along with accelerated HPC and graphics workloads.
Volta is NVIDIA’s 7th generation GPU architecture and is built with 21 billion transistors. Its performance is equivalent to 100 CPUs for deep learning. It’s 5 times better than Pascal, the current generation NVIDIA GPU architecture, in terms of teraflops and 15 times better than Maxwell architecture. The overall performance is 4 times improved than Moore’s law would have predicted.
NVIDIA Launches GPU Computing Architecture named Volta (Image Courtesy: NVIDIA)
The Tesla V100 GPU carries some groundbreaking technologies that surpass the performance of previous generations GPUs, thereby allowing the shattering of the 100 teraflops barrier of deep learning performance. These include:
• Tensor Cores equipped with 640 Tensor Cores, V100 delivers 120 teraflops of deep learning performance.
• New GPU architecture with over 21 billion transistors pairs CUDA cores and Tensor Cores within a unified architecture, providing the performance of an AI supercomputer in a single GPU.
• NVLink™ provides the next generation of high-speed interconnect linking GPUs, and GPUs to CPUs, with up to 2x the throughput of the prior generation NVLink.
• 900 GB/sec HBM2 DRAM achieves 50 percent more memory bandwidth than previous generation GPUs, essential to support the extraordinary computing throughput of Volta.
• Volta-optimized software, including CUDA, cuDNN and TensorRT™ software, which leading frameworks and applications can easily tap into to accelerate AI and research.
The new Volta platform will become the new standard for high-performance computing. Pairing CUDA® cores and the new Volta Tensor Core within a unified architecture will enable a single server with Tesla V100 GPUs to replace hundreds of commodity CPUs for traditional HPC.
Filed Under: News