Today Exxact Corporation announced its planned production of HPC solutions using the NVIDIA Tesla P100 GPU accelerator for PCIe. Exxact will be integrating the Tesla P100 into their Quantum family of servers, which are currently offered with either NVIDIA Tesla M40 or K80 GPUs. The NVIDIA Tesla P100 for PCIe-based servers was introduced at the recent 2016 International Supercomputing Conference and is anticipated to deliver massive leaps in performance and value compared with CPU-based systems. NVIDIA stated the new Tesla P100 will help meet unprecedented computational demands planted on modern data centers.
Accelerated computing is the only path forward to keep up with researchers’ insatiable demand for HPC and AI supercomputing,” said Roy Kim, senior group product manager of accelerated computing at NVIDIA. “Deploying CPU-only systems to meet this demand would require large numbers of commodity compute nodes, leading to substantially increased costs without proportional performance gains. Dramatically scaling performance with fewer, more powerful NVIDIA Tesla P100 powered nodes puts more dollars into computing instead of vast infrastructure overhead.”
HPC data centers need to support the growing demands of scientists and researchers while staying within a tight budget. The old approach of deploying lots of commodity compute nodes requires vast interconnect overhead that substantially increases costs without proportionally increasing data center performance.
With Exxact’s Quantum Series servers featuring the Tesla P100 GPUs, scientists and researchers will be able to optimize the total cost of ownership for highly-demanding computing requirements with a solution that provides reliability, serviceability, and density.The NVIDIA Tesla P100 accelerators are the world’s most advanced data center GPUs ever built; designed to boost throughput and save money for HPC and hyperscale data centers. Powered by the brand new NVIDIA Pascal™ architecture, NVIDIA Tesla P100 enable a single PCIe-based node to replace up to half-rack of commodity CPU nodes by delivering lightning-fast performance in a broad range of HPC applications. Handling the same workload with far fewer nodes means customers can save up to 70% in overall data center costs.
With over 400 applications accelerated, including 9 out of top 10 HPC applications, every HPC customer can now deploy accelerators in their data center to keep up with ever-growing user demands while saving money.
NVIDIA Tesla P100 for PCIe is reimagined from silicon to software, crafted with innovation at every level. It features four groundbreaking technologies that deliver a dramatic jump in performance:
- New Pascal Architecture: Delivering 4.7 and 9.3 TeraFLOPS of double and single precision for HPC, 18.7 TeraFLOPS of FP16 for deep learning
- CoWoS with HBM2: Unifying data and compute into single package for up to 3X memory bandwidth over prior-generation solution.
- Page Migration Engine: Parallel programming got simpler by enabling datasets beyond the physical limits of GPU memory.
Tesla P100 Accelerator Specification:
- 4.7 TeraFLOPS double-precision performance with NVIDIA GPU Boost
- 9.3 TeraFLOPS single-precision performance with NVIDIA GPU Boost
- 18.7 TeraFLOPS half-precision performance with NVIDIA GPU Boost
- Up to 720 GB/s memory bandwidth with CoWoS HBM2 Stacked Memory
- Up to 16 GB of CoWoS HBM2 Stacked Memory
- Enhanced Programmability with Page Migration Engine and Unified Memory
- ECC protection for increased reliability
- Server-optimized for best throughput in the data center
The NVIDIA Tesla P100 GPU accelerator for PCIe will be available beginning in Q4 2016 from Exxact Corporation. Exxact Quantum Series servers featuring the Tesla P100 will also be available in Q4 2016. Users that do not plan on going taking the data center route or wish to test the NVIDIA Tesla P100 before the Tesla P100 variant for PCIe ships may consider NVIDIA’s P100 powered DGX-1 supercomputer. It features eight NVIDIA Tesla P100 accelerators, interconnected with NVLink, delivering 170 teraFLOPS of half-precision peak performance, which is equivalent to 250 CPU-based servers.