Nvidia has announced its PCIe Tesla P100 graphics processing unit (GPU) for servers, set for release in the last quarter of the year.
According to Nvidia, the latest addition to its Tesla Accelerated Computing Platform enables the creation of "super nodes" that provide the throughput of more than 32 commodity CPU-based nodes and deliver up to 70 percent lower capital and operational costs.
"Accelerated computing is the only path forward to keep up with researchers' insatiable demand for HPC and AI supercomputing," Ian Buck, vice president of accelerated computing at Nvidia, said.
"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 Tesla P100-powered nodes puts more dollars into computing instead of vast infrastructure overhead."
The Tesla P100 is available in a PCIe form factor and is compatible with most GPU-accelerated servers.
The company said a single Tesla P100-powered server delivers higher performance than 50 CPU-only server nodes when running the AMBER molecular dynamics code, and is faster than 32 CPU-only nodes when running the VASP material science application.
It delivers 4.7 teraflops double-precision performance, 9.3 teraflops single-precision performance, and 18.7 teraflops half-precision performance with Nvidia GPU Boost technology.
The Tesla P100 GPU accelerator is available in two configurations: 16GB of CoWoS HBM2 stacked memory; delivering 720GB/sec of memory bandwidth and 12GB of CoWoS HBM2 stacked memory; and delivering 540GB/sec of memory bandwidth.
Last month, Nvidia revealed its latest GeForce GTX 1080 GPU offering, which the chip giant said is faster than both the GeForce GTX 980 Ti and GeForce GTX TITAN X with three times the efficiency, thanks to its 9 teraflops of performance and packing 8 gigabytes of GDDR5X memory.
The GeForce GTX 1080 is the first gaming GPU based on Nvidia's Pascal architecture, which the technology giant said is purpose-built to be the engine of computers that learn, see, and simulate the world.
"These gains in performance and power efficiency are enabled by the marvels of the Pascal architecture," the company said previously. "The first of these marvels is the introduction of cutting-edge 16nm FinFET chip construction. This new, smaller chip design uses fewer watts of power and emits less heat, enabling us to crank up the core clock speed of the GPU, which is key for increasing a graphics card's performance."