NVIDIA Tesla P100 16GB GPU

Description

The Tesla P100 16GB GPU is a dual-slot card outfitted with the same features as its counterpart, the NVIDIA Tesla P100 12GB GPU, with the exception of the memory specifications. The P100 graphics processor is built on a 16nm process featuring Pascal architecture to power thousands of High-Performance Computing (HPC) applications within quantum chemistry, medical science, financial modeling, and many other fields. Other Tesla P100 GPUs:

Performance

Pascal-powered with 15,300 million transistors, the 16GB PCIe graphics card supports the same features as its complement. It delivers 4.7 TFLOPS of double-precision performance, 9.3 TFLOPS of single-precision performance, and 18.7 TFLOPS of half-precision performance. It also supports 3584 CUDA cores, 224 texture mapping units, and 96 ROPs to provide lightning fast performance in a wide variety of applications like object identification, speech recognition, and search result enhancement.

Memory

This GPU is the first NVIDIA GPU to support the second generation of HBM2 with error-correcting code (ECC) protection. The 16GB memory also features Chip-on-Wafer-on-Substrate technology, just like the 12GB card. The only difference between the two PCIe-based GPUs is the memory specification. The 16GB CoWoS HBM2 memory runs at 715MHz and is connected to a 4096-bit memory interface with a 732GB/s bandwidth. The GPU also runs at a frequency of 1190MHz, which can be boosted to 1329MHz.

Features

Providing the power to tackle the most computationally intensive AI and HPC applications, this is the first NVIDIA Pascal GPUs to support Page Migration Engine. This allows developers to concentrate on high-performance tuning and less on data movement management and allows applications to scale beyond the physical memory size of the GPU with virtual memory paging support. By permitting developers to see a single memory space for the whole node, unified memory technology significantly enhances efficiency. The dual-slot card draws 250W from an 8-pin power connector, but there is no display connectivity for this unit, as it’s not meant to have monitors connected to it.

Summary

As one of three variants within the NVIDIA Tesla P100 family, the Nvidia Tesla P100 16GB GPU can take on any complex workload with its Pascal architecture and advanced technology. With 16GB of memory, 3584 CUDA cores, NVIDIA’s Page Migration Engine, and HBM2 memory for maximum flexibility, this card offers substantial credentials for a number of high-performance workloads.