IT Creations Partners

NVIDIA Tesla P100 12GB GPU - System Overview

Description

NVIDIA Tesla P100 12GB GPU Front

The Tesla P100 12GB PCIe GPU has a full-height form factor and is packed with 3584 NVIDIA CUDA cores. The P100 is the first NVIDIA GPU with a 16nm manufacturing process and Gen2 High Bandwidth Memory (HBM2). Thousands of High-Performance Computing (HPC) applications are accelerated by NVIDIA Tesla graphics cards in several fields, including medical science, financial modeling, quantum chemistry and many others. There are two other versions of the NVIDIA Tesla V100:

NVIDIA Tesla P100 16GB GPU

NVIDIA Tesla P100 SXM2 GPU

Performance

The Tesla P100 is the first full-size NVIDIA GPU based on the 16nm manufacturing process. It also delivers about 4.7 TFLOPS of double-precision floating point performance, 9.3 TFLOPS of single-precision performance, and 18.7 TFLOPS of half-precision performance. The NVIDIA Tesla P100 12GB GPU supports 3584 CUDA cores, 224 texture mapping units, and 96 ROPs. With power like this, the graphics card enables a single node to replace half a rack of CPU nodes in a wide variety of applications by providing lightning fast performance. Applications include identifying objects in images, recognizing speech, and enhancing search results.

Memory

This is the first NVIDIA GPU to feature the second generation of High Bandwidth Memory (HBM2). HBM2 memory supports error-correcting code memory (ECC) protection for increased reliability. Memory on the NVIDIA P100 12GB GPU is driven by Chip-on-Wafer-on-Substrate (CoWoS) with HBM2 technology offering 3x the memory performance over the previous Maxwell generation architecture. The 12GB CoWoS HBM2 memory runs at 715MHz and is connected to a 3072-bit memory interface with a 549GB/s bandwidth. The GPU also runs at a frequency of 1190MHz, which can boost up to 1329MHz.

Features

NVIDIA Tesla P100 12GB GPU Power Connection

Installed in many of the world’s top supercomputers to accelerate complex workloads, the Tesla P100 is one of the first NVIDIA Pascal GPUs to include Page Migration Engine. This hardware supports Unified Memory page faulting and migration, which allows applications to go beyond the GPU’s physical memory size to virtually limitless levels for efficient processing. No hardware outputs are provided because the NVIDIA P100 PCIe GPU is a server GPU and not designed for powering monitors directly. The card has no moving parts and uses a passive heatsink cooler, for increased reliability and reduced power consumption. This design is dependent on the server’s internal airflow to cool the GPU, including the heat from the 250W of power that the card uses to operate.

Summary

As one of the first GPUs to lead innovative technology, the NVIDIA Tesla P100 12GB GPU delivers maximum performance and flexibility for a diverse set of applications. With 3584 NVIDIA CUDA cores, HBM2 memory, and Page Migration Engine, this GPU can easily help researchers and scientists complete complex projects previously thought unsolvable.

Check Availability

NVIDIA Tesla P100 12GB GPU - Specifications

Memory
    GPU Memory: 12 GB HBM2

  • Memory Interface: 3072-bit
  • Memory Bandwidth: Up to 549.1 GB/s
Cores
  • NVIDIA CUDA Cores: 3,584
Performance
    FP16 (half) performance:

  • 19.05 TFLOPS (2:1)
  • FP32 (float) performance

  • 9.526 TFLOPS
  • FP64 (double) performance

  • 4.763 TFLOPS (1:2)
System Interface
  • PCIe 3.0 x16
Power Consumption
  • 250W
Form Factor
  • Board Height: 4.4"
  • Board Length: 10.5"
  • Dual Slot
Engine
    Compute APIs

  • CUDA
  • DirectCompute
  • OpenCL
  • OpenACC

Check Availability

Get a quote for NVIDIA Tesla P100 12GB GPU

If you know what you want but can't find the exact configuration you're looking for, have one of our knowledgeable sales staff contact you. Give us a list of the components you would like to incorporate into the system, and the quantities, if more than one. We will get back to you immediately with an official quote.

[email protected]

NVIDIA Tesla P100 12GB GPU - Documentation

Check Availability