pchardware.org
Categories

NVIDIA · 2017

NVIDIA V100 GPU accelerator

The NVIDIA V100 is a high-end GPU accelerator based on the Tesla architecture, released in 2017 for data center and AI workloads.

The NVIDIA V100 GPU accelerator, released in December 2017, is built on the Tesla architecture with the GV100 chip on a 12 nm process. It features 5120 shading units and 640 Tensor/AI cores, with base and boost clocks of 1246 MHz and 1530 MHz respectively. The card is equipped with 32 GB of HBM2 memory on a 4096-bit bus, delivering 900 GB/s bandwidth. It has a TDP of 300 W and requires a single 8-pin power connector. The V100 uses a PCIe 3.0 x16 interface and has no display outputs, as it is designed for compute acceleration. It succeeded the NVIDIA P100 and was succeeded by the NVIDIA A100.

Specifications

Architecture unverified Tesla
GPU chip unverified GV100
Process node unverified 12 nm
Shading units unverified 5120
Tensor / AI cores unverified 640
Base clock unverified 1246 MHz
Boost clock unverified 1530 MHz
Memory size unverified 32 GB
Memory type unverified HBM2
Memory bus unverified 4096 bit
Memory bandwidth unverified 900 GB/s
Board power (TDP) unverified 300 W
Power connectors unverified 1x 8-pin
Bus interface unverified PCIe 3.0 x16
Display outputs unverified None
Release date unverified 2017-12-07
Release year unverified 2017
Launch MSRP unverified 10000 USD
Successor unverified NVIDIA A100
Predecessor unverified NVIDIA P100