The A100, based on the NVIDIA Ampere architecture, is the core of NVIDIA's data center platform. The A100 delivers up to 20x higher performance compared to previous solutions. Each A100 GPU can be divided into seven instances, allowing for quick adaptation to ever-changing demands. The A100 80GB offers the world's fastest memory bandwidth (over 2 TB/s), enabling it to handle the largest models and datasets.
General characteristics | |
---|---|
Video card type | Quadro |
Model GPU | NVIDIA A100 80 GB |
Graphics Processor | A100 |
Interface | PCI-E 16x 4.0 |
Technology GPU | 7 nm |
Specifications | |
GPU frequency | 1410 MHz |
Video memory size | 80 GB |
Video memory type | HBM2E |
Video memory frequency | 1512 MHz |
Video memory bus width | 5120 bit |
Math block | |
Number of general purpose processors | 14592 |
NVIDIA Tensor Cores | 432 |
Number of texture units | 432 |
Additionally | |
The need for additional power | 8-pin |
Recommended PSU Wattage | 700 W |
TDP | 300 W |
Number of occupied slots | 2 |