NVIDIA® A100 PCIe «NEW»
NEWThe NVIDIA® A100 PCIe delivers 40GB memory, third generation tensor cores, and the ability to create up to 7 vGPU’s with NVIDIA’s Multi-Instance GPU Ampere architecture.
The A100 PCIe is currently scheduled for general availability and shipment late September 2020, however this is subject to change given the current global environment. We will up-date this page as new information is known. Contact us for a quote and to reserve units in a pre-order arrangement. XENON also has special server builds designed for the A100 – The new XENON NITRO GX29A and XENON NITRO GX49A.
Get a QuoteIn performance numbers below:
* Structural sparsity enabled;
** SXM GPUs via HGX A100 server boards; PCIe GPUs via NVLink Bridge for up to 2 GPUs.
GPU Architecture
NVIDIA Ampere
Double-Precision Performance
- FP64: 9.7 TFLOPS
- FP64 Tensor Core: 19.5 TFLOPS
Single-Precision Performance
- FP32: 19.5 TFLOPS
- Tensor Float 32 (TF32): 156 TF | 312 TFLOPS*
Half-Precision Performance
312 TFLOPS | 624 TFLOPS*
Bfloat16
312 TFLOPS | 624 TFLOPS*
Integer Performance
- INT8: 624 TOPS | 1,248 TOPS*
- INT4: 1,248 TOPS | 2,496 TOPS*
GPU Memory
40 GB hBM2
Memory Bandwidth
1.6 TB/sec
Error-Correcting Code
Yes
Interconnect Interface
- PCIe Gen4: 64 GB/ sec Third generation
- NVIDIA® NVLink®: 600 GB/sec** PCIe
Form Factor
PCIe
Multi-Instance GPU (MIG)
Up to 7 GPU instances
Max Power Consumption
250 W
Delivered Performance for Top Apps
90%
Thermal Solution
Passive
Compute APIs
CUDA®, DirectCompute, OpenCLTM, OpenACC®