|
gptkbp:instance_of
|
gptkb:Graphics_Processing_Unit
|
|
gptkbp:architectural_style
|
gptkb:Ampere
|
|
gptkbp:cooling_system
|
Active Cooling
|
|
gptkbp:form_factor
|
PC Ie
SX M4
|
|
gptkbp:gpu
|
4.0
|
|
https://www.w3.org/2000/01/rdf-schema#label
|
NVIDIA A100
|
|
gptkbp:is_a_framework_for
|
gptkb:Graphics_Processing_Unit
gptkb:Caffe
gptkb:Py_Torch
MX Net
|
|
gptkbp:is_compatible_with
|
NVIDIADGX Systems
NVIDIAHGX Systems
|
|
gptkbp:key
|
Scalability
High Throughput
High Memory Bandwidth
Multi-instance GPU
Versatile Workloads
|
|
gptkbp:manufacturer
|
gptkb:DJ
|
|
gptkbp:market
|
High-End GPU
|
|
gptkbp:network
|
gptkb:battle
|
|
gptkbp:number_of_cores
|
432
6912
|
|
gptkbp:performance
|
156 TFLOPS
19.5 TFLOPS
624 TOPS
ML Perf
SPECACCEL
|
|
gptkbp:power_consumption
|
400 W
|
|
gptkbp:predecessor
|
gptkb:NVIDIA_V100
|
|
gptkbp:ram
|
40 GB
HB M2
|
|
gptkbp:release_date
|
gptkb:2020
|
|
gptkbp:successor
|
gptkb:NVIDIA_H100
|
|
gptkbp:supports
|
gptkb:Tensor_RT
gptkb:cu_DNN
gptkb:CUDA
NVIDIANGC
|
|
gptkbp:target_audience
|
gptkb:physicist
Enterprises
Data Scientists
AI Developers
Cloud Providers
|
|
gptkbp:target_market
|
Data Centers
|
|
gptkbp:use_case
|
High-Performance Computing
AI Inference
AI Training
|
|
gptkbp:bfsParent
|
gptkb:DJ
gptkb:CUDA
|
|
gptkbp:bfsLayer
|
3
|