NVIDIA A100 is the flagship data-center GPU of the Ampere generation, announced in 2020 — and it became the de facto standard of AI compute for several years. With 40 or 80GB of HBM2 memory, the A100 carried both the training and serving of large models, underpinning GPT-3, the early Llama family and many other landmark systems. Even after H100 arrived, A100s remained popular in production thanks to their price-performance and 'mature, available' status. Today they are still the backbone of many startups' small-to-medium-scale training and inference workloads.
MEVZU N°124ISTANBULYEAR I — VOL. III
Glossary · Beginner · 2020
NVIDIA A100
NVIDIA's Ampere-generation GPU launched in 2020, the workhorse of deep learning for years.
- EN — English term
- NVIDIA A100
- TR — Turkish term
- NVIDIA A100