NVIDIA H100 is the flagship Hopper-generation GPU announced in 2022, and it quickly became the standard hardware for frontier-scale model training. With its Transformer Engine, FP8 precision support and 80GB of HBM3 memory, it delivers up to 3x speedups over the A100 on certain workloads; GPT-4, Claude, Gemini and Llama 3 were all largely trained on H100 clusters. NVLink and NVSwitch allow thousands of H100s to be wired together as a single logical system, ushering AI infrastructure into the 'supercomputer' era. Its scarcity through 2023-2024 turned into a global GPU crunch and made supply itself a strategic issue.
MEVZU N°124ISTANBULYEAR I — VOL. III
Glossary · Beginner · 2022
NVIDIA H100
The Hopper-generation GPU launched in 2022 — the standard hardware for frontier model training.
- EN — English term
- NVIDIA H100
- TR — Turkish term
- NVIDIA H100