NVIDIA GPU architectures
NVIDIA GPU architectures are the chip design generations, such as Ampere, Hopper, and Blackwell, that define a GPU's era and features.
NVIDIA GPU architectures are the microarchitecture generations that NVIDIA ships its chips in. Each generation sets the transistor process, core design, and feature set for every card built on it. A new one arrives roughly every one to two years and spreads across the full product line, from workstation cards to data center accelerators.
Ampere launched in 2020 and covers the RTX 30 series along with the A100 and A6000. Ada Lovelace followed in 2022 with the RTX 40 series, the RTX 6000 Ada, and the L40. Hopper, also from 2022, is data center only, spanning the H100 and H200. Blackwell arrived in 2024 and 2025 across the RTX 50 series, the B100, B200, and GB200.
When specing a system, the architecture name tells you generation level performance, memory bandwidth, and feature support such as FP8 or FP4. Check it before comparing individual model numbers.
Sources
Source | Publisher |
|---|---|
NVIDIA | |
NVIDIA | |
NVIDIA |
- Publisher
NVIDIA
- Publisher
NVIDIA
- Publisher
NVIDIA
Last verified August 29, 2026.
- nvidia ampere
- ada lovelace
- hopper architecture
- blackwell architecture
- nvidia gpu generations
- rtx 40 series architecture
- gpu microarchitecture
- nvidia chip architecture