Hopper (microarchitecture)

<p>Hopper is a <a href="/facts/Graphics_processing_unit/PTK1RQVp">graphics processing unit</a> (GPU) <a href="/facts/Microarchitecture/NFRd4NUL">microarchitecture</a> developed by <a href="/facts/Nvidia/8QfCgqbu">Nvidia</a>. It is designed for datacenters and is used alongside the <a href="/facts/Ada_Lovelace_(microarchitecture)/mb9xGgK6">Lovelace</a> microarchitecture. It is the latest generation of the line of products formerly branded as <a href="/facts/Nvidia_Tesla/SsHo1w98">Nvidia Tesla</a>, now Nvidia Data Centre GPUs.
</p><p>Named for computer scientist and <a href="/facts/United_States_Navy/xYK0fUou">United States Navy</a> <a href="/facts/Rear_admiral/3f3BNjuy">rear admiral</a> <a href="/facts/Grace_Hopper/MEhbGazn">Grace Hopper</a>, the Hopper architecture was leaked in November 2019 and officially revealed in March 2022. It improves upon its predecessors, the <a href="/facts/Turing_(microarchitecture)/7fJP3rM3">Turing</a> and <a href="/facts/Ampere_(microarchitecture)/GIAEqPWS">Ampere</a> microarchitectures, featuring a new <a href="/facts/Streaming_Multiprocessor/PTK1RQVp">streaming multiprocessor</a>, a faster memory subsystem, and a <a href="/facts/Transformer_(deep_learning_architecture)/cDbjx6a8">transformer</a> acceleration engine.
</p>

Hopper (microarchitecture) open-in-new

Hopper (microarchitecture)