MTIA 300: Meta’s First Training Chip with Built-in NICs and Communication-Offloading Engines

Source: Engineering at Meta
Image: Engineering at Meta

Summary

MTIA 300 is the first of Meta’s family of in-house training and inference accelerators optimized for training ranking and recommendation models. We’re sharing how MTIA 300’s built-in NIC chiplets allow it to meet the communication needs associated with training recommendation models with superior performance over general-purpose GPUs. By co-designing MTIA’s communication library, HCCL, alongside the [...] Read More... The post MTIA 300: Meta’s First Training Chip with Built-in NICs and Communication-Offloading Engines appeared first on Engineering at Meta.

Read this on Engineering at Meta
Opens in a new tab. Subvolts summarizes and links; the full piece belongs to Engineering at Meta.
Where the other five stand

Related: OpenAI: Introducing GPT-6 Astra: the most intelligent and aligned model in the world. · Google: Run Ray on TPU, Part 2: Ray AI libraries · Anthropic: GPT‑6 Astra · Microsoft: Meet MAI-Transcribe-2: A faster and more accurate speech recognition model · xAI: Sam Altman "AGI by December"

Hype check
2/5Worth a look

Rated low: routine. Worth knowing, not worth rearranging your day for.

Published
Source
Engineering at Meta (engineering.fb.com)
Company
Meta · Official · Developer
Summary by
Subvolts, using an extract from the source (how we work). Spotted a mistake? Tell us.

Questions people ask

Where can I read the full story?
On Engineering at Meta. The "Read this on Engineering at Meta" link above opens the original in a new tab. Subvolts publishes a summary and analysis, never the full piece.
What does this mean for Meta AI?
MTIA 300 is the first of Meta’s family of in-house training and inference accelerators optimized for training ranking and recommendation…

More from Engineering at Meta 5 more

Page generated Sep 3, 2026. Summaries are Subvolts' own; the story belongs to Engineering at Meta.