Create

Sign in to ReadmeX

or

How we handle your data: Privacy Policy

ReadmeX

Your AI reading companion

What would you like to catch up on?

Pick a briefing, or ask your own question.

Your community and people briefings stay personal to you.

Story

DeepX details DX-M2 edge AI chip

AI summary

DeepX announced that its next-generation DX-M2 edge AI chip received a CES 2027 artificial-intelligence category innovation award. According to IT之家, the chip uses Samsung Foundry’s 2nm process and includes an 80 TOPS NPU with support for up to 128-bit LPDDR5X-9600 memory. DeepX claims it can run LLM and VLM models with more than 10 billion parameters locally at 5W. The company also says DX-M2 has completed mask tapeout, physical chips are expected in early 2027, and a packaged integrated-memory version is planned.

In-depth insight

The announcement provides concrete process, memory, power, tapeout and availability details for DX-M2, along with the planned integrated-memory package version.

IT之家 11 月 11 日消息,韩国边缘能效 AI 芯片开发商 DEEPX 昨日宣布,其下一代芯片 DX-M2 荣获 CES 2027 人工智能类别创新奖。

DX-M2 基于三星晶圆代工 2nm 工艺,集成 80TOPs NPU 算力,至高支持 128bit 的 LPDDR5X-9600 内存,可本地运行超 10B 参数的 LLM 和 VLM 模型,同时功耗仅有 5W。除单芯片外,DX-M2 还将提供封装集成内存的版本。

DEEPX 表示,DX-M2 近日成功实现“光罩(掩膜)流片”,这为 2027 年初首批芯片的面世打下了基础。

#DeepX#DX-M2#Samsung Foundry

Original · 1 min readRead the original
Group chat

No comments yet. Start the conversation.