DEEPX Verifies AI Inference Technology With Intel, Expands Next-Gen NPU Business
TECHWORLD ·
✦ AI Summary
DEEPX announced on the 12th at COEX in Seoul that it had signed an MOU on AI semiconductor technology cooperation with Intel Korea.
On the same day, DEEPX also signed a separate tripartite MOU involving KTNF.
The cooperation aims to integrate and optimize Intel OpenVINO and DEEPX NPU, and to establish a heterogeneous computing-based AI server cooperation framework that also includes server systems.
BizCrushWant to follow Korean talks and live streams in real time?Try CapCrush on the web
DEEPX and Intel Korea are pursuing cooperation on AI inference technology. DEEPX announced on the 12th at COEX in Seoul that it had signed an MOU on AI semiconductor technology cooperation with Intel Korea.
In addition to the MOU with Intel Korea, DEEPX also signed a separate tripartite MOU involving KTNF on the same day. KTNF is a domestic server specialist.
The purpose of the tripartite MOU is to establish a cooperation framework for AI servers based on heterogeneous computing that encompasses AI semiconductors, software, and server systems. Through this, the scope of cooperation has expanded to server systems.
The key area of cooperation is the integration and optimization of Intel OpenVINO and DEEPX NPU. OpenVINO is an open-source software toolkit that supports the efficient execution of AI models across a variety of computing environments.
The two companies will review compatibility between AI model conversion and execution environments for using DEEPX NPU within the OpenVINO environment, and will seek optimization measures. The plan is to create a development environment that makes it easier to apply DEEPX NPU while preserving the use of existing AI models and development tools.
The cooperation will expand beyond existing software integration to physical AI and edge AI platforms. The two companies plan to verify the technical potential of heterogeneous computing platforms by combining Intel x86 processors' host and control functions with DEEPX NPU's perception and inference functions.
In the process, the two companies will pursue demonstrations in robotics and seek ways to expand practical applicability in the robotics sector. The cooperation targets also include the development of next-generation generative AI solutions, and they will review the use of DEEPX's next-generation generative AI NPU, DX-M2, now under development. They will also review the feasibility of enterprise AI appliances and on-device AI services.
DX-M2 is an AI semiconductor currently under development based on Samsung Electronics' 2-nanometer process. DX-M2 aims to run large language models (LLM) and vision-language models (VLM) in a low-power environment of about 5W. Through this, DEEPX plans to lay the groundwork for generative AI use in edge environments where power and space are limited.
The tripartite agreement with KTNF involved has significance in expanding the scope of cooperation to server systems. Through this, the companies are pursuing heterogeneous computing-based cooperation aimed at applying AI semiconductors and software in real server environments, in a move that seeks system-level applicability beyond verification at the individual chip level.
DEEPX said CEO Kim Nok-won met with members of Intel's OpenVINO development team in Silicon Valley in the U.S. last October. At that meeting, the two sides discussed integrating DEEPX's NPU with OpenVINO and ways to optimize AI models.
Source: TECHWORLD · Park Kyu-chan
Original: https://www.epnc.co.kr/news/articleView.html?idxno=409732
References
This article was produced with the help of an automated content generation algorithm.
Source: TECHWORLD
View originalThis article was summarized and organized by BizCrush based on the original article from TECHWORLD. For exact quotations and full details, please refer to the original article.
Keep reading
BizCrush