News from the AI & ML world
DeeperML
@insidehpc.com
//
MiTAC Computing Technology and AMD are strengthening their partnership to deliver cutting-edge solutions for AI, HPC, cloud-native, and enterprise applications. MiTAC will showcase this collaboration at COMPUTEX 2025, highlighting their shared vision for scalable and energy-efficient technologies. This partnership, which began in 2002, leverages AMD EPYC processors and Instinct GPUs to meet the evolving demands of modern data centers. Rick Hwang, President of MiTAC Computing Technology, emphasized their excitement in advancing server solutions powered by AMD's latest processors and GPUs, stating that it's key to unlocking new capabilities for their global customer base in AI and HPC infrastructure.
Specifically, MiTAC and AMD are developing next-generation server platforms. One notable product is an 8U server equipped with dual AMD EPYC 9005 Series processors and support for up to 8 AMD Instinct MI325X GPUs, offering exceptional compute density and up to 6TB of DDR5-6400 memory, ideal for large-scale AI model training and scientific applications. Additionally, they are offering a 2U dual-socket GPU server, that supports up to four dual-slot GPUs with 24 DDR5-6400 RDIMM slots and tool-less NVMe storage carriers, it offers high-speed throughput and flexibility for deep learning and HPC environments.
Meanwhile, Nvidia is preparing to compete with Huawei in the Chinese AI chip market by releasing a budget-friendly AI chip. This strategy is driven by the need to maintain relevance in the face of growing domestic competition and navigate export restrictions. The new chip, priced between $6,500 and $8,000, represents a significant cost reduction compared to the previously banned H20 model. This reduction involves trade-offs, such as using Nvidia's RTX Pro 6000D foundation with standard GDDR7 memory and foregoing Taiwan Semiconductor's advanced CoWoS packaging technology.
ImgSrc: insidehpc.com
References :
Specifically, MiTAC and AMD are developing next-generation server platforms. One notable product is an 8U server equipped with dual AMD EPYC 9005 Series processors and support for up to 8 AMD Instinct MI325X GPUs, offering exceptional compute density and up to 6TB of DDR5-6400 memory, ideal for large-scale AI model training and scientific applications. Additionally, they are offering a 2U dual-socket GPU server, that supports up to four dual-slot GPUs with 24 DDR5-6400 RDIMM slots and tool-less NVMe storage carriers, it offers high-speed throughput and flexibility for deep learning and HPC environments.
Meanwhile, Nvidia is preparing to compete with Huawei in the Chinese AI chip market by releasing a budget-friendly AI chip. This strategy is driven by the need to maintain relevance in the face of growing domestic competition and navigate export restrictions. The new chip, priced between $6,500 and $8,000, represents a significant cost reduction compared to the previously banned H20 model. This reduction involves trade-offs, such as using Nvidia's RTX Pro 6000D foundation with standard GDDR7 memory and foregoing Taiwan Semiconductor's advanced CoWoS packaging technology.
ImgSrc: insidehpc.com
Share:
References :
- insideAI News: MiTAC Computing Technology Corporation, a server platform designer and manufacturer, will showcase its strategic collaboration with AMD at COMPUTEX 2025 (Booth M1110).
- www.artificialintelligence-news.com: Nvidia is preparing to go head-to-head with Huawei to maintain its relevance in the booming AI chip market of China.
- insidehpc.com: MiTAC Computing Technology Corporation, a server platform designer and manufacturer, will showcase its strategic collaboration with AMD at COMPUTEX 2025 (Booth M1110).
Classification:
- HashTags: #MiTAC #AMD #AIInfrastructure
- Company: MiTAC Computing Technology
- Target: Data Centers
- Product: AI Chips
- Feature: AI Infrastructure
- Type: AI
- Severity: Informative