- AMD is developing the “Ryzen AI Halo,” a high-performance Mini-PC tailored for local AI development.
- This device aims to challenge Apple’s Mac Studio and Nvidia’s specialized hardware by offering a Windows-based alternative for running LLMs.
- The system is powered by the Ryzen AI Max+ 395, featuring a 16-core processor.
New information indicates that AMD is preparing to launch the Ryzen AI Halo. Essentially, it’s a Mini-PC based on the Strix Halo platform, designed for local AI developers.
The AMD team is attempting to carve out a niche for itself among established manufacturers in this industry, as well as in more closed systems such as the NVIDIA DGX Spark. Because of Macs’ unified memory architecture, Mac Studios and Mac Minis are increasingly being purchased by more traditional users to run AI locally rather than by AI enthusiasts.
In terms of hardware, the AMD Ryzen AI Halo is almost identical to any other computer. The difference is that it provides augmented consumer hardware for AI. Of course, this is the ideal excuse for AMD to offer consumer technology originally meant for high-performance laptops and Mini-PCs at AI hardware pricing.
The AMD Ryzen AI Halo includes the AMD Ryzen AI Max+ 395 APU. This features a 16-core Zen 5 CPU with 32 threads. This CPU has a base frequency of 3.00 GHz and can turbo up to 5.10 GHz. It contains 16 MB of L2 cache, 64 MB of L3 cache, a base TDP of 55 W, and a customizable cTDP of 45-120 W.
TSMC manufactured it using a 4 nm technology, and it supports AVX-512, which is useful for certain CPU workloads. However, the key selling point of this platform is its combination of CPU, iGPU, NPU, and unified memory. This combination makes it a feasible choice for running AI locally.
In AI, the CPU, iGPU, and AMD XDNA2 NPU combine to offer up to 126 total TOPS. This isn’t especially amazing given that it’s designed to meet standards such as Copilot+. What is genuinely important for huge LLMs is the inbuilt iGPU, which can access nearly all of the system’s RAM.
As a result, AMD claims that its Ryzen AI Halo can run models with up to 200 billion parameters locally. Ryzen AI Max+ datasheet suggests up to 235 billion parameters in specific models and quantization setups.
The fine print is key. These figures are dependent on quantization, context, model format, and acceptable performance. In other words, it does not imply that any 200-billion-parameter model will operate at data center rates.
Thank you! Please share your positive feedback. 🔋
How could we improve this post? Please Help us. 😔
[Editor-in-Chief]
Sajjad Hussain is the Founder and Editor-in-Chief of Tech4Gamers.com. Apart from the Tech and Gaming scene, Sajjad is a Seasonal banker who has delivered multi-million dollar projects as an IT Project Manager and works as a freelancer to provide professional services to corporate giants and emerging startups in the IT space.
Majored in Computer Science
13+ years of Experience as a PC Hardware Reviewer.
8+ years of Experience as an IT Project Manager in the Corporate Sector.
Certified in Google IT Support Specialization.
Admin of PPG, the largest local Community of gamers with 130k+ members.
Sajjad is a passionate and knowledgeable individual with many skills and experience in the tech industry and the gaming community. He is committed to providing honest, in-depth product reviews and analysis and building and maintaining a strong gaming community.
Join Our Community
