AMD Strengthens Its AI Position: Ryzen AI MAX+ 395 Outperforms RTX 5080 in DeepSeek R1 Benchmarks
AMD presented Ryzen AI MAX+ 395, demonstrating it significant advantage before NVIDIA RTX 5080 in AI performance tests. New APU Strix Halo с 128GB of shared storage offers 3 times faster processing speed tokens in LM Studio when working with large language models.
The key factor of superiority is the Strix Halo capability use all 128GB of memory for processing AI tasks, while RTX 5080 limited to 16GB VRAM. This allows the AMD chip cope effectively with models that require a large amount of memory.
In real-world testing, the Ryzen AI MAX+ 395 demonstrates 2,5-3 times better performance compared to the RTX 5080 in models with more parameters. At the same time, Energy consumption the new APU is total 55 WWhereas RTX 5080 consumes 360W.
AMD is pushing Strix Halo as a powerful AI computing solution for laptops, competing not only with NVIDIA but also with Intel. The new processor supports up to 50 TOPS of computing power based on the XDNA 2 NPU, and also allows you to use up to 96 GB of variable video memory.
The Ryzen AI MAX+ 395 is expected to be a major milestone in the evolution of AI integration in consumer laptops and Ultrabooks, allowing for local execution of Llama, DeepSeek, and Google Gemma models without a cloud connection.
