AMD Radeon RX 7900 XTX Surpasses NVIDIA GeForce RTX 4090 in DeepSeek’s AI Inference Tests: A Guide to Running R1 on Your Local AMD Systems

AMD’s Radeon RX 7900 XTX is showing impressive capabilities in handling the DeepSeek R1 AI model, outpacing NVIDIA’s GeForce RTX 4090 in various inference benchmarks.

AMD Leads Charge with Enhanced Support for DeepSeek’s R1 Model

DeepSeek’s latest AI model has made quite a splash in the tech world. While many are curious about the kind of computational heft required to train such a model, it turns out that even everyday users can get remarkable performance from it thanks to AMD’s "RDNA 3" Radeon RX 7900 XTX. AMD has demonstrated DeepSeek’s R1 inference benchmarks, comparing their top-tier RX 7000 series GPU to NVIDIA’s offering and highlighting better results across several models.

DeepSeek performs exceptionally well on the @AMDRadeon 7900 XTX. Discover how to deploy on Radeon GPUs and Ryzen AI APUs here: pic.twitter.com/5OKEkyJjh3 — David McAfee (@McAfeeDavid_AMD) January 29, 2025

For AI tasks, consumer GPUs have proven valuable for many, mainly because they historically offered good performance per dollar compared to standard AI accelerators. Running models on a local machine also means your privacy remains intact, an important factor given the privacy concerns surrounding DeepSeek’s AI models. Luckily, AMD has released a detailed guide for operating DeepSeek R1 distillations on their GPUs, outlined as follows:

1. Ensure you’re using the 25.1.1 Optional Adrenalin driver or later.
2. Download LM Studio version 0.3.8 or newer from lmstudio.ai/ryzenai.
3. Install LM Studio, bypassing the onboarding screen.
4. Navigate to the discover tab.
5. Select your DeepSeek R1 Distill. Smaller models like Qwen 1.5B offer lightning-fast performance (a great starting point), while larger models deliver superior reasoning capabilities.
6. On the right side, select “Q4 K M” quantization and click “Download.”
7. After downloading, return to the chat tab, choose the DeepSeek R1 distill from the menu, and ensure “manually select parameters” is checked.
8. In the GPU offload layers, push the slider to maximum.
9. Click model load.
10. Engage with a reasoning model that runs entirely on your local AMD setup!

If these steps aren’t yielding results, AMD has provided a handy YouTube tutorial that breaks down each step. Be sure to watch it and confidently execute DeepSeek’s language models on your local AMD hardware, keeping your data safe and secure. With NVIDIA and AMD’s upcoming GPUs, we anticipate a significant boost in inferencing power, thanks to dedicated AI engines that will further support such demanding tasks.