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

AMD’s Radeon RX 7900 XTX is setting new standards by outperforming NVIDIA’s GeForce RTX 4090 in running the DeepSeek R1 AI model during inference benchmarks.

### AMD Leading the Way with DeepSeek’s R1 Model

The latest offering from DeepSeek has captivated the tech world, leaving many curious about the computational power required for its operation. However, it appears that AMD’s “RDNA 3” Radeon RX 7900 XTX GPU is accessible enough for the average user to achieve optimal performance. AMD has released impressive benchmark results for their flagship RX 7000 series GPU, which dominates in performance when compared to NVIDIA’s equivalent across a variety of models.

For those looking to utilize consumer GPUs for AI purposes, AMD’s setup offers a noteworthy solution, providing excellent performance at a reasonable price compared to conventional AI hardware. Running models locally also ensures user privacy, a significant consideration with DeepSeek’s offerings. Microsoft’s detailed guide walks you through setting up DeepSeek R1 distillations on AMD GPUs, detailed below:

1. Ensure your system runs the 25.1.1 Optional Adrenalin driver or newer.
2. Download LM Studio version 0.3.8 or later from lmstudio.ai/ryzenai.
3. Install LM Studio, and bypass the onboarding screen.
4. Access the discover tab.
5. Select your desired DeepSeek R1 Distill. Start with the Qwen 1.5B for speed, though larger distills offer enhanced reasoning capabilities.
6. On the right, ensure “Q4 K M” quantization is selected, then click “Download”.
7. Post-download, switch to the chat tab, select the DeepSeek R1 distill, and check “manually select parameters.”
8. Maximize the GPU offload layers using the slider.
9. Load the model.
10. You can now run a reasoning model entirely on your local AMD setup!

If you hit any bumps using these steps, AMD has prepared a detailed tutorial on YouTube breaking down each part of the process. Watching this can help you maximize the potential of DeepSeek’s LLMs on your AMD hardware, protecting your data from external misuse. As AMD and NVIDIA gear up to release future GPUs, expect a significant boost in inferencing capabilities, driven by built-in dedicated AI engines tailored for such tasks.