AMD Instinct MI250X meets the minimum VRAM requirement for Q4 inference of google/embeddinggemma-300m. Review the quantization breakdown below to see how higher precision settings impact VRAM and throughput.
AMD Instinct MI250X can run google/embeddinggemma-300m with Q4 quantization. At approximately 605 tokens/second, you can expect Excellent speed - conversational response times under 1 second.
You have 127GB headroom, which is sufficient for system overhead and smooth operation.
| Quantization | VRAM needed | VRAM available | Estimated speed | Verdict |
|---|---|---|---|---|
| Q4 | 1GB | 128GB | 605.34 tok/s | ✅ Fits comfortably |
| Q8 | 1GB | 128GB | 427.66 tok/s | ✅ Fits comfortably |
| FP16 | 1GB | 128GB | 236.10 tok/s | ✅ Fits comfortably |