AMD Instinct MI300X meets the minimum VRAM requirement for Q4 inference of codellama/CodeLlama-34b-hf. Review the quantization breakdown below to see how higher precision settings impact VRAM and throughput.
AMD Instinct MI300X can run codellama/CodeLlama-34b-hf with Q4 quantization. At approximately 286 tokens/second, you can expect Excellent speed - conversational response times under 1 second.
You have 175GB headroom, which is sufficient for system overhead and smooth operation.
| Quantization | VRAM needed | VRAM available | Estimated speed | Verdict |
|---|---|---|---|---|
| Q4 | 17GB | 192GB | 286.45 tok/s | ✅ Fits comfortably |
| Q8 | 35GB | 192GB | 183.93 tok/s | ✅ Fits comfortably |
| FP16 | 70GB | 192GB | 101.45 tok/s | ✅ Fits comfortably |