Can the Instinct MI50 32GB run Darwin-27B-RSI?
Darwin-27B-RSI·26.9B·Instinct MI50 32GB·32 GB VRAM
Yes. The Instinct MI50 32GB runs Darwin-27B-RSI fully in VRAM at Q6_K (about 26 GB of 32 GB), at roughly 25 tokens/s.
- best quant
- Q6_K
- needs
- 26 GB
- usable
- 32 GB
- est. speed
- ~25 tok/s
Estimates for an 8K context; offload assumes 32 GB of system RAM.Try other contexts on the Darwin-27B-RSI page →
Every quantisation of Darwin-27B-RSI on the Instinct MI50 32GB
| Quant | Memory | Fit | Speed |
|---|---|---|---|
| Q2_K | 13 GB | fits | ~64 tok/s |
| Q3_K_M | 15 GB | fits | ~43 tok/s |
| Q4_K_M | 19 GB | fits | ~35 tok/s |
| Q5_K_M | 23 GB | fits | ~29 tok/s |
| Q6_K pick | 26 GB | fits | ~25 tok/s |
| Q8_0 | 31 GB | tight | ~20 tok/s |
| FP16 / BF16 | 55 GB | offload | ~1.4 tok/s |
Speed is a bandwidth-based estimate for generating text (1024 GB/s); prompt processing and long contexts are slower.
Measured speed on the Instinct MI50 32GB
No community reports for this pair yet.
Report your own speed →Run Darwin-27B-RSI on the Instinct MI50 32GB
Using Q6_K, the best fit for Instinct MI50 32GB.
ollama run hf.co/bartowski/FINAL-Bench_Darwin-27B-RSI-GGUF:Q6_KOllama’s ROCm/CUDA builds don’t cover this GPU. Recent Ollama releases use Vulkan when its Vulkan backend and driver are installed (older ones needed OLLAMA_VULKAN=1); llama.cpp or LM Studio with Vulkan are the safer bet.
llama-server -hf bartowski/FINAL-Bench_Darwin-27B-RSI-GGUF:Q6_KNeeds a Vulkan build of llama.cpp. Downloads the GGUF on first run.
docker run -d --name llama --device /dev/dri -p 8080:8080 \
-v ~/.cache/llama.cpp:/root/.cache/llama.cpp \
ghcr.io/ggml-org/llama.cpp:server-vulkan \
-hf bartowski/FINAL-Bench_Darwin-27B-RSI-GGUF:Q6_K --host 0.0.0.0 --port 8080 -ngl 99Server and web UI on port 8080.
lms get bartowski/FINAL-Bench_Darwin-27B-RSI-GGUF@Q6_KThen load it from the LM Studio app or with `lms load`.
from transformers import pipeline
pipe = pipeline("text-generation", model="FINAL-Bench/Darwin-27B-RSI", dtype="auto", device_map="auto")
messages = [{"role": "user", "content": "Explain self-hosting in one paragraph."}]
out = pipe(messages, max_new_tokens=512)
print(out[0]["generated_text"][-1]["content"])Chat models that fit the Instinct MI50 32GB comfortably
All that fit →| Model | Params | Best quant | Memory | Speed |
|---|---|---|---|---|
| 20.9BA4.2B | Q8_0 | ≈13 GBFits in memory | ~126 tok/s | |
| 31.2BA4.6B | Q5_K_M | ≈29 GBFits in memory | ~172 tok/s | |
| 32.8B | Q6_K | ≈29 GBFits in memory | ~21 tok/s | |
| 36BA3B | Q5_K_M | ≈26 GBFits in memory | ~264 tok/s | |
| 30.5BA3B | Q6_K | ≈26 GBFits in memory | ~228 tok/s | |
| 24BA2B | Q8_0 | ≈26 GBFits in memory | ~265 tok/s |
Other GPUs that run Darwin-27B-RSI
Frequently asked questions
- Can the Instinct MI50 32GB run Darwin-27B-RSI?
- Yes. The Instinct MI50 32GB runs Darwin-27B-RSI fully in VRAM at Q6_K (about 26 GB of 32 GB), at roughly 25 tokens/s.
- How fast does Darwin-27B-RSI run on the Instinct MI50 32GB?
- An estimate from memory bandwidth gives about 25 tokens/s at Q6_K.
- Which quantisation of Darwin-27B-RSI should I use on the Instinct MI50 32GB?
- Q6_K: the largest one that fits. Expected quality: near-lossless compared with the full-precision weights.
Memory and speed are estimates from the model's size and the card's memory bandwidth; real numbers depend on the runtime, drivers and context length. Community figures are medians of user reports.