Qwen3.8-27B and Radeon R9700 and 13-year-old Pentium = actual good performance

A developer running Qwen3.8-27B on an ASUS Turbo Radeon AI Pro R9700 with 32 GB of GDDR6 VRAM in a 13-year-old Pentium G3420 system reports good performance, achieving usable speeds for code analysis tasks. The setup uses Ubuntu 26.04 on a USB SSD, llama.cpp Vulkan build b10586, and Unsloth Dynamic 3.0 GGUF quants, with the model outperforming expectations despite the outdated CPU and motherboard.

Updated August 23, 2026: new Unsloth GGUF files and updated benchmark results. It’s time to go big or go home I’ve bought ASUS Turbo Radeon AI Pro R9700, an RDNA4 graphics card with a whopping 32 GB of GDDR6 VRAM, right before yet another price increase. Sadly, the only desktop computer I have is an ancient custom build with a 4th-gen Intel Pentium G3420, 8 GB DDR3 RAM, and a single PCIe 3.0 slot. The CPU socket is damaged; the PC works only when I push down on the CPU cooler while booting so that POST won’t fail . I don’t have any SATA disks anymore, and of course NVMe wasn’t a thing back in the Haswell days. Yesterday, a new Qwen3.8 model was released. It’s a dense model with 27 billion active parameters, vision support, and a built-in MTP multi-token prediction layer. If benchmarks are to be believed, it’s better than Claude Opus 4.6. I have some real-world experience with Qwen3.6‑35B‑A3B, the sparse / MoE version that runs at acceptable speeds on a laptop iGPU, so expectations are really high. Let’s try it out. First impressions I’ve installed Ubuntu 26.04 on a USB SSD and downloaded all the updates. This cursed setup works just fine, even if some apps report that there may be compatibility issues: Warning: radv is not a conformant Vulkan implementation By the way, a USB‑C cable matters A LOT. I have two such cables, both with no logos or markings. The first one is too slow, at only 42 MB/s. The other one is 10x faster, but every time you look at it wrong, it would lose a connection and the OS dies. I had to update VBIOS to make the graphics card quiet. This update changes the minimum fan speed from 30% to 12% about 1000 RPM . Unfortunately, VBIOS can be updated only on Windows, but you can temporarily install Microsoft malware directly on SSD https://gist.github.com/Alee14/e8ce6306a038902df6e7a6d667544ac9 without losing any data. There is no official AMD control panel for Linux, but you can use a wonderful LACT app https://flathub.org/en/apps/io.github.ilya zlobintsev.LACT instead. I’ve set the fan curves to start at 12% and never exceed 40%. You can also change power limits. The default is 300 W, but I haven’t noticed any performance issues at 210 W, and it helps keep the temperatures below 85°C without thermal throttling. Qwen3.8 has multiple reasoning modes, with xhigh being the default. It behaves like a caveman, which is a downgrade for me. With Qwen3.6 I could easily analyze the reasoning and refine the prompt early if the model hallucinated. So it’s helpful to switch to a medium reasoning mode. We need answer user question. Need analyze code, explain functionality, rate quality. We need think in English same language. Need inspect code carefully. … Testing Flathub’s moderation code https://github.com/flathub-infra/website/blob/1430248c92c81b485f8cbf5fc57340924ae55ff0/backend/app/moderation.py is approximately 2350 lines of text including empty lines in a 90 kB file, which is about 19700 input tokens. I don’t think the model has been benchmaxxed against this file, so it should show the real performance. How long would it take to churn through a large chunk of code and answer these two questions: What does this code do? How would you rate it? I’ve limited the output to 2048 tokens and applied other llama-cli parameters: --load-mode none --n-gpu-layers all --flash-attn on --fit off --ctx-size 131072 --cache-type-k q8 0 --cache-type-v q8 0 --temperature 1.0 --top-k 20 --top-p 0.95 --min-p 0 --repeat-penalty 1 --presence-penalty 0 --reasoning on --reasoning-effort medium --spec-type draft-mtp --spec-draft-type-k q8 0 --spec-draft-type-v q8 0 --spec-draft-n-max 2 --n-gpu-layers-draft all Tested with the official llama.cpp Vulkan build b10586 https://github.com/ggml-org/llama.cpp/releases/tag/b10586 , AMD Lemonade llamacpp-rocm build b1314 https://github.com/lemonade-sdk/llamacpp-rocm/releases/tag/b1314 , and Unsloth Dynamic 3.0 GGUF quants https://huggingface.co/unsloth/Qwen3.8-27B-GGUF , with 128k context size and an average MTP acceptance rate of 0.65 at 2 tokens ahead. Sampler parameters are from the model card. Note that Q8 K XL quant won’t fit in 32 GB VRAM, so Q8 K L is used instead. Results Wide tables can be scrolled on mobile. | Quant | Vulkan pp/s | Vulkan tg/s | Vulkan VRAM GB | ROCm pp/s | ROCm tg/s | ROCm VRAM GB | || |---|---|---|---|---|---|---|---|---| Without MTP | |||||||| | Q4 K M | 755 | 26 | 19.4 | 809 | 24 | 19.9 | || With MTP | |||||||| | Q4 K M | 648 | 39 | 20.1 | 766 | 36 | 20.6 | || | Q4 K XL | 645 | 37 | 21.1 | 743 | 34 | 21.6 | || | Q6 K L | 621 | 32 | 28.2 | 619 | 29 | 28.6 | || | Q8 K L | 650 | 30 | 30.5 | 726 | 29 | 31.0 | Observations People report they can also generate about 30-40 tokens per second on a modern hardware. Looks like you can buy a powerful graphics card, pair it with mediocre components, and still be happy. No need to pay extra for a modern CPU and expensive DDR5 RAM, yet. Qwen3.8 is based on a battle-tested qwen35 architecture and requires no custom patches for llama.cpp. It doesn’t crash, it has valid reasoning but way too long and with primitive language , and responses are correct and coherent. MTP works, resulting in huge performance improvements on Vulkan. ROCm has faster prompt processing, but token generation speed is slightly slower, and it requires a little more VRAM. I think it’s not worth trying to use 6-bit quants. Their performance is similar to 8-bit quants, if not worse, and it’s not like I’m going to run out of RAM. What didn’t work AMD claims that Qwen3.8‑27B is capable of up to 50tg/s https://www.amd.com/en/blogs/2026/run-qwen-3-8-27b-on-amd-ryzen-ai-max-and-radeon-graphics-cards-day-0.html on a single R9700 on Windows. But it doesn’t work for me. Both Vulkan and ROCm fail to allocate memory when loading a model. I’ve tried hipfire, ROCmFPX, and vLLM with various container runtimes, but they do not work since Pentium G3420 doesn’t support AVX. Tough luck.