alt Hacker NewsSince it might be helpful to some, here's my current commandline for llama.cpp running on an RTX 4090 with my monitor moved to the iGPU to free up all of its VRAM.
llama-server -m Qwen3.8-27B-IQ4_NL.gguf --mmproj mmproj-BF16.gguf -c 170000 --parallel 1 -ngl -1 --cache-type-k q8_0 --cache-type-v q8_0 -b 1024 -ub 512 --flash-attn on --no-context-shift --no-mmproj-offload --spec-type draft-mtp --spec-draft-n-max 5 --spec-default --cache-type-k-draft q4_0 --cache-type-v-draft q4_0 --threads 24 --jinja --reasoning on -fit off
Identical to the qwen3.6 config. With a prompt like "svg owl" (which can reuse quite a lot compared with creative writing or similar, so ngram-mod shines), I get about 70-80t/s like this, with a memory overclock of about 1.5GHz
Replies
Hm, I have a 4090 as well, and:
$ build/bin/llama-server -m Qwen3.8-27B-IQ4_NL.gguf --mmproj mmproj-BF16.gguf -c 170000 --parallel 1 -ngl -1 --cache-type-k q8_0 --cache-type-v q8_0 -b 1024 -ub 512 --flash-attn on --no-context-shift --no-mmproj-offload --spec-type draft-mtp --spec-draft-n-max 5 --spec-default --cache-type-k-draft q4_0 --cache-type-v-draft q4_0 --threads 24 --jinja --reasoning on -fit off
0.02.993.689 E ggml_backend_cuda_buffer_type_alloc_buffer: allocating 911.53 MiB on device 0: cudaMalloc failed: out of memory
Update: Oh, it works after I stop Xorg. But nvidia-smi only showed Xorg using 200M out of the 24G, so why would a 911M alloc fail?
Any chance I could run it on a GeForce RTX 5060 Ti 16 GB, (64 gb ram)
Thanks for posting! Have you had any success with running without kv cache quantization? Is there a noticeable difference in quality without any? I would assume that would eat into context but 170k is pretty generous!
Do you find it useful or worthwhile to split a large LLM across two GPUs on a desktop?
If you've tried it, what worked well and what didn't? I'm especially interested in mismatched VRAM setups, e.g. a 16 GB GPU + a 24 GB GPU.
How much overhead did you see from inter-GPU transfers, and did the extra usable VRAM outweigh the performance hit?
Lol at that command. Why is this stuff so hard to run locally? I've spent a few days trying to figure it all out and haven't been able to. LM Studio doesn't work behind proxies. Ollama is confusing and doesn't seem to support Qwen3? And Llama.cpp is your command.
I just want to run `<some-command> <model-name>` with some default parameters set and for it to run locally.
> --cache-type-k q8_0 --cache-type-v q8_0
In my tests, even Q8 quantization for the KV cache comes with notable drops in performance for longer tasks. It does provide more context length in limited RAM budgets, but the longer context tasks are where KV quantization starts to show problems. It’s basically unnoticeable for simple and short tasks.
> --spec-draft-n-max 5
5 is a lot of tokens to draft. Are you really seeing acceptance rates to support that? When I tested it, 2-3 was the peak. Anything more started reducing performance except on highly predictable short outputs.