Running Qwen3.8-27B-Q4 at max context on a 32 GB GPU while avoiding kvcache quantization

I can't stand kv cache quantization. Even at q8_0, I can feel the difference. But realistically, when running Qwen3.8-27B-UD-Q4_K_XL on my 32 GB GPU, I only have room for ~170k tokens (with mtp and mmproj enabled). It's a lot of context, but Qwen3.8 eats through it on xhigh effort. I've been wanting