Anyone try XeStrata on Intel Arc?

A user running XeStrata on a single Intel Arc B70 for about two weeks reported decode throughput gains over upstream Strata v0.1.45 across four tasks at 192k context, INT8 KV and MTP spec 4, with the same GGUF and three seeds: bugfix rose from 37.7 to 51.4 tok/s, code from 32.7 to 49.2, invoice from 31.7 to 66.2 and incident from 28.7 to 52.9. The user cautioned the figures are largely one-shot runs rather than averages over many prompts, and that answers on checkable tasks were equally correct though not token-identical. The user singled out the engine's conversation cache, which parks finished conversations' KV state in 25 GiB of RAM and retains up to 16 conversations to resume instead of re-prefilling, as the most valuable feature for agentic use.

I’ve been running it for about two weeks on a single B70, and it has become my daily driver My setup is pretty modest with Qwen3.8-Flash-Next Swift 1.5 IQ3 S . The experts mostly live in system RAM and the B70 holds the hot ones, I compared this to Qwen 3.8 27B and ended up sticking with this The main driver was the performance difference between Strata and XeStrata I had claude create some arbirtrary workloads for a quick comparison and got the following results Absolute numbers for both models through llama-swap, 192k context, INT8 KV, MTP spec 4 Strata vs XeStrata same GGUF, context, MTP, prefill and reserve, 3 seeds, thinking on Decode tok/s, upstream Strata v0.1.45 → XeStrata: ┌──────────┬────────┬──────────┐ │ Task │ Strata │ XeStrata │ ├──────────┼────────┼──────────┤ │ bugfix │ 37.7 │ 51.4 │ ├──────────┼────────┼──────────┤ │ code │ 32.7 │ 49.2 │ ├──────────┼────────┼──────────┤ │ invoice │ 31.7 │ 66.2 │ ├──────────┼────────┼──────────┤ │ incident │ 28.7 │ 52.9 │ └──────────┴────────┴──────────┘ Speed was there but should be noted that these are pretty much one shot runs, not an average over a large number of prompts. Answers on the checkable tasks were equally correct. The text was not token-identical Conversation cache This is the feature I like most for agentic use. The engine parks finished conversations’ KV state in RAM I give it 25 GiB, and retain up 16 conversations. When a conversation comes back, it resumes instead of re-prefilling: For agents and coding tools that re-send a long prefix every turn, this matters more than raw decode speed. Things I hit