About 2.85x: Arc B580 runs a large AI model faster with XeStrata
One test lifts speed from 10.9 to 31.1 tokens per second, with important memory conditions.
A test showed that new software makes an Intel graphics card write an AI model's answers about three times faster. It needs a computer with a lot of memory.
Hype check
The noise matches the real change.
Our editorial opinion, based on the sources below. How we score
Your call: hype or real?
Real = it will matter in daily use. Hype = louder than it is. One vote, and it can’t be changed.
Our view and readers’ views are opinions, not statements of fact about a product.
Why it matters
It suggests better software can help Intel cards run models on the device.
Watch: we wait for other tests to confirm it.
The details
- XeStrata is an open-source inference engine derived from Strata that uses XMX units on Intel GPUs.
- The test used an ASRock Arc B580 12GB with a Core Ultra 5 250KF Plus, 64GB RAM and Ubuntu 26.04.
- The model was Qwen3.8-Flash-Next at IQ3_S quantisation with 64K context.
- Result: 31.1 per second against 10.9 with Strata 0.2.1.
- The model needs about 50GB of system RAM, and the result covers text generation only, not a general .
Receipts
One trusted outlet reports it. Not confirmed yet. Open the sources and check us.
Written with the help of AI tools and checked against the sources above. How we work
See a mistake? Tell us and we’ll fix it.
