The Cloud-Bound Isle

Get the best builds by email. No account needed.

Explore

Apps & Tools

The Cloud-Bound Isle

Optimized SYCL kernels for Qwen models on Intel XPU hardware

byAJ

Is this your creation?

Open source

Description

A collection of high-performance SYCL and SYCL-TLA kernels for vLLM, optimizing Qwen3.8-Flash-Next and Qwen4-27B on Intel GPUs. The project introduces fused operators and quantization support (FP8/INT4), delivering up to 1.8x faster prefill times as verified by comprehensive benchmarks, with implementation assistance from OpenAI Codex.

Descriptions, tags, and model credits may be AI-generated or inferred from public sources and can be incomplete or wrong. Learn more.

Get the best builds by email. No account needed.

Explore

Apps & Tools