以下部分引用 AI 总结，现阶段 AI 仍然有幻觉。请以内容的原文为准。

数据新鲜度群聊 07-14 ✓卖方 07-14 ✓Wrap 断档15天 ✕News 07-15 ✓X 断档12天 ✕页面生成 07-15

Intel Xe-Forge

4 atoms · 跨 1 天 · 首见 2026-05-27 · 最近 2026-05-27

三色: 🟦 fact 4 · 🟥 take 0 · stance ▲2/▼0/◆0

来源: X 4

时态: fresh:4

标签: 好数字:3 · 好信源:3

2026-05-27 ⭐ · geometric mean speedup
The evaluation results of Xe-Forge on 97 Level-2 KernelBench kernels and Flash Attention on the Intel Arc Pro B70 achieve a 1.17× geometric mean speedup over PyTorch eager, with 67% of kernels improving.
tags: 好数字·好信源 · → daily
value: qty=1.17× · date=evaluation on 97 Level-2 KernelBench kernels and Flash Attention on Intel Arc Pro B70
2026-05-27 ⭐ · nine kernels exceed 5× speedup
nine kernels exceed 5× (up to 82×), and Flash Attention achieves 2×–13.3× speedups across all tested configurations without regression.
tags: 好数字·好信源 · → daily
value: qty=up to 82×
2026-05-27 · applies up to nine optimization stages
Xe-Forge applies up to nine optimization stages, from algorithmic restructuring and operator fusion through block-pointer modernization, GPU-specific tuning, and open-ended discovery.
tags: 好数字 · → daily
value: qty=9 stages
2026-05-27 · optimization stage driven by CoVeR agent
Each optimization stage is driven by a Chain-of-Verification-and-Refinement (CoVeR) agent that iteratively generates optimizations, verifies them against runtime compilation, correctness checks, and validates them on real hardware, and refines them based on error feedback.
tags: 好信源 · → daily