跳到正文
研踪PaperTrail

论文库

检索计算机领域论文,按研究方向、作者或关键词筛选。

部分采集异常
正在读取账号与研究记录…

全站更新概况

+1471新增论文
+1164其中 AI 相关
+159已有论文修订
+97新增 AI 导读

arXiv 最近完成扫描:2026-11-20 08:48 · 全站最近入库:2026-11-20 12:52各信源进度 →

统计口径与数据新鲜度

统计 2026-11-19 14:33 — 2026-11-20 14:33(北京时间) · 点击数字查看对应简报;AI 相关包含在新增论文内。

最近完成检查:2026-11-20 08:48最近发现新增版本:2026-11-20 12:52库内论文最新更新时间:2026-11-20 02:59

每分钟刷新已保存的数据。空检查不代表采集异常;新收录可能包含历史论文。AI 导读按新增版本计数。

28,342 已收录论文23,539 AI 相关8,682 有修订112 HF 社区发现

浏览论文库

时间范围跟随排序:版本更新、本站收录或首次发表。最新收录可能包含历史论文。部分信源异常,查看状态 →

找到 1 篇论文 · 按版本更新时间排序上轮扫描起点 2026-11-20 08:46(北京时间)
新论文 · v1arXiv:2611.17526

User-Schedulable Approximations for Quantized Tensor Programs

Alex Reinking

原文摘要

Quantization lets language models run on edge devices, but it sharply increases the number of tensor kernels a library must maintain: every weight format must be paired with every activation format on every instruction set and GPU, with an encoder and a decoder that agree on the stored bytes. In GGML, the tensor library used by llama.cpp, these kernels are written by hand, and only 8 of the 54 weight-activation pairs it runs on our Arm CPU have a tiled matrix-multiplication kernel. We extend Halide's separation of algorithm and schedule to the representation of values. Each approximation is a codec, an encoder paired with a decoder, composed from reusable units: quantizers, integer representations, bit packers and layouts. A new directive, approximate_by, applies a codec to one stage of a program and returns the encoder and decoder as ordinary stages that Halide can schedule, so both decoders fuse into the register tiles of a matrix product. Optional contracts, such as a round-trip law, are checked by property tests; the law exposed an inconsistent block sum in GGML's q8_1. One algorithm and one schedule library generate 347 kernels for 31 of GGML's and KleidiAI's formats on Arm CPUs and Apple GPUs, covering 38 weight-activation pairs for which GGML has no kernel, in 3.7x less code than GGML and KleidiAI. On an Apple M3 Max, our kernels reach at least 95% of the speed of GGML's format-specific kernels in 91.0% of the cases that have one, with a median speedup of 1.13x, and are faster still where GGML falls back to blocked dot products. Counting Apple's BLAS library, which is usually faster for larger f32 products, we still reach that threshold in 86.9% of CPU cases. A port to x86 with Intel's AMX adds 793 lines of code and matches GGML in 83.5% of its cases.

cs.PL · cs.LG · cs.PF