English(EN) Spent way too much time trying to turn `float*` into a proper tensor library. But it finally works: single-header, plain C99, GPU acceleration, capable of train

开发者创建带 GPU 加速的 C99 张量库

作者 PulseAugur 编辑部 · [1 个来源] · 2026-07-14 09:49

一位开发者创建了一个单头文件、纯 C99 的张量库，提供 GPU 加速，在某些任务上的性能可与 PyTorch 相媲美。该库旨在训练各种神经网络架构，包括 DNN、CNN 和 LSTM。未来的开发计划包括增加对 OpenCL、量化和 transformer 模型支持。 AI

影响为 AI 开发提供了一个轻量级、高性能的张量库，可能对嵌入式系统或性能关键型应用有用。

排序理由这是一个针对专业库的单开发者项目，而非重大的行业发布或研究论文。

在 Mastodon — fosstodon.org 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。我们如何撰写摘要 →

报道来源 [1]

Mastodon — fosstodon.org TIER_1 English(EN) · [email protected] · 2026-07-14 09:49

Spent way too much time trying to turn `float*` into a proper tensor library. But it finally works: single-header, plain C99, GPU acceleration, capable of train

Spent way too much time trying to turn `float*` into a proper tensor library. But it finally works: single-header, plain C99, GPU acceleration, capable of training DNNs, CNNs, LSTMs - sometimes even beats pytorch on performance! Still lots of things to add: OpenCL, quantisation, …

链接 zserge.com/…/tensor github.com/…/utensil

报道来源 [1]

Spent way too much time trying to turn `float*` into a proper tensor library. But it finally works: single-header, plain C99, GPU acceleration, capable of train

相关实体

相关话题