【计算机体系架构和算子优化】
【请简述SIMD和SIMT的区别,以及在AI算子开发中如何利用它们提高吞吐】
https://www.doubao.com/thread/xGE0dB0uq4uvLnUMfhttps://www.doubao.com/thread/xGE0dB0uq4uvLnUMf【昇腾芯片的达芬奇架构(DaVinciArchitecture)中,Cube单元和Vector单元分别负责什么任务】
https://www.doubao.com/thread/xTFbSnue7tm3j730mhttps://www.doubao.com/thread/xTFbSnue7tm3j730m【如何在代码层面减少访存延迟?请结合L1/L2Cache缓存一致性进行说明】
https://www.doubao.com/thread/x9Rln5RmTc9feN3iwhttps://www.doubao.com/thread/x9Rln5RmTc9feN3iw【在CUDA中,什么是BankConflict?如何有效规避】
https://www.doubao.com/thread/xvMepYkVvEGWakkhMhttps://www.doubao.com/thread/xvMepYkVvEGWakkhM【简述算子融合(OperatorFusion)的原理,它能解决AI模型推理中的哪些瓶颈】
https://www.doubao.com/thread/xEvXDU5MnlRYYhE8Bhttps://www.doubao.com/thread/xEvXDU5MnlRYYhE8B
【请解释什么是Tiling技术,并说明在昇腾CANN中如何进行切分优化】
https://www.doubao.com/thread/xPi9zZq3SxBAsRT8Thttps://www.doubao.com/thread/xPi9zZq3SxBAsRT8TDeepSeekChat with DeepSeek AI – your intelligent assistant for coding, content creation, file reading, and more. Upload documents, engage in long-context conversations, and get expert help in AI, natural language processing, and beyond. | 深度求索(DeepSeek)助力编程代码开发、创意写作、文件处理等任务,支持文件上传及长文本对话,随时为您提供高效的AI支持。
https://chat.deepseek.com/share/adojev9waz8j9a9jedDeepSeekChat with DeepSeek AI – your intelligent assistant for coding, content creation, file reading, and more. Upload documents, engage in long-context conversations, and get expert help in AI, natural language processing, and beyond. | 深度求索(DeepSeek)助力编程代码开发、创意写作、文件处理等任务,支持文件上传及长文本对话,随时为您提供高效的AI支持。
https://chat.deepseek.com/share/q7utzaesi6euef89n7
DeepSeekChat with DeepSeek AI – your intelligent assistant for coding, content creation, file reading, and more. Upload documents, engage in long-context conversations, and get expert help in AI, natural language processing, and beyond. | 深度求索(DeepSeek)助力编程代码开发、创意写作、文件处理等任务,支持文件上传及长文本对话,随时为您提供高效的AI支持。https://chat.deepseek.com/share/htdxs38e65izm60sy6
【请解释什么是Tiling技术,并说明在昇腾CANN中如何进行切分优化】
https://www.doubao.com/thread/xPi9zZq3SxBAsRT8Thttps://www.doubao.com/thread/xPi9zZq3SxBAsRT8TDeepSeekChat with DeepSeek AI – your intelligent assistant for coding, content creation, file reading, and more. Upload documents, engage in long-context conversations, and get expert help in AI, natural language processing, and beyond. | 深度求索(DeepSeek)助力编程代码开发、创意写作、文件处理等任务,支持文件上传及长文本对话,随时为您提供高效的AI支持。
https://chat.deepseek.com/share/adojev9waz8j9a9jedDeepSeekChat with DeepSeek AI – your intelligent assistant for coding, content creation, file reading, and more. Upload documents, engage in long-context conversations, and get expert help in AI, natural language processing, and beyond. | 深度求索(DeepSeek)助力编程代码开发、创意写作、文件处理等任务,支持文件上传及长文本对话,随时为您提供高效的AI支持。
https://chat.deepseek.com/share/q7utzaesi6euef89n7
DeepSeekChat with DeepSeek AI – your intelligent assistant for coding, content creation, file reading, and more. Upload documents, engage in long-context conversations, and get expert help in AI, natural language processing, and beyond. | 深度求索(DeepSeek)助力编程代码开发、创意写作、文件处理等任务,支持文件上传及长文本对话,随时为您提供高效的AI支持。https://chat.deepseek.com/share/htdxs38e65izm60sy6
【针对半精度(FP16)运算,硬件层面是如何进行溢出处理和精度补偿的】
https://www.doubao.com/thread/x8f2Dlryx99EfRVyWhttps://www.doubao.com/thread/x8f2Dlryx99EfRVyW【如何评估一个算子的算力利用率(Utilization)?你会关注哪些核心指标?】
https://www.doubao.com/thread/xD6V2ZACKNhisD976https://www.doubao.com/thread/xD6V2ZACKNhisD976【C++中智能指针的实现原理是什么?在多线程环境下使用shared_ptr是否安全?】
https://www.doubao.com/thread/xYRBLCEnJlhZPYRUrhttps://www.doubao.com/thread/xYRBLCEnJlhZPYRUr【请解释虚函数表(vtable)的结构,并说明多态在底层是如何实现的】
https://www.doubao.com/thread/xPx2cVwRnRsFG4yxOhttps://www.doubao.com/thread/xPx2cVwRnRsFG4yxO【如何使用C++实现一个高性能的内存池】
https://www.doubao.com/thread/xhmBOBXsp1AP0DbUyhttps://www.doubao.com/thread/xhmBOBXsp1AP0DbUy