8 Pull requests merged by 4 people

• check vulkan fp16 uniform support and implement lfp conversion without fp16u

  #5287 merged Jan 17, 2024

• fix cast armv7 sigbus when loading fp16 model

  #5292 merged Jan 17, 2024

• pnnx convert some cudnn conv2d variants

  #5289 merged Jan 17, 2024

• pnnx handle index_put with empty indices and scalar values

  #5288 merged Jan 16, 2024

• Add Dimensity 9300 MT6989 ncnn benchmark

  #5284 merged Jan 15, 2024

• add orangepi zero2 ncnn benchmark

  #5277 merged Jan 12, 2024

• Add Xeon Phi 3120A results

  #5276 merged Jan 12, 2024

• pnnx model binary over 4g

  #5274 merged Jan 11, 2024

1 Pull request opened by 1 person

• disable signal based detectisa if being debugged

  #5280 opened Jan 12, 2024

3 Issues closed by 3 people

• ncnn::create_gpu_instance() 析构时发生崩溃

  #5291 closed Jan 17, 2024

• ncnn是不是不支持索引和切片？

  #5261 closed Jan 13, 2024

• The C compiler identification is unknown 找不到编译器

  #5275 closed Jan 11, 2024

10 Issues opened by 9 people

• 编译Debug版本时候出错误

  #5295 opened Jan 17, 2024

• 使用mingw编译ncnn源码时，出现LONG PVOID等未定义

  #5294 opened Jan 17, 2024

• pnnx export not support reshape op

  #5293 opened Jan 17, 2024

• Android7.1平台运行模型同时进行录音会出现录音异常问题

  #5290 opened Jan 16, 2024

• 是否支持在国产化电脑上部署ncnn呢？比如麒麟、统信

  #5286 opened Jan 15, 2024

• pnnx转换模型，pnnx模型输出与torch 原模型输出不一致

  #5285 opened Jan 15, 2024

• Build for Ubuntu16.04 ： why protobuf not find，是未找到protobuf 吗

  #5283 opened Jan 14, 2024

• Problem after last update

  #5282 opened Jan 12, 2024

• Convert efficientnetv2-s_in21k-pre_3rdparty_in1k from onnx to ncnn issue

  #5281 opened Jan 12, 2024

• NCNN Int8 quantizated model load failed.

  #5279 opened Jan 12, 2024

1 Unresolved conversation

Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.

• 同一模型结构 使用cpu可以正常推理使用gpu会直接拉满内存程序卡死

  #5259 commented on Jan 17, 2024 • 2 new comments