nihui
nihui
使用ncnn的example复现吗 代码：https://github.com/Tencent/ncnn/blob/master/examples/yolov8.cpp 模型：https://github.com/nihui/ncnn-assets/tree/master/models
https://github.com/Tencent/ncnn/pull/5953
请尝试使用这个 pr 分支的代码测试下能否解决卡死问题 https://github.com/Tencent/ncnn/pull/5953
@1027663760 @xiaoxingbobo 临时workaround：尝试加载模型前这样设置，能不能避免问题 ```cpp #if defined _WIN32 // workaround for windows + nvidia gpu + driver > 566 const ncnn::GpuInfo& gpu_info = ncnn::get_gpu_info(gpu_device); if (gpu_info.vendor_id() == 0x10de && gpu_info.driver_id() ==...
定位到这个conv实现有可能越界写数据了 ``` Convolution [ 40, 32, 8 *16] -> [ 40, 32, 3 *8] kernel: 1 x 1 stride: 1 x 1 ``` 关联 ``` Concat 16_153 3 1 15_145_bn_mish...
另一种需要改ncnn源码 allocator.h 的workaround方法，能尽量少降速 ```cpp explicit VkWeightAllocator(const VulkanDevice* vkdev, size_t preferred_block_size = 0); // 8M ```
https://github.com/Tencent/ncnn/pull/6102
> > [@1027663760](https://github.com/1027663760) [@xiaoxingbobo](https://github.com/xiaoxingbobo) > > 临时workaround：尝试加载模型前这样设置，能不能避免问题 > > #if defined _WIN32 > > // workaround for windows + nvidia gpu + driver > 566 > > const ncnn::GpuInfo& gpu_info...
@dependabot rebase
https://github.com/Tencent/ncnn/pull/6439