Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

CUDNN_STATUS_NOT_INITIALIZED error #4

Open
opentld opened this issue Aug 9, 2019 · 10 comments
Open

CUDNN_STATUS_NOT_INITIALIZED error #4

opentld opened this issue Aug 9, 2019 · 10 comments

Comments

@opentld
Copy link

opentld commented Aug 9, 2019

I0809 19:18:18.471977 15264 net.cpp:84] Creating Layer mobilenet0_conv25
I0809 19:18:18.477895 15264 net.cpp:406] mobilenet0_conv25 <- mobilenet0_batchnorm24
I0809 19:18:18.478863 15264 net.cpp:380] mobilenet0_conv25 -> mobilenet0_conv25
I0809 19:18:19.578912 15264 net.cpp:122] Setting up mobilenet0_conv25
I0809 19:18:19.578912 15264 net.cpp:129] Top shape: 1 256 20 20 (102400)
I0809 19:18:19.580909 15264 net.cpp:137] Memory required for data: 183910400
I0809 19:18:19.581923 15264 layer_factory.cpp:58] Creating layer mobilenet0_batchnorm25
I0809 19:18:19.587956 15264 net.cpp:84] Creating Layer mobilenet0_batchnorm25
I0809 19:18:19.588887 15264 net.cpp:406] mobilenet0_batchnorm25 <- mobilenet0_conv25
I0809 19:18:19.589884 15264 net.cpp:380] mobilenet0_batchnorm25 -> mobilenet0_batchnorm25
I0809 19:18:19.591877 15264 net.cpp:122] Setting up mobilenet0_batchnorm25
I0809 19:18:19.596879 15264 net.cpp:129] Top shape: 1 256 20 20 (102400)
I0809 19:18:19.597873 15264 net.cpp:137] Memory required for data: 184320000
I0809 19:18:19.597873 15264 layer_factory.cpp:58] Creating layer mobilenet0_batchnorm25_scale
I0809 19:18:19.599881 15264 net.cpp:84] Creating Layer mobilenet0_batchnorm25_scale
I0809 19:18:19.601862 15264 net.cpp:406] mobilenet0_batchnorm25_scale <- mobilenet0_batchnorm25
I0809 19:18:19.602859 15264 net.cpp:367] mobilenet0_batchnorm25_scale -> mobilenet0_batchnorm25 (in-place)
I0809 19:18:19.608254 15264 layer_factory.cpp:58] Creating layer mobilenet0_batchnorm25_scale
I0809 19:18:19.608254 15264 net.cpp:122] Setting up mobilenet0_batchnorm25_scale
I0809 19:18:19.609251 15264 net.cpp:129] Top shape: 1 256 20 20 (102400)
I0809 19:18:19.610247 15264 net.cpp:137] Memory required for data: 184729600
I0809 19:18:19.610247 15264 layer_factory.cpp:58] Creating layer mobilenet0_relu25
I0809 19:18:19.611243 15264 net.cpp:84] Creating Layer mobilenet0_relu25
I0809 19:18:19.614934 15264 net.cpp:406] mobilenet0_relu25 <- mobilenet0_batchnorm25
I0809 19:18:19.620775 15264 net.cpp:367] mobilenet0_relu25 -> mobilenet0_batchnorm25 (in-place)
I0809 19:18:19.625766 15264 net.cpp:122] Setting up mobilenet0_relu25
I0809 19:18:19.625766 15264 net.cpp:129] Top shape: 1 256 20 20 (102400)
I0809 19:18:19.626761 15264 net.cpp:137] Memory required for data: 185139200
I0809 19:18:19.634972 15264 layer_factory.cpp:58] Creating layer mobilenet0_conv26
I0809 19:18:19.636997 15264 net.cpp:84] Creating Layer mobilenet0_conv26
I0809 19:18:19.641957 15264 net.cpp:406] mobilenet0_conv26 <- mobilenet0_batchnorm25
I0809 19:18:19.642958 15264 net.cpp:380] mobilenet0_conv26 -> mobilenet0_conv26
F0809 19:18:19.647945 15264 cudnn_conv_layer.cpp:53] Check failed: status == CUDNN_STATUS_SUCCESS (1 vs. 0) CUDNN_STATUS_NOT_INITIALIZED

net_.reset(new Net(model_file, TEST)); running to mobilenet0_conv26
error occurs
Why???

@wzj5133329
Copy link
Owner

It may have been a problem with your caffe on cuda

@xiakj
Copy link

xiakj commented Oct 14, 2019

It may have been a problem with your caffe on cuda

Project/MobileNet-YOLO-master2/include
这包含的是什么?

@wzj5133329
Copy link
Owner

It may have been a problem with your caffe on cuda

Project/MobileNet-YOLO-master2/include
这包含的是什么?

这个是安装caffe的路径。。。

@xiakj
Copy link

xiakj commented Oct 15, 2019

It may have been a problem with your caffe on cuda

Project/MobileNet-YOLO-master2/include
这包含的是什么?

这个是安装caffe的路径。。。

我在Windows10上配置的caffe,将您的代码放到工程里面,编译通过了,在执行的时候报错:
I1015 09:42:25.221655 11960 net.cpp:406] mobilenet0_conv21 <- mobilenet0_batchnorm20
I1015 09:42:25.221655 11960 net.cpp:380] mobilenet0_conv21 -> mobilenet0_conv21
F1015 09:42:25.300750 11960 cudnn_conv_layer.cpp:53] Check failed: status == CUDNN_STATUS_SUCCESS (1 vs. 0) CUDNN_STATUS_NOT_INITIALIZED
*** Check failure stack trace: ***

请问这是什么原因?谢谢!

@wzj5133329
Copy link
Owner

It may have been a problem with your caffe on cuda

Project/MobileNet-YOLO-master2/include
这包含的是什么?

这个是安装caffe的路径。。。

我在Windows10上配置的caffe,将您的代码放到工程里面,编译通过了,在执行的时候报错:
I1015 09:42:25.221655 11960 net.cpp:406] mobilenet0_conv21 <- mobilenet0_batchnorm20
I1015 09:42:25.221655 11960 net.cpp:380] mobilenet0_conv21 -> mobilenet0_conv21
F1015 09:42:25.300750 11960 cudnn_conv_layer.cpp:53] Check failed: status == CUDNN_STATUS_SUCCESS (1 vs. 0) CUDNN_STATUS_NOT_INITIALIZED
*** Check failure stack trace: ***

请问这是什么原因?谢谢!

你在配置caffe时需要根据自己电脑的capacity设置 CUDA_ARCH

@xiakj
Copy link

xiakj commented Oct 15, 2019

It may have been a problem with your caffe on cuda

Project/MobileNet-YOLO-master2/include
这包含的是什么?

这个是安装caffe的路径。。。

我在Windows10上配置的caffe,将您的代码放到工程里面,编译通过了,在执行的时候报错:
I1015 09:42:25.221655 11960 net.cpp:406] mobilenet0_conv21 <- mobilenet0_batchnorm20
I1015 09:42:25.221655 11960 net.cpp:380] mobilenet0_conv21 -> mobilenet0_conv21
F1015 09:42:25.300750 11960 cudnn_conv_layer.cpp:53] Check failed: status == CUDNN_STATUS_SUCCESS (1 vs. 0) CUDNN_STATUS_NOT_INITIALIZED
*** Check failure stack trace: ***
请问这是什么原因?谢谢!

你在配置caffe时需要根据自己电脑的capacity设置 CUDA_ARCH

我的环境是win10+cunda10.1+rtx2060。我首先配置的caffe-windows,成功。然后将你的这个工程添加到caffe工程下的examples目录下新建的项目,然后单独编译这个项目,通过。在运行的时候,执行
/* Load the network. */
net_.reset(new Net(model_file, TEST));
到这里的时候,就是解析模型文件,就报这个错误。
Rtx2060显存是6G的,应该也够用吧,调用的是mnet.prototxt。

@wzj5133329
Copy link
Owner

It may have been a problem with your caffe on cuda

Project/MobileNet-YOLO-master2/include
这包含的是什么?

这个是安装caffe的路径。。。

我在Windows10上配置的caffe,将您的代码放到工程里面,编译通过了,在执行的时候报错:
I1015 09:42:25.221655 11960 net.cpp:406] mobilenet0_conv21 <- mobilenet0_batchnorm20
I1015 09:42:25.221655 11960 net.cpp:380] mobilenet0_conv21 -> mobilenet0_conv21
F1015 09:42:25.300750 11960 cudnn_conv_layer.cpp:53] Check failed: status == CUDNN_STATUS_SUCCESS (1 vs. 0) CUDNN_STATUS_NOT_INITIALIZED
*** Check failure stack trace: ***
请问这是什么原因?谢谢!

你在配置caffe时需要根据自己电脑的capacity设置 CUDA_ARCH

我的环境是win10+cunda10.1+rtx2060。我首先配置的caffe-windows,成功。然后将你的这个工程添加到caffe工程下的examples目录下新建的项目,然后单独编译这个项目,通过。在运行的时候,执行
/* Load the network. */
net_.reset(new Net(model_file, TEST));
到这里的时候,就是解析模型文件,就报这个错误。
Rtx2060显存是6G的,应该也够用吧,调用的是mnet.prototxt。

应该不是显存不够用,比如我2080TI的CUDA_ARCH应该设置为7.5,你要找到对应的算力表
https://developer.nvidia.com/cuda-gpus ,添加一行加上这个算力

@xiakj
Copy link

xiakj commented Oct 15, 2019

It may have been a problem with your caffe on cuda

Project/MobileNet-YOLO-master2/include
这包含的是什么?

这个是安装caffe的路径。。。

我在Windows10上配置的caffe,将您的代码放到工程里面,编译通过了,在执行的时候报错:
I1015 09:42:25.221655 11960 net.cpp:406] mobilenet0_conv21 <- mobilenet0_batchnorm20
I1015 09:42:25.221655 11960 net.cpp:380] mobilenet0_conv21 -> mobilenet0_conv21
F1015 09:42:25.300750 11960 cudnn_conv_layer.cpp:53] Check failed: status == CUDNN_STATUS_SUCCESS (1 vs. 0) CUDNN_STATUS_NOT_INITIALIZED
*** Check failure stack trace: ***
请问这是什么原因?谢谢!

你在配置caffe时需要根据自己电脑的capacity设置 CUDA_ARCH

我的环境是win10+cunda10.1+rtx2060。我首先配置的caffe-windows,成功。然后将你的这个工程添加到caffe工程下的examples目录下新建的项目,然后单独编译这个项目,通过。在运行的时候,执行
/* Load the network. */
net_.reset(new Net(model_file, TEST));
到这里的时候,就是解析模型文件,就报这个错误。
Rtx2060显存是6G的,应该也够用吧,调用的是mnet.prototxt。

应该不是显存不够用,比如我2080TI的CUDA_ARCH应该设置为7.5,你要找到对应的算力表
developer.nvidia.com/cuda-gpus ,添加一行加上这个算力

我没有用那个预编译好的caffe,用的makefile.config进行编译配置的,这里面有架构设置:

For CUDA >= 9.0, comment the *_20 and *_21 lines for compatibility.

CUDA_ARCH := -gencode arch=compute_30,code=sm_30
-gencode arch=compute_35,code=sm_35
-gencode arch=compute_50,code=sm_50
-gencode arch=compute_52,code=sm_52
-gencode arch=compute_60,code=sm_60
-gencode arch=compute_61,code=sm_61
-gencode arch=compute_61,code=compute_61
我的cuda是10.1,所以将_20和_21删掉了,这里应该就是你所说的CUDA_ARCH.
实在没办法了,我现在在服务器上(有4块RTX2080)配置caffe,等编译成功了再试试。

@xiakj
Copy link

xiakj commented Oct 16, 2019

It may have been a problem with your caffe on cuda

Project/MobileNet-YOLO-master2/include
这包含的是什么?

这个是安装caffe的路径。。。

方便加个实时联系方式吗,关于部署的问题我想咨询您一下,谢谢!

@Hope-y0909
Copy link

你好,我也遇到了这个问题,请问您最后是怎么解决的

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

4 participants