caffe make runtest error(core dumped)Check failed: status == CUDNN_STATUS_SUCCESS (6 vs. 0)

参考了谷歌论坛

=================================================================

简单讲就是GPU的加速性能不够,CUDNN只支持CUDA Capability 3.0以上的GPU加速

==================================================================

在实验室台式机上(I7+GeForce GTX TITAN Black With CUDA CAPABILITY 3.5)成功配置caffe并测试数据无误、训练数据也开始以后,也打算笔记本上(I5+GEFORCE GT540M--With CUDA CAPABILITY 2.1)配上caffe.

于是安装了cuda和cudnn加速,make和make test都过了,而make runtest时报错,大概是这样滴错误

[plain] view plaincopy
  1. <pre name="code" class="plain">[----------] 6 tests from CuDNNConvolutionLayerTest/1, where TypeParam = double  
  2. [ RUN      ] CuDNNConvolutionLayerTest/1.TestSimpleConvolutionGroupCuDNN  
  3. F1014 08:55:30.083176 23568 cudnn_conv_layer.cpp:30] Check failed: status == CUDNN_STATUS_SUCCESS (6 vs. 0)  CUDNN_STATUS_ARCH_MISMATCH  
  4. *** Check failure stack trace: ***  
  5.     @     0x2b082d0a8daa  (unknown)  
  6.     @     0x2b082d0a8ce4  (unknown)  
  7.     @     0x2b082d0a86e6  (unknown)  
  8.     @     0x2b082d0ab687  (unknown)  
  9.     @           0x739689  caffe::CuDNNConvolutionLayer<>::LayerSetUp()  
  10.     @           0x42c1f0  caffe::Layer<>::SetUp()  
  11.     @           0x49d776  caffe::CuDNNConvolutionLayerTest_TestSimpleConvolutionGroupCuDNN_Test<>::TestBody()  
  12.     @           0x68d613  testing::internal::HandleExceptionsInMethodIfSupported<>()  
  13.     @           0x6840b7  testing::Test::Run()  
  14.     @           0x68415e  testing::TestInfo::Run()  
  15.     @           0x684265  testing::TestCase::Run()  
  16.     @           0x6875a8  testing::internal::UnitTestImpl::RunAllTests()  
  17.     @           0x687837  testing::UnitTest::Run()  
  18.     @           0x41e9d0  main  
  19.     @     0x2b0831cf1ec5  (unknown)  
  20.     @           0x4261c7  (unknown)  
  21.     @              (nil)  (unknown)  
  22. make: *** [runtest] Aborted (core dumped) caffe::NetTest_TestReshape_Test<>::TestBody()  
  23.     @           0x68d613  testing::internal::HandleExceptionsInMethodIfSupported<>()  
  24.     @           0x6840b7  testing::Test::Run()  
  25.     @           0x68415e  testing::TestInfo::Run()  
  26.     @           0x684265  testing::TestCase::Run()  
  27.     @           0x6875a8  testing::internal::UnitTestImpl::RunAllTests()  
  28.     @           0x687837  testing::UnitTest::Run()  
  29.     @           0x41e9d0  main  
  30.     @     0x2b2a9f778ec5  (unknown)  
  31.     @           0x4261c7  (unknown)  
  32.     @              (nil)  (unknown)  
  33. make: *** [runtest] Aborted (core dumped)  



在内网找了半天无果,终于在墙外找到了解决办法,不,是问题所在。因为笔记本上的GT540M的CUDA Capability是2.1,而官方的cudnn加速是不支持3.0以下的版本的,因此只能在Makefile.config中注释掉USE_CUDNN这行,重新执行以下

[plain] view plaincopy
  1. make clean  
  2. make all -j4  
  3. make test -j4  
  4. make runtest -j4  


最后除了make runtest中2 DISABLED TESTS之外,没有其他问题。make runtest中出现几个测试例子不过不影响使用。
展开阅读全文

没有更多推荐了,返回首页