[ITK] gpu mean filter segfault
Wei Liu
weiliu620 at gmail.com
Tue Sep 13 14:54:25 EDT 2016
Hi community,
It's more like a question to developers. I'm using itkGPUMeanImageFilter
and got a segfault. Here are the steps I took:
- Compile latest github version of ITK with GPU option on
- Wrote a program to apply gpu mean filter on a 3D MRI brain image.
Run it in gdb (to show the location of segfault) and here is the output:
---------- gdb output ------------------------
(gdb) run -i ../data/MRHead.nrrd -o tmp.nrrd
Starting program: /home/weiliu/build/gpu_mean_smoothing -i
../data/MRHead.nrrd -o tmp.nrrd
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
Platform : NVIDIA CUDA
[New Thread 0x7ffff465e700 (LWP 12782)]
[New Thread 0x7fffebbff700 (LWP 12783)]
[New Thread 0x7fffeb2b6700 (LWP 12784)]
[New Thread 0x7fffeaab5700 (LWP 12785)]
[New Thread 0x7fffea2b4700 (LWP 12786)]
[New Thread 0x7fffe9ab3700 (LWP 12787)]
[New Thread 0x7fffe92b2700 (LWP 12788)]
GeForce GTX 980 Ti
Maximum Work Item Sizes : { 1024, 1024, 64 }
Maximum Work Group Size : 1024
Alignment in bits of the base address : 4096
Smallest alignment in bytes for any data type : 128
cl_khr_byte_addressable_store cl_khr_icd cl_khr_gl_sharing
cl_nv_compiler_options cl_nv_device_attribute_query cl_nv_pragma_unroll
cl_nv_copy_opts cl_khr_global_int32_base_atomics
cl_khr_global_int32_extended_atomics cl_khr_local_int32_base_atomics
cl_khr_local_int32_extended_atomics cl_khr_fp64
Defines: #define DIM_3
#define PIXELTYPE float
Program received signal SIGSEGV, Segmentation fault.
0x0000000000670f96 in itk::GPUMeanImageFilter<itk::Image<float, 3u>,
itk::Image<float, 3u> >::GPUGenerateData (this=0x1b01f00)
at
/home/weiliu/packages/ITK/Modules/Filtering/GPUSmoothing/include/itkGPUMeanImageFilter.hxx:75
75 typename GPUOutputImage::SizeType outSize =
otPtr->GetLargestPossibleRegion().GetSize();
--------- end of gdb output --------------------
So where should I start to debug? I'm on Ubuntu 14.04 LTS, 64 bit, with
CUDA 7.5.
Thanks for your input.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/community/attachments/20160913/e1bb300b/attachment.html>
More information about the Community
mailing list