Am 18.10.2014 um 08:22 schrieb Lukas Fleischer:
After a more thorough analysis by Jan, we found out that the issue is
caused by a bug in the new OpenCL benchmark code. On modern CPUs, the
benchmark is executed on every start of ImageMagick which leads to huge
number of __memcpy_avx_unaligned() calls with a small block size. Jan
prepared a patch and he will submit it upstream.
If you wish to use the current imagemagick version without a recompile
you can set the opencl cache to a missing directory. Then imagemagick
cannot write the cachefile and skips the benchmark.
env MAGICK_OPENCL_CACHE_DIR=/tmp/CACHE_DOESNT_EXIST convert ...
