Fabrizio Levati wrote:
About to optimise input curves and matrix again ............................................................................
............................................................................ ........................colprof: Warning - xfit_fit: Conjgrad failed to converge
Looks like this is a bug (typo) in the profiling code, it's just that different compilers create code that responded differently to it. The fix should solve the performance problem, and it may improve the quality of some profiles. (On my MacBook the fixed code takes about 1m13s for the IFRA26L point set). The fix will be in the next release, but I've also updates <http://www.argyllcms.com/V1.0.1_patches.txt> to reflect the changes for those who can compile from source. Graeme Gill.