On 05/22/2015 04:39 PM, Lennart Sorensen wrote:
That seems to require registration.:)
My bad! It does indeed. I am not sure if I can send the file directly though.
This presentation seemed to indicate that at least there are some biases towards big endian in the vector instructions: http://llvm.org/devmtg/2014-10/Slides/Schmidt-SupportingVectorProgramming.pdf
That is a very interesting presentation, thanks for sharing. It is also curious because we got some help from Bill himself since we work at the same department here at IBM. :) Indeed I stumbled onto some of those biases when trying to build pixman on LE (vec_perm, vec_mergeh etc for instance). -- Fernando Seiti Furusato IBM Linux Technology Center _______________________________________________ Pixman mailing list Pixman@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/pixman
