Hi all, I was working on some unit tests for the expm1 builtin cl function tonight, and I noticed that in piglit-util-cl.c the code to probe float/double results and check that the result is within a given ulp sees to be treating the ulp as an actual tolerance instead of as a proper ulp value.
I'll try to see what I can do to fix it, but I wanted to let people know about it, as fixing the bug may cause a bunch of math rests to start failing if we have implementations that do not meet the precision requirements. --Aaron
_______________________________________________ Piglit mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/piglit
