On Sat, Jul 05, 2014 at 02:40:55AM +0100, Al Viro wrote: > d) at least on EV6 and EV67 DNOD *still* trips INV. According to the > manual suppression of INV by DNOD is optional. And while their text > might be interpreted as "INV is suppressed if operation with denorm > wouldn't result in something unpleasant" (which would apply to > sqrt(DBL_MIN/2)), the same behaviour happens on DBL_MIN/2 + DBL_MIN/2, > where the result is a good finite value, so it really looks like DNOD > doesn't suppress INV at all on these processors. > > Does anybody have 21364 to run some tests on?
In fact, DNOD is simply not implemented on those guys - if you try to set it, the bit still reads zero. Worse, according to Compiler Writer's Guide for the 21264/21364, "Alpha architecture FPCR bit 47 (DNOD) is not implemented by the 21264 or 21364". In other words, it looks like FPCR.DNOD is something from (never-produced) 21464.
