🚀 go-pugleaf

Paul Hsieh <qed@pobox.com> wrote:
>of.  Nevertheless, if you have a good one it should document the CPUID 
>instruction which you can test function 1, bit 4 of edx.  That's how you 
>know whether or not your CPU supports the TSC register.
>
>RDTSC operates in all rings; It does not run in 16 bit v86 mode.
>
>You don't have to guess.  Use the CPUID instruction above -- it tells you 
>all the additional features that your processor supports.

I'm suprised you didn't tell me how to check for cpuid support as well,
or mention any of the ways to figure out whether or not a program is
running on a processor that can acutally execute a cpuid check.  I'll
admit to being lazy and usually assuming my code is running on something
more modern than a 386, though that's not a wise assumption in some
markets.

Do you really think I'm so stupid I don't know all this already?  What I
do lack is the resources to go out and do characterization that should
be done by x86 clone manufacturers anyway; if I download a K6
instruction set listing---from AMD's website, of all places---and
discover rdtsc is *not* listed, it's only sensible to assume no K6 has
rdtsc and code accordingly.

>> FWIW, I've also had invalid opcode faults on rdtsc on an early K6,
>Impossible.  Unless you are in 16 bit v86 mode (ask Intel about that.)

No, it was not 16 bit v86 mode: wake up, Paul.  Nothing's perfect, and
it's not like I had a K6 ICE and the time to see what was going on,
particularly when AMD's documentation gave no indication the processor
did support rdtsc (that was a little over a year ago; perhaps they've
fixed this by now, but I've moved on to other things).

Todd
--
Anti-spam boggling in effect.  Edit reply-to address before sending mail.