[kaffe] Re: Darwin/x86 134 of 144 tests Failed

Kiyo Inaba inaba at src.ricoh.co.jp
Wed Sep 8 20:20:49 PDT 2004

Hi Dalibor,

In general, your comment covers almost all points I have to mention.

I have one additional comment.

Dalibor wrote:
>A very crude (and factually wrong, anyway, but as we are guessing upper 
>bounds here, the advantages of RISC systems do not do harm) estimation 
>would be to look at the lowest CPU megahertz counts people are testing 
>on: if it takes 5 seconds with jit on a 200 MHz ix86 then it may take 
>5*10*10 secs on a 20 MHz m68k with intrp.

This calc matches to what we get for the last several years, but if
I can turn soft-float back again, it may introduce one more factor.
I have no idea right now, how intensively floating is used in regresion,
but I am sure it may make the execution slower :-)

By the way, for clarity, even right now, someone can test soft-float
if you have proper CPU with floating-point exception. In this case,
the porter needs not do anything to enjoy slowness of that CPU. But
as far as M68k series is concerned, the famous EC problem makes us
not be able to use floating-point emulation, and for 683xx series
may not have FP at all. So the only way is to make 'HAVE_NO_FLOATING_POINT'
back again for jit3 (and hopefully jit).

P.S. If it works, I can finally start 683xx ucLinux port for kaffe ;->
P.P.S. This port can be 'smallest hardware in the world', I think.

More information about the kaffe mailing list