Vampire Speedtests

Posted by ALB42 on 3. April 2017One Comment

Tried some (Integer) Speedtest, some people asked me about. Of course I’m also interested if my feeling (double 060/50 speed). I used some code from Benchmarks Game. Especially the tests: fannkuch 10 and BinaryTree 15. (68060/50Mhz= A1200 with Blizzard 1260, MorphOS = MacMini 1.4Ghz, UAE/AROS/Linux = Athlon FX 8120 3.16 Ghz)

machine BinaryTree fannkuch
Time [s] Speed rel. to 060 Time [s] Speed rel. to 060
68060/50Mhz 42.9 1.0 59.6 1.0

VampireA600 23.4 1.8 28.3 2.1

MorphOS 68k emul 3.0 14.3 33.0 1.3

WinUAE OS3.9 1.4 30.6 2.0 29.8

MorphOS PPC 1.3 33 2.3 25.9

AROS i386 0.6 71.5 0.7 85.1

Linux x86_64 0.3 143 0.2 298

So my feeling was more or less right, it’s around double speed to 060, so it’s a nice step but far far away from UAE. More interesting for me would be FPU processing but at this point the comparison would be very dramatic and a little bit unfair because Vampire still have no FPU support (even it’s everywhere announced, but seems even next version will not have FPU).

One comment to "Vampire Speedtests"

  1. magorium sagt:

    hi ALB,

    Thank you for the overview of your benchmarks.

    The results are very interesting.

    For instance, i was surprised to see that your MOS setup is about the same speed of WinUAE (incomparable HW of course, but interesting to see nonetheless).

    AROS native would of course be no match, but am surprised to see twice the increase in result for 64 bit hosted (and even more for the fannkuch test).

    Vampire being about twice the speed of a 1260, isn’t too bad. I take it that for this test, no use was made of optimized Vampire instructions ?

    Sorry to hear that FPU support keeps itself waiting, as it would make some comparisons a bit more fair and make things more usable.

    regards,
    M.

Schreibe einen Kommentar

Deine E-Mail-Adresse wird nicht veröffentlicht. Erforderliche Felder sind mit * markiert