Re: [hatari-devel] Hatari WinUAE speed comparisons with Falcon030 / 16MHz

[ Thread Index | Date Index | More Archives ]

A trivial question, but are you disabling interrupts for the duration of your test?

Yes, all but TimerC, which is configured at 200Hz for timing purposes.

Each test is also executed a second time, with the critical part of the timed fragment absent, with that time measurement subtracted from the original test. This compensates for the test harness time, and yields the time for the critical fragment only. This is why the cycle measurements for the trivial cases appear nearly exact.

There are a few other tricks used to deal with cases where opcodes overlap, but the more interesting cases don't occur here anyway.


Mail converted by MHonArc 2.6.19+