View Single Post
Old 01 November 2013, 21:49   #298
Toni Wilen
WinUAE developer
Join Date: Aug 2001
Location: Hämeenlinna/Finland
Age: 42
Posts: 19,518
Originally Posted by amilo3438 View Post
Well, do you remember this ( already solved problem ... at first it was on same (false) diagnosis!
Everything is possible.

EDIT: Now one thing bugging me ... Can it happen a CPU instruction with bad timing, once repaired (if ever), causes errors in some of the "already repaired" demos?!
Only if there is some other bug and both bugs cancel each other.

Just thinking ...
What would be the fastest way to check does the all CPU instructions use correct timings?
Is it possible to print out the values and compare it with values with those in a book?
Or maybe make some program that will save timings from a real A500 and then compare these values on WinUAE?
There is no documentation that shows cycle-exact timings (except 68000 microcode in US patent document which unfortunately is from some pre-production variant and may have differences) and instruction internal timing can't be timed with software.

Each instruction variant, including address modes should be checked because 68000 is microcoded CPU, unexpected differences are possible.

It will take too much time and it is easy to make mistakes when analyzing results. Testing existing programs is much easier.

Madness 2 / Anarchy still seems to be the best test case because error is not small (= it has to be instruction that is executed many times).

EDIT2: Just tested the demo Last A500 (Demo) by Oblivion again with 68010, 68020 and 68030 CPUs on same A500 freq. with CE=ON ... got exactly same glitches on rotation cube part as with 68000! ... Is it possible that all emulated CPUs suffer from the same timing problem?!

EDIT3: But one interesting case: On the demo Hi Freaks (Intro) by Supplex & The Silents ... when using 68010 instead 68000, problem is gone!!!
It still isn't interesting and it won't help. It won't really narrow down possible problem, unless you start assuming and I won't (at least not too much).

btw, shifted scrollers should be 100% fixed now in today's

Last edited by Toni Wilen; 01 November 2013 at 21:59.
Toni Wilen is online now  
Page generated in 0.05358 seconds with 9 queries