 No.06278
Tafoid Administrator
Jun 22, 2010, 00:59
edited on: Jun 22, 2010, 01:32
|
Added more driver based on reports from Bryan Ischo: http://www.mameworld.info/ubbthreads/showthreaded.php?Cat=&Number=224832 All single performance related bugs are consolidated to this one bug. |
|---|---|
 No.06279
Haze Senior Tester
Jun 22, 2010, 05:38
edited on: Jun 22, 2010, 05:43
|
As Luca states lordgun.c is because the entire video rendering was rewritten between versions to be more accurate. SH2 (especially multiple SH2) games have taken a noticable hit, hence coolridr, and stv.c., psikyo4 is single sh2, but as noted, performance now seems to dip significantly in places (likewise puzzloop in suprnova.c) So there is definitely some issue with the SH2 under the new system. attckufo isn't just slower, it has various on-screen glitches now knckhead which he lists as being significanlty *faster* now is actually running too slow, and has many sprite glitches. (likewise Atomiswave which he lists as faster now doesn't boot at all) You didn't copy these here as regressions, but you probably should. The only reason they now run faster is because they're broken. dgpix and limenko are both Hyperstone games, maybe they're having the same issue with the old speedups as GStream was (where the code to handle the idle skip had become slow due to cpu name string lookups) nss/sfcbox/snesb.c are all SNES, so will all be the same issue. |
|
No.06281
Tafoid Administrator
Jun 22, 2010, 13:03
|
I checked speed slowdowns by the numbers.. that's it. I took his list and tested per game for substantial speed drops and posted what I saw to be valid issues. I didn't look at the 'faster' games he has listed. I'll look it over today and check for issues. If there are bugs that you know have shown up that are not speed related and you have no clue on how to fix, please make a report with some screenies to show the obvious regressions. I can't do it all alone. It would seem that most of the major slowdowns are with either speedup hacks not being used .. or in the case of SNES family stuff, the cycle-eating routines are likely stuffed. There are also a number of games/drivers that had some lower loss (up to 10%) which still might be of concern but I have them documented locally in case it becomes an issue. |
 No.06778
aaron Developer
Oct 19, 2010, 13:47
|
I'm calling this resolved. The source of the major slowdowns is fixed. Any outliers are just slower. |