Okay, I'm probably going to lose all my karma, but here goes...
I've got a 7500 that's been upgraded to a G3/338 with 224M of RAM. The video card is an ATI XClaim VR. The system is 8.6. So far, so good, it meets the requirements posted. However, I've been having a problem with the game being sluggish and jittery. There wasn't a quick-and-ready frames-per-second display on the demonstration game, but it seemed to be going from 14 FPS when no scrolling was involved, down to what feels like 8 FPS when scrolling or worse. At times, during game play, there would be a distinct half-second pause during walking, fighting, watching butterflies, etc. The effect wouldn't be as bad, except for the speed throttle doesn't seem to be coping, as moving will cause the butterfly to crawl, and as soon as the scrolling stops, the butterfly shoots off.
The random stutters, of course, are mostly other apps. Quitting background tasks helped things, and playing around with spriteworld and WaitNextEvent proved the hit (Multiple Layers Demo gave 60FPS w/o WaitNextEvent, 45 with). I RTFM and I had to say adios to any map alpha mask in order to improve the feel, so at least it's faster than PoG was for me. But it still feels too sluggish.
I tested PoG on a P2/350/64/win98 (Yes, I know I'm pushing things here) and it still seemed to have speed issues. I couldn't fully test what I made with Coldstone, as it caused the computer to reboot when trying full screen, and gave an unhandled exception when running in a window.
Cutting to the chase:
-
I know my system is what could euphemistically be called 'legacy'. Is it simply too old to run Coldstone's games?
-
If so, then what was the minimum system configuration the testers ran on, and could you update FAQ question 5 to reflect this?
-
If not, is there a configuration or setting I am doing wrong?
-
Otherwise, might I suggest that the engine allow for 8-bit graphics (to aid the blitter) and/or ignoring background apps (using Get instead of Wait)
-
Finally, I'm still mourning the loss of a non-rectangular border, but testing I did seems to indicate that with the mask present, the mask testing is done everywhere. In most cases, the region that needs masking is only a dozen or two pixels high, the rest either all background or all foreground. Could you allow for the artist to define region(s) inside the map area that need mask testing, so the blitter doesn't have to unnecessarily test large blank areas?
I know my system is the exception rather than the rule, and otherwise you've made a rather great product. Keep up the good work.
