08193: spacfury: Space Fury gameplay slows down in second level after first ship docking.

ID	Category [?]	Severity [?]	Reproducibility	Date Submitted	Last Update
08193	Speed	Major	Always	Jan 24, 2022, 18:12	Jan 28, 2022, 05:01

Tester	john_iv	View Status	Public	Platform	MAME (Official Binary)
Assigned To		Resolution	Open	OS	Windows 10/11 (64-bit)
Status [?]	Confirmed			Driver	sega/segag80v.cpp
Version	0.239	Fixed in Version		Build	64-bit
		Fixed in Git Commit		Github Pull Request #

Summary	08193: spacfury: Space Fury gameplay slows down in second level after first ship docking.
Description	Playing spacfury once you defeat the first wave of fighters you get to dock with an upgraded weapon system. Once you dock the next round appears to experience a slowdown on the entire screen, enemy and your own ship movement. Emulated speed is 100% on my new i7-12700K. Slowdown happens both during and without speech in-game.
Steps To Reproduce	1. mame.exe spacfury -norc 2. Play past first level, dock with new weapon. 3. Start second level. 4. Move around screen and rotate. Observe sort of a 'chugging' slowdown. INP attached. (most noticeable in level 2 when I start doing circles)
Additional Information
Github Commit

Flags
Regression Version
Affected Sets / Systems	spacfury

Attached Files	spacfuryslow.inp (27,223 bytes) Jan 24, 2022, 18:12 Uploaded by john_iv spacfury inp

No.19715 Haze Senior Tester Jan 24, 2022, 20:56	regression version would help with this one
No.19716 john_iv Senior Tester Jan 24, 2022, 20:59 edited on: Jan 24, 2022, 21:00	Yeah, I don't have too many older .exes to bisect. Maybe in a bit unless someone leaps to it. ;) Also interesting that there are no performance dips in the bench results of that game which I use in my suite.
No.19719 Tafoid Administrator Jan 25, 2022, 02:02 edited on: Jan 25, 2022, 02:03	I suspect this is completely a result of the netlist sound additions (replacing samples) in MAME 0.224 Not sure if there is anything we can do about it. If it were something unrelated to sound - it should repro in 0.223 or earlier?
No.19720 john_iv Senior Tester Jan 25, 2022, 04:17 edited on: Jan 25, 2022, 04:18	Confirmed. .224 has the slowdown, so not a regression. I'm getting 300% unthrottled on .239 during playback of the slowdown areas so it's not the host CPU dipping. .223 does not have the slowdown. Maybe we can get a visit from Aaron.
No.19721 Haze Senior Tester Jan 25, 2022, 12:02	it's an interesting case though, the speed isn't dropping below 100%, so it isn't due to the demands of the new sound emulation. something is messing up the timing elsewhere, maybe a status bit, or an unrelated change in the driver.
No.19725 Haze Senior Tester Jan 26, 2022, 18:56	It was traced back to the waitstate support that was added at the same time, so the emulated machine is expected to have more slowdowns now, but maybe it's too aggressive / incorrectly implemented.