03461: starfire, starfirea, starfir2: Scratchpad Ram test not available in emulation

ID	Category [?]	Severity [?]	Reproducibility	Date Submitted	Last Update
03461	DIP/Input	Minor	Always	Oct 4, 2009, 20:22	Dec 9, 2012, 13:26

Tester	TrevEB	View Status	Public	Platform	MAME (Official Binary)
Assigned To	hap	Resolution	Fixed	OS
Status [?]	Resolved			Driver	exidy/starfire.cpp
Version	0.134	Fixed in Version	0.147u4	Build	C2D
		Fixed in Git Commit		Github Pull Request #

Summary	03461: starfire, starfirea, starfir2: Scratchpad Ram test not available in emulation
Description	Scratchpad ram test is noted on page 7 & 17 of the owners manual as a complete test of the main ram done by disabling the Non-maskable Interrupt signal to the cpu. Jumper J6 at location 4G on the mainboard must be open. This would be an added option to complete emulation of the system tests. Also of note on Page 6 is the shorting of resistor R3 at location 2A of the mainboard if the daughterboard has 11 chips (starfire) versus 10 chips (starfirea) Does the driver account for this in the test? I have no idea. It behaves like it does.
Steps To Reproduce
Additional Information
Github Commit

Flags	Verified with Original
Regression Version
Affected Sets / Systems	starfire, starfirea, starfir2

Attached Files

No.05423 phantomdj Tester Jan 6, 2010, 07:04	Careful analysis of the game's code reveals where the self tests occur and the check for the NMI's to determine if a scratchpad RAM test should occur. You can fool the game's code into allowing the scratchpad RAM test with the following breakpoint set in the debugger console of MAME: bp 51fd,1,{a=b@8367 ; g} From what I can tell, the scratchpad RAM test is set to run 256 times, but the loop does not contain a watchdog kicker (unlike Starfire2). So the game will reset before this test completes all of these loops. I don't know if disabling the NMI's also affects the need to kick the watchdog. It seems like it does, or else this test would also reset on an actual machine, which I doubt. So it appears that we need something in the MAME driver, like a dipswitch, which can emulate the disabling of the NMI's which is done on the actual board by shorting a resistor and/or changing a jumper.

No.05423

phantomdj

Tester

Jan 6, 2010, 07:04

Careful analysis of the game's code reveals where the self tests occur and the check for the NMI's to determine if a scratchpad RAM test should occur.

You can fool the game's code into allowing the scratchpad RAM test with the following breakpoint set in the debugger console of MAME:

bp 51fd,1,{a=b@8367 ; g}

From what I can tell, the scratchpad RAM test is set to run 256 times, but the loop does not contain a watchdog kicker (unlike Starfire2). So the game will reset before this test completes all of these loops.

I don't know if disabling the NMI's also affects the need to kick the watchdog. It seems like it does, or else this test would also reset on an actual machine, which I doubt.

So it appears that we need something in the MAME driver, like a dipswitch, which can emulate the disabling of the NMI's which is done on the actual board by shorting a resistor and/or changing a jumper.