nesdev.org

Posted: **Wed Feb 16, 2011 12:43 pm**

I'm trying to figure a way to test whether emulators *truly* support WRAM bigger than 8K.

The test I came up with was to write a different value to the last byte of each 8K WRAM bank and then when all 4 banks have been written, read the values back from each bank and check that they are still set correctly.

The logic behind this is that if the WRAM bank-switching is not implemented, the value from each bank will read the same value (the last value that was written) and therefore I can trap that condition.

However, I've tested this in FCEUX, NesterJ and VirtuaNES and the odd thing is, despite the fact that I *know* these emulators don't support WRAM>8K, the test doesn't catch.

Can anyone suggest another test?

Here's the code from my test:

Code: Select all

	lda #WRAM_BANK_00
	jsr setMMC1r1
	lda #$01
	sta WRAM0_TEST	
	lda #WRAM_BANK_01
	jsr setMMC1r1
	lda #$02
	sta WRAM1_TEST	
	lda #WRAM_BANK_02
	jsr setMMC1r1
	lda #$04
	sta WRAM2_TEST	
	lda #WRAM_BANK_03
	jsr setMMC1r1
	lda #$08
	sta WRAM3_TEST
		
	lda #WRAM_BANK_00
	jsr setMMC1r1
	lda WRAM0_TEST
	cmp #$01
	bne @lockUp	
	lda #WRAM_BANK_01
	jsr setMMC1r1
	lda WRAM1_TEST
	cmp #$02
	bne @lockUp
	lda #WRAM_BANK_02
	jsr setMMC1r1
	lda WRAM2_TEST
	cmp #$04
	bne @lockUp
	lda #WRAM_BANK_03
	jsr setMMC1r1
	lda WRAM3_TEST
	cmp #$08
	beq @noLock

@lockUp:	jmp @lockUp
	
@noLock:

'setMMC1r1' sets the WRAM bank
WRAM0_TEST is a RAM variable in bank 0
WRAM1_TEST is in bank 1 etc

Posted: **Wed Feb 16, 2011 1:28 pm**

Mmh maybe it's possible that those emulators *does* emulate 32k WRAM, but does only *save* 8k ? If they would emulate 8k or 16k this code would obviously detect that.

Also this might seems like a stupid question, but does WRAM0_TEST WRAM1_TEST, WRAM2_TEST and WRAM3_TEST equals the same adress (exept the bank) ? If they're not, it's no wonder why your program doesn't lead to the expected results.

Posted: **Wed Feb 16, 2011 1:45 pm**

Bregalad wrote:Mmh maybe it's possible that those emulators *does* emulate 32k WRAM, but does only *save* 8k ? If they would emulate 8k or 16k this code would obviously detect that.

Also this might seems like a stupid question, but does WRAM0_TEST WRAM1_TEST, WRAM2_TEST and WRAM3_TEST equals the same adress (exept the bank) ? If they're not, it's no wonder why your program doesn't lead to the expected results.

Doh, they were *supposed* to be pointing to the same address. Would help if I'd setup the RAM segment properly in my CA65 config file.

Thanks for being the voice of clarity Bregalad.

Everyone else feel free to ignore this thread. *shame*

Posted: **Wed Feb 16, 2011 6:51 pm**

neilbaldwin wrote: Everyone else feel free to ignore this thread. *shame*

Any chance you could put up the generated ROM for those of us with emulators not in your "tested on" list to test with?

Yes...I'm being lazy.

Posted: **Thu Feb 17, 2011 12:57 am**

NESICIDE wrote:
neilbaldwin wrote: Everyone else feel free to ignore this thread. *shame*
Any chance you could put up the generated ROM for those of us with emulators not in your "tested on" list to test with?

Yes...I'm being lazy.

Already done

http://blog.ntrq.net/?p=494

Posted: **Thu Feb 17, 2011 1:01 am**

now that's what i call a "nag screen"

nesdev.org

Testing For 32KB WRAM Support in an Emulator

Testing For 32KB WRAM Support in an Emulator