RAM Init Pattern

[groepaz]

As the thing came up on Lemon [1].... it would be nice if some of you could dump a snapshot of the initial (non initialized) RAM, so we can find out what common/typical patterns are and tweak the emulation accordingly. We did this for the C64 a while ago [2] and it exposed a few interesting things.

* i am not sure how to do this on vic20, on the C64 you can do poke43,0:poke44,0:poke45,255:poke46,255:poke55,255:poke56,255:save"mem",8,1 ... perhaps it works the same? please tell if you know

* start vic20 without any expansions, then save said dump. please tell what ASSY your vic20 has, and what the type of RAM is. then post it all here

* if you can tell other programs than "AE" which misbehave depending on the RAM init pattern, please name them too

thanks!

[1] https://www.lemon64.com/forum/viewtopic.php?p=902588
[2] https://sourceforge.net/p/vice-emu/code ... itpattern/

[brain]

When you say non initialized ram, are you talking about RAM before the RAM test runs? Or, are you asking for the default state of RAM once BASIC has initialized and the cursor appears?

Jim

[groepaz]

when the cursor appears (the RAM test doesnt change the contents anyway). it will then likely contain SOME initialized pages (0-$3ff), but also enough pages that are untouched so we can compare them against other dumps

[brain]

OK. I wonder if there's a way with SRAM and some of my uC stuff to read the actual state of RAM without even needing to use the save routines.

Jim

[DarwinNE]

We did this for the C64 a while ago [2] and it exposed a few interesting things.

When I worked on my games for the unexpanded VIC, I ran into situations where the game ran flawlessly on VICE but not on the machine. IIRC, it was caused by memory locations that I forgot to initialize and were containing 0 on VICE but something different on the real thing. It was a coding error, but I could not notice it in the emulator because of this.

If it can help, I will be more than happy to report observations of what happens on my PAL VIC20 (CR) (just not this weekend).

BTW, I have an oscilloscope and a logic analyzer available, if there is the need for hardware testing (I don't think it's the case here, but just in case...)

[groepaz]

Whats needed are as much RAM dumps from different machines/boards/ram types as possible... plus a list of programs that fail in one way or another because they rely on it. (please only stuff from "back in the days" - the reason being that we will select a pattern that makes as much of them work correctly as possible)

[eslapion]

RAM init patterns vary greatly between static and dynamic RAM.

One VIC-20 game which is sensitive to RAM init patterns is Jelly Monsters. The original version will work ok on the old 2 prong VIC but it will have problems with later VIC-20s which have the C64 style power plug because they also use 2kx8 SRAMs instead of 11 2114.

IMHO, the simple way to have 'normal' RAM init patterns you can dump is to set a VIC-1110 for BLK5 so BASIC startup doesn't affect it.

One could also have a single 8k x8 SRAM IC visible in BLK 5 too.

[groepaz]

Anyone?

[ops]

Here is mine.

Edit 2020-03-09: Assy 250403, Rev D

[groepaz]

Thanks!

Anyone else? we need a few more to draw any meaningful conclusions

[srowe]

Here's mine from my original (two pin). Note, it has an internal 3K expansion fitted to the VIC address bus.

[groepaz]

please tell what ASSY it is.... all i know about "original two pin" is that it exists

[srowe]

Didn't realise this would need a total tear-down

324002-02 Rev D

[vicist]

MEM-324003 = pet keyboard WGB 1606
MEM-250403 = cr version UKB 126095

3583 bytes free on both machines.

MEM.zip

(57.75 KiB) Downloaded 56 times

[DarwinNE]

Sorry for the delay: here is mine. PAL VIC_20 CR version Ser. No. WGC 182127

Here is a picture of the memory chips:

Hope it helps.