8bitwizard wrote:Maybe that wasn't a picture of Pete Sampras after all? (it wasn't my photo, so I never saw the actual cartridge)
It was indeed Micromachines 2 on the picture
Something strange about this game is that it seems to need the external RAM being initialized with the string "PETETEST01234567" repeately written all over the memory, otherwise (by default, external RAM is initialized to 0x00) it will rewrite the memory with theses strings, then read the whole memory again and finally overwrite every byte with 0x00 if we return the previous written values (this does not happen if we return 0x00 to each byte read).The result will be that characters names are all messed up in the game menus.
I modified my sourcecode to initialize external RAM with these strings and as I expected, the game don't reset the memory to 0x00 anymore and now put correct values into external RAM (I can see now ASCII values, used by the game)
I also found why I can't get those Micro Machines games to backup/load data correctly. DIsassembling the code, it appeared that the bit position for SDA_OUT was not the same as the one used for SDA_IN so I was always returning 0.
Regarding other games, the only ones than still give me problem are the ones from NBA Jam serie. It seems that they also use a different bit position for SDA_OUT and SDA_IN but I didn't managed to get it through disassembly
hum ...and for romantic people, here's what can be found written in external RAM by NBA Jam, which I know was originally develloped by Steve Snake
00000008:53 6e 61 6b 65 20 31 39 39 33 Snake 1993
000000e5:24 6c 6f 76 65 73 20 47 69 6c 6c $loves Gill
the "funny" thing is that I found this (very) old message in Eidolon's archive while digging for some genesis technical tips: http://eidolon.dnsalias.net/ta/199912/m ... /1077.html
eight years ago, Steve Snakes wrote:
Well, coding all day can get a little dull....
I'm just glad that nobody has emulated the back-up chip used in NBA Jam / Jam TE......
EDIT: it seems that Rev00 of NBA Jam TE has bugged eeprom support which prevents backup data to be saved correctly. With Rev 01 version, this is finally working