So I tested things on the HW available, and its good that all variations show same results. 50Hz : A34A / A34B (rapidly changing) 60Hz : 88AD / 88AE (88AE appears only momentairly every once in a while like Stealth described) AND some 10MHz overclock results : 50Hz : D179 / D17A (rapidly changing) 60Hz : AF56 / AF55 (rapidly changing) As you can see, 50Hz at stock speed is close 60Hz in 10MHz in a typical VBL driven game setup :P EDIT : and ROM (actually whole address range) access is fixed speed, which is managed by the I/O controller.... ROM speed does not matter, if its too slow, the game crashes, if too fast, things work, but not faster, exactly as fast as with a slow ROM. !DTACK is generated in the I/O controller or 32X not in the cartridge by ROM.
On both my CDX and Model 2 Genesis (US), I get 88AD with the occasional flick to 88AE every few seconds, just like the others. I think it's safe to say that all NTSC models are going to give the same result.
Running from ROM NTSC Running from RAM NTSC (by adding a small penalty when RAM is accessed) Running from ROM PAL Running from RAM PAL (untested, Tiido!!) Done by emulating emulating 481 cycles per lines instead of 488.
50Hz - A1BA / A1BB 60Hz - 875E / 875D 10MHz 50Hz - D07C / D07A / D07B / D07D 60Hz - AE81 / AE80 / AE82 / AE83 I find it funny that RAM has difference... it seems some part of the address area has different timings....
Ahha....so the last screenshot, which was totally untested at the time posted, matches exactly with my results.
well, this thing only touches few aspects about the HW, and my no means its definitive.... if you want something evil, I'll dig out my other, serious benchmark, that will show WILDLY different results accross emulators Things are always constant too, and no difference between 50 and 60Hz since VDP is not relied on Things take about 20 seconds to perform......
/me laughs at Sik. TmEE, can you PLEASE at least tell me the results of just Regen compared with real thing? (on MSN/IRC/whatever).
I have not done any test on Regen, since Regen did not exist when I made that benchmark, and I have remembered about it only recently... I need to clean things up a little and do a proper 3rd release
You didn't make any benchmark so you can't laugh. Not to mention that your emulator gave different results than real hardware, regardless if this benchmark is flawed or not =P
I'd rather code a "serious benchmark" or none at all. And didn't you see the screenshots? /me laughs at Sik a bit more
Talks the one who kept insisting me in making the benchmark ROM, so you still have some fault. Besides Regen was doing it wrong when I had posted the ROM anyways. Not to mention that it's still a bit off in two of the tests =P
Sorry for resurrecting an old thread, but someone expressed interest in the 68k Bench I had on my Sega CD shelf. The following zip file has ISOs for US, EU, and JP. All source is included in the ISO file structure. I only had to modify one line of Sik's source to make it relocatable to execute from RAM instead of a ROM (and wrap for use on the Sega CD). Nothing amazing, but I guess no reason to let it rot on my shelf and hard drive. Link: 68KBENCHISO.ZIP (100K)
