vitorvilela

What if Gradius III used FastROM instead of SA-1?

Added 2020-05-31 23:59:48 +0000 UTC

You can give it a try now! While there is no noteworthy progress on my other SA-1 works, you can check it out one of my initial experiments with Gradius III. The BPS patch is attached on this post.

Initially I was having a lot of trouble for getting SA-1 Root working on Gradius III and in hope to have at least a little idea on how the game would run on the chip, I coded a FastROM patch for it that increased the SNES CPU clock speed from 2.68 MHz to 3.58 MHz when executing code and loading data from the ROM.

The CPU itself can run at either 1.78 MHz, 2.68 MHz or 3.58 MHz depending on which feature/device the system is accessing. Accessing the WRAM (main memory) is always 2.68 MHz, while internal registers and PPU (picture chip) always 3.58 MHz. Reading the controllers (joypad, multitap, mouse, etc.) is always at 1.79 MHz. But depending on the cart, the SNES CPU can read the ROM at either 2.68 MHz or 3.58 MHz. Since most of the code and all data is read from the ROM, that could give a small performance boost to the CPU. But is it enough to eliminate slowdown?

Given that 3.58 MHz is around 33% faster than 2.68 MHz, you can expect some minor differences on places that has almost no slowdown (the speed throttle is reduced) and on places that it has extreme slowdown (like during the boss rush battle). Keep in mind that the game will not run 33% faster because there are CPU commands that already runs at 3.58 MHz like arithmetic commands and others will stay at 2.68 MHz like RAM read/write operations, don't forgetting that the game awaits for the video synchronization period for rendering the next frame.

What are your thoughts about the patch? Have you noticed any significant speed improvement? Comment it now! Given the feedback I can consider doing the same for my next works, since it's technically an intermediate step between the original game and a full SA-1 implementation.