Playstation and Saturn (32-bit) memory tech

My assumption is that the buffer was a thoughtful addition to the memory pipeline to hedge against long loading times and excessive CD-ROM reads. As the game is loaded into main RAM, VRAM, and Audio RAM and the player is playing, the CD-Rom can "casually" seek the next level's data and pre-load 512KB of it into the buffer to give it a head start when needed.

Also figured it could be a reason why the some of the Saturn 2D fighters loaded faster between bouts than the Playstation versions. Ancillary things like the ending portraits or level transition screens (i.e. the world map and fighter quote screens) could sit in buffer RAM rather than waste precious main RAM for the actual gameplay. This would have been a luxury the Playstation didn't have.
 
I played Deep Fear only for two hours, but plan play all game. Very interesting, graphics good, but could’ve been better. Interesting setting and story. Very good was what game have a lot of cinematic videos. Maybe because of that game is on two CDs.
And yes, I said that before and can say it again, in my opinion Saturn was amazing system. If that would’ve been used at maximum, graphics would’ve been a lot better. There are some home brew projects showing that, especially Unreal port.
 
Came back to this topic after some time. I've been specifically looking at the 512KB CD-Rom buffer in the Saturn compared to the Playstation's 32KB. This is a significant difference, but I rarely hear much discussion about it.

Did we really see a tangible difference in games due to the buffer space? Was Saturn much faster in loading games? It seems fairly hit or miss from some of the examples that I've seen, with the exception of 2D fighter games. Are some of the faster loading times seen in some of the 2D fighters more of a result of the extra buffer space, or the extra VRAM, or both?

I especially think about this because texture/data streaming became an important technique later in the Playstation's library. I believe that Crash Bandicoot was one of the early pioneers of this technique (streaming 32KB chunks of data from the CD at a time, according to lead programmer Andy Gavin). I wonder if 32KB was the limit precisely because of the buffer available in the system? Would the Saturn have been able to really up the ante with its 16x increase in buffer size?

16x is probably too big an increase to expect; after all it's not just the size of the memory buffer that would've contributed to the performance, but the entire subsystem itself, the drivers, the instruction set for the processor(s), memory bus width, even how the game itself was coded to support data streaming would've had an impact.

So, not every game would get that increase, BUT maybe if more games optimized for it, perhaps more multiplats on Saturn could've had faster load times than PS1 equivalents in theory. But there are two very big implications that could have negated that right out of the gate. First is lack of audio compression on Saturn; second is no MPEG hardware decoding. At some point the CPU has to access and process that data for it to actually be of use, even if it's sitting right there in the 512 KB buffer, and on Saturn tasks like MPEG decoding had to be done via software.

So if the graphics data, for example, are streamed MPEG video frames, I don't think you're getting a 16x stream rate boost on Saturn over PS1 just because the buffer is 16x larger. Could there still be a boost of some type in favor of Saturn? Possibly.
 
Back
Top