Home | Forums | What's new | Resources | |
A Saturn Programming Challenge |
the Sega Saturn Pope - Oct 8, 2022 |
< Prev | 1 | 2 | 3 | Next> |
nando | Oct 22, 2022 | |||
Funny enough, Fighters Megamix ruined VF3 for me when I got the Japanese launch day Dreamcast. Of course I got VF3, and of course the graphics were pretty good. But I'd basically already played a better version of it on the Saturn (gameplay wise), with much more variety and different modes to try. I really didn't play VF3 very much because I'd already learned all of the moves and most of the characters. Also the DC controller wasn't good for fighting games, but that's another story... So in the end I'm not sad we didn't get VF3 on Saturn. FMM was a better game overall, and it's unlikely the graphics would have been much different anyway. |
adbevers | Oct 27, 2022 | |||||
I'm not so convinced that Fighter's Megamix was developed in conjunction with Virtua Fighter 3, and not the *outcome* of the Virtua Fighter 3 development itself--Unless there's direct evidence otherwise, the timing seems to line up, and it would make sense for Yu Suzuki and crew to do a quick feasibility test using the Viper engine to get the Desert Stage up and running while they were primarily focusing on "Virtua Fighter RPG". Similarly to the reasons why Scud Race was cancelled for the Dreamcast, I believe Sega saw the extent of what they could reasonably achieve with a port, decided that it wouldn't approach a respectable approximation with the time/resources they could allocate, and decided to offer Saturn owners an alternative that wouldn't undermine the quality standards of the property (like they did with Scud Race > Daytona USA 2001) That said, I often wish we could've seen something with Virtua Fighter 3 that pushed the Saturn to the extent something like Digital Dance Mix or Last Bronx were able--Ever consider doing a raw "polygon pushing" demo @XL2...? |
the Sega Saturn Pope | Oct 27, 2022 | |||
Adbevers, since you were active during that time, do you have any idea how to get in touch with Sam Pettus? His site/book is where I am getting a lot of information from. Like so many others, he's difficult to get in touch with, having vanished after publishing the Service Games book and a piece of fiction. XL2, you're the man when it comes to Saturn programming - is it even possible to get the audio processors and the VDPs to "talk" to one another? Put differently, is it within the realm of possibility for the sound suite to generate visuals? Is there any way we could use hardware diagnostics to test this on Digital Dance Mix? Please pardon my lack of technical knowledge here. |
TrekkiesUnite118 | Oct 27, 2022 | |||||
Define Audio Processors. Define Talk to one another. Define Sound Suite. Define Generate Visuals. |
adbevers | Oct 27, 2022 | |||||
The name sounds familiar, but I basically lost all contact within the community when the site I worked on (dimension-sega and dimension-s.com) shut down around the time Sega transitioned from the Saturn to the Dreamcast. I stuck around the scene on Dave Z's site for a little while, but as is usually the case life/school/gainful employment caught up with me, Sega pulled out of the hardware business, and I lost the few ties I did have. |
XL2 | Oct 27, 2022 | |||||
No, it's not, the 68k cpu can't access the rest of the system by itself from what I remember from the technical documents. It's also way too weak to be useful for 3D graphics. But it can help with 3D : 3D sound that is. There is even a third party library for it. But even for the SH2 it takes so few ressources to do 3D sound that I find it a bit useless to offload it the 68k. if you want to do more fancy tricks like doppler effects and all, then perhaps using the 68k for it could be justified. |
the Sega Saturn Pope | Oct 29, 2022 | |||
OK, and none of the other sound chips could really be utilized either? The sound DSP versus the "regular" DSP? I think, if that is the case, then we can discount anything that source has to say on the matter. This includes that Rich Leadbetter or others saw VF3 Saturn at the Tokyo Game Show. This also discredits the idea that Shining Force 3, Burning Rangers, etc. use any of the audio processors for nonaudio purposes. Sam Pettus' book (Service Games: the Rise and Fall of Sega) details a "considerably scaled down" Saturn conversion that was finished on July 3rd, 1998. That's a pretty specific date and I am trying to find corroborating sources. If true, I would imagine that it is as people in this thread have detailed, i.e., pretty similar to Fighters Megamix in terms of visuals. Maybe even a bit worse graphically due to the need for more detail in the arenas. I will keep trying to find out more related to the fate/existence of this version. I am still going to prioritize attempting contact with those who worked on VF2 (Saturn and arcade), VF3 (arcade), Digital Dance Mix (Saturn), Fighters Megamix (Saturn), Shenmue 1/2 (Dreamcast/Xbox), and any combination of the above. |
rorirub | Oct 31, 2022 | |||||
I recall Fighters Megamix including a document on the CD where they flat out state that the game is an apology to VF3 fans. And that game felt more like Vipers with VF characters, the gameplay felt different from VF (even if you put it to virtua fighter mode). It also had no real arcade mode.
Most of these rumours saying that a game uses the audio chips is just magazine articles from the 90s that confuse the SCU DSP with the sound DSP, or confuse the Slave SH2 for the 68k. SF3 uses the SCU DSP for sure, I think it was to help data decompression for the fight scenes. Most games run the standard 68k sound library on the 68k. There's something like a dozen titles that do something else, of which VF2 and Kids use a custom sound engine and I think Exhumed has the 68k idle and uses the SH2 for sound instead. The only game I know of where it is confirmed that the 68k is used for number crunching is Doom, not exactly the best reference for using the 68k for game processing. |
the Sega Saturn Pope | Oct 31, 2022 | |||||
The manual here is apology-free. I doubt there was one in the EU/JP release either since it was only officially canceled in September '98. I see what you mean though I think. On the CD as in you pop it into your computer and get a .txt file like so many audio CDs in the 90's? Does anyone have a physical copy of Megamix that can confirm this is the case? |
Knight0fDragon | Oct 31, 2022 | |||||
not the manual, on the disc itself probably the abs.txt or bib.txt file |
XL2 | Oct 31, 2022 | ||||
That's the read me file, I tried to translate some parts with Deepl translator. Here is some part of it, while they do give apologies I guess it's more a Japanese cultural thing than a recognition that VF3 won't happen (I really doubt that at the end of 1996 they could know already that it wouldn't happen) :
|
jollyroger | Nov 14, 2022 | |||
So, I had looked into this a few years ago and then stopped for lack of time. The Yamaha SCSP has a programmable DSP that can operate at ~44000 Hz, it has an internal multiply/add pipeline (integer) and due to its capabilities for effects like reverb and echo, it can operate on circular buffers of audio data. Technically speaking it may be therefore possible to use it to perform some linear algebra, feed it with the 68k and control the whole lot with one of the SH2s… Does it make any sense to use it for such a purpose? I really doubt it. At some point I still want to try it just for a laugh though As far as accelerators are concerned, they have attempted this multiple times. On top of the supposed cartridge mentioned in the interview, they had a prototype ST-V board with two SHARC DSPs and an extra VDP1. They never used it commercially, as the Dreamcast was already close to production and way faster, but as we all know they used a similar design with their NAOMI2… Fun stuff… |
antime | Nov 14, 2022 | |||
For each audio sample generation, the DSP will execute its 128-step program top to bottom, without loops, branches, or any other kind of flow control. There's no way I can think of to synchronize with the DSP except trying to use the timers to count samples and hope for the best. Except for bragging rights, I can't really see a reason for trying to use it for non-audio tasks (but if you do, also think about what it might be possible to calculate using the modulation slots). |
jollyroger | Nov 14, 2022 | |||
You are 100% correct, there is absolutely no reason other than "because"... |
nando | Nov 14, 2022 | |||
A related topic I've been thinking about for a while - how about developing some sort of standard Saturn benchmark suite that could be used to test this kind of stuff? Basically a standard demo (or several demos) of typical tasks - polygons per second, textured fill rate, sprite manipulation, physics calculations, VDP stuff, etc - that has all the assets built, and you'd just have to compile it using whatever crazy algorithms and DSP tricks people can think of to make things run better? Even as a learning tool. Start it basic - a pure flat shaded cube rendering speed test, and then add layers of effects and other tests on top of that foundation. |
nando | Nov 14, 2022 | |||||
Probably, but who knows? There must be many things people have never tried. I think part of what attracts people to the Saturn indie scene is the possibility of "what if?". Saturn developers in 1995 had a lot going against them - bad managers, time crunch, poor support, lack of tools, little knowledge of 3D development, etc. What if you removed all of those restrictions? |
XL2 | Nov 15, 2022 | |||
The coolest thing I have seen with the SCSP DSP is the ADPCM decompression someone did as a test. He could decompress and play up to 8 compressed samples with it. The PS1 had hardware support for 24 compressed samples AFAIK, so 8 isn't that impressive, but still cool enough. |
jollyroger | Nov 15, 2022 | |||||
Definitely, that is very impressive use of the DSP capabilities. |
< Prev | 1 | 2 | 3 | Next> |