/vr/ - Retro Games

>>4578589
>Sega of Japan won
They won because their alternative was more viable.
Had the 32x been chosen Sega's console would have bombed even more than it did in America and Europe, while not having the popularity the saturn had in Japan. Having a standalone 32X console as the successor of the megadrive was a short-sighted decision.

>>4578589
Except Kalinsky wanted to keep running Genesis into 1995 or 1996 and launch a new console then.

Possible technology that could have been available:

https://forum.digitpress.com/forum/showthread.php?135014-a-SEGA-quot-64X-quot-upgrade-for-Saturn-that-never-happened-and-some-quot-what-ifs-quot

>>4578589
The 32X wasn't enough of a leap over the Genesis to ever be worth it in the first place, and was a poorly shortsighted move that cost Sega marketshare and brand value. And it was a lame attempt at combatting the Jaguar, which was a total flop. They would have been better off doing nothing.

>>4578589
32X should have been cancelled mid- development and remaining work on it split into two factions: 1) those being added to the Saturn dev team and 2) folks pushed to the Sega CD development team.

Can someone tell me if the sega cd project started before or after nintendo had plans to make the snes cd-addon?

>>4582040
The sega cd started because of the turbografx 16
and the snes cd started because of the sega cd/turbografx

https://segaretro.org/History_of_the_Sega_Mega-CD

>>4581976
Oh God why is Sega of Japan so fucking pigheaded?

Sega gonna Sega

>>4578589
32X and Saturn were released like within a week of one another. As per interviews, Sega believed that the 32X would be a "budget console" and the Saturn a "premium console". A market split between two entirely different architectures obviously didn't go over well with developers who practically ignored the 32X.

>>4581976
>Lockheed-Martin Real3D chip in a consumer-grade upgrade
How can anyone post that kind of shit while being serious? Model1/2/3 boards were hella expensive even by arcade board standards thanks to said Lockheed-Martin chips. How could they hope selling to consumers a console based on the most expensive part of said board?
>>4582249
'Cause they knew it wouldn't have been possible to make a model2-based console that didn't have a higher pricetag than even the neo geo AES.

>>4582317
>A market split between two entirely different architectures
Both 32X and Saturn use twin Hitachi SH-2 CPUs though. The only other differences in architecture is the former draws polygons in software and the latter in hardware.

>>4582725
The best possible outcome for the Saturn would have been Kalinske successfully securing the N64's chip from SGI and actually wiring some decent RAM to it. It would have performance on-par with the Model 2 that way, at a low low price.

>>4582743
Better than the actual saturn, but not on-par with the Model 2.

>>4582761
True, I guess so. But making it on par with the Model 2 wouldn’t have been too hard. If the N64’s FPU in the CPU was removed and replaced with RSP, and SGI’s chip was just two RDP pixel pipelines, that would match (possibily exceed) the Model 2 for performance easily.

>>4582743

They evaluated it once and thought it looked and sounded like crap. They looked at it again and said the same thing. Can't really fault them there.

>>4582725

Well, they were already selling model 3 level tech in 1998(DC), so model 2 tech is possible in 1996.

>>4582919
>They evaluated it once and thought it looked and sounded like crap. They looked at it again and said the same thing.
They only looked at it once. And it was quite infamously a political decision rather than a technical one. SoA found no fault with what SGI were presenting. Not to mention it would have been foolish to dismiss the world's biggest 3D graphics technology company on the basis of hardware that was 2 years out from finished.

>Can't really fault them there.
SGI managed to fit into one chip the equivalent of what took Sega about 5 chips on Saturn. And it was significantly higher performance at that. SGI's chip has about 8 times the alpha blending performance of Sega's VDP1.

Wake me up when the Saturn could do this. Now imagine what it could do in Sega's hands with a CD drive and no gimped memory.

SoJ were quite frankly, idiots. And as long as they could have put fast enough RAM into the system in order to keep up with the graphics, Sega could have made a home Model 2 system with a few tweaks as specified here >>4582775

>>4582775
>that would match (possibily exceed) the Model 2 for performance easily.
>easily
The SGI chipset would have to be more than 5 times more powerful that it is when used in the nintendo 64 to match (not even exceed) some of the Model 2 hardware characteristics.
>>4582924
The leap forward in term of consumer 3D rendering chip between 1994 and 1996 is meaningless compared to what happened between 1996 and 1998. You didn't have the same kind of competition before and after the introduction of the voodoo.
>>4583026
No, Sega of Japan saw (rightly so) Sony as a major menace and made the right decisions by readying some hardware to match the newly coming playstation. When the SGI chipset was presented to Sega of Japan, it had major issues an was still at the stage of prototype. They didn't design the best chipset around and kinda rushed the saturn release a bit too much, but waiting 2 more years, using an even more expensive chipset and having faster (more expensive) RAM than the N64 had would have made the system so expensive that it wouldn't have been affordable and would have lost more marketshare than it did (don't forget that a CD drive + all that is needed to handle it is WAY more expensive than a simple card-edge connector).

>>4583026
>They only looked at it once. And it was quite infamously a political decision rather than a technical one.

Wrong. Literally the first result in google:
http://www.sega-16.com/2006/07/interview-tom-kalinske/


>I remember Joe Miller and I were talking about this, and we had been contacted by Jim Clark, the founder of SGI (Silicon Graphics Inc.), who called us up one day and said that he had just bought a company called MIPS Inc. which had been working on some things with some great R&D people, and it just so happened that they came up with a chip that they thought would be great for a video game console. We told them that in the U.S., we don’t really design consoles; we do the software, but it sounded interesting and we would come over and take a look at it. We were quite impressed, and we called up Japan and told them to send over the hardware team because these guys really had something cool. So the team arrived, and the senior VP of hardware design arrived, and when they reviewed what SGI had developed, they gave no reaction whatsoever. At the end of the meeting, they basically said that it was kind of interesting, but the chip was too big (in manufacturing terms), the throw-off rate would be too high, and they had lots of little technical things that they didn’t like: the audio wasn’t good enough; the frame rate wasn’t quite good enough, as well as some other issues.

So, the SGI guys went away and worked on these issues and then called us back up and asked that the same team be sent back over, because they had it all resolved. This time, Nakayama went with them. They reviewed the work, and there was sort of the same reaction: still not good enough.

>>4583026
>They only looked at it once. And it was quite infamously a political decision rather than a technical one.

Wrong. Literally the first result in google:
http://www.sega-16.com/2006/07/interview-tom-kalinske/

>I remember Joe Miller and I were talking about this, and we had been contacted by Jim Clark, the founder of SGI (Silicon Graphics Inc.), who called us up one day and said that he had just bought a company called MIPS Inc. which had been working on some things with some great R&D people, and it just so happened that they came up with a chip that they thought would be great for a video game console. We told them that in the U.S., we don’t really design consoles; we do the software, but it sounded interesting and we would come over and take a look at it. We were quite impressed, and we called up Japan and told them to send over the hardware team because these guys really had something cool. So the team arrived, and the senior VP of hardware design arrived, and when they reviewed what SGI had developed, they gave no reaction whatsoever. At the end of the meeting, they basically said that it was kind of interesting, but the chip was too big (in manufacturing terms), the throw-off rate would be too high, and they had lots of little technical things that they didn’t like: the audio wasn’t good enough; the frame rate wasn’t quite good enough, as well as some other issues.

>So, the SGI guys went away and worked on these issues and then called us back up and asked that the same team be sent back over, because they had it all resolved. This time, Nakayama went with them. They reviewed the work, and there was sort of the same reaction: still not good enough.

>>4583026

32X was basically a model 1, and Saturn already was a model 2 lite. It could run Daytona. It could run HoTD. All it needed was time in the form of Moore's law to close the gap.

>>4583703
>The SGI chipset would have to be more than 5 times more powerful that it is when used in the nintendo 64
SGI’s chip: 62.5 MPixel/s fill rate
Model 2: 110 MPixel/s fill rate

It only has twice the fill rate. This is with a comparable number of features turned on by the way. So doubling the pixel pipelines already puts it beyond Model 2 level. Also Nintendo downclocked it from 66 to 62.5 before launch, so that’s any ther thing.

As for geometry processing
SGI’s chip: 128-bit vector unit at 62.5MHz
Model 2: A chain link of like 8 FPU each with 32-bit precision running at 16 MHz each

So not too far off enough.

>using an even more expensive chipset
Here’s where you are wrong. SGI’s chipset was highly intergrated. Saturn had to have several chips to match its functionality. Nintendo’s manufacturing costs on the N64 were reputed to be 4 or 5 times lower than Saturn’s.

>having faster (more expensive) RAM than the N64 had would have made the system so expensive that it wouldn't have been affordable
That’s where the savings would have gone. All it needed to match or exceed Model 2 was twin RAMBUS channels to GPU (still at 4 MB) and some shitty cheap DRAM for CPU (2 MB would be fine).

They rejected SGI’s chip because they had no faith they would fix it (they eventually did) and because behind the scenes there was a classic not-invented-here mentality.

>>4584754
Alright they did check it, but it was clearly political. I can’t see how even a flawed prototype SGI chip would have been worse than say, VDP1, a garbage chip which Sega green lighted as OK.

https://www.youtube.com/watch?v=pOWZbydnlZE

>>4584764
32X wasn't anywhere near a Model 1. The ports to that system make that clear.

>>4584764
>32X was basically a model 1
The 32x was way less powerful than the model1. Sure, you could do decent model 1 game ports, but compare the 32X version with the original.
>>4585281
You're only comparing how fast it can write already rendered pixels, not the actual polygon calculation capabilities.
The model 2 can map 3 times the number of polygons the N64 can when using effects more advanced effects (and trilinear filtering instead of bilinear), and between 500k~900k with specular lighting depending on the texel count. It can transform 1.2 million flatshaded polygons using floating-point arithmetic (2 million fixed point), up to 880k gouraud-shaded polygon (1.5 million fixed-point).
You guys are seriously undersestimating Model 1 and 2 arcade hardware.

>>4585390
Those polygon figures account for the Model 2’s superior memory bandwidth as compared to other 5th gen consoles. They are not ‘memory neutral’ figures which I was trying to go for under the assumption that Sega would have improved the RAM.

Fill rate (at a given feature set level) is a good indicator of rendering throughput. It’s not just writing already rendered polygons - it’s including the entire pixel pipeline from rasterisation, to texturing, to filtering, blending, etc. The other part of the equation is transform and lighting, which doesn’t seem to be so far behind on N64 even as it stands. This is to be expected as the 128-bit vector unit it had was the most powerful component of the system relatively speaking (and it had to be, given it was tasked to process sound too).

The N64 can do trilinear filtering in 2 pixel cycles, exactly like the Model 2. I think you are underestimating just how good SGI’s chip was and just how badly it was gimped by the memory Nintendo paired with it. It truly demonstrated an improvement in graphics fabrication technology since the release of Model 2 in 1993.

Again, Sega were idiots not to use SGI’s chip and attach some fast memory to it.

>>4583026
What's this game? It looks nice

>>4585586
Gran Turismo

>>4578589

32x is a lot better than people give it credit for, it just gets shit on for a few (understandable) reasons:

>normies don't remember it because it sold way too poorly
>tiny library
>Doom port's music is astonishingly bad despite the fact that the system could have done better, and Doom is one of the best and most influential games ever made so all the hardcore gaming nerds who know of the 32x know of Doom 32x having awful music. Leaves a nasty stench on everyone's 32x memories.

The games we did get were a good step up from Genesis games (especially the 3D ones), and those were just early games. The console died too fast for us to see it's full potential, there wasn't enough time for developers to master the hardware.

the 32x could have ruined my childhood but a benevolent vendor told us not to buy it and wait for the saturn

>>4585767
Nigger the 32X sucks because it barely has games.
>The games we did get were a good step up from Genesis games
Quit being a contrarian, they looked and played barely different from Genesis games. The handful of 3D games were also shit besides Virtua Racing and VF1.

Even as a Segafag I can admit it was an abortion of am add-on. The Sega CD has a more interesting library than the 32X

>>4584764
>32X was basically a model 1, and Saturn already was a model 2 lite.
Please tell us you aren't this retarded.
>It could run Daytona. It could run HoTD.
Yeah, you are retarded. Don't invoke moores law without sounding fucktarded.

>>4585586
World Driver Championship /vr/s favorite gran turismo like on the 64

>>4585803

It also had the Star Wars arcade game and another, even better space shooter that I forgot the name of.

I'll concede that both the Genesis and Sega CD are better due to their libraries, and I would never recommend anyone actually buy the hardware. But it's worth emulating just to see how it's not a complete abortion, it's more like an autistic child that showed promise before being drowned by his mother.

Sega CD, X32 and Saturn are all examples of how a company that's doing fine shoot itself in all limbs because it had to be first to the punch.

>>4586771
>Sega CD
>first to the punch
What is PC Engine CD

>>4587408
That's right, kids! The Turbo Duo is the FIRST CD game system on the market! Don't listen to those FEKA goons!

>>4586771

You forgot Dreamcast.

>>4587489
I'd like to think they kind of wised up by then but yeah, the Dreamcast had some flaws Sega could've fixed.

>>4584764
The model 2 was infinitely more powerful than the sega saturn, I don't think either of your comparisons are valid outsite of the dates coinciding and thus ports from each arcade board were made to the individual systems.

>>4587423
It doesn't even compare!

>>4585808
>It wouldn't have had at least double the performance in 18 months when they would have had 2 years to do it

Kill yourself my man. What do you think Moore's law does?

>>4588219

It's retarded to say it's infinitely more powerful when it could already run the games, it just had to catch up in terms of frame rate and draw distance while having the advantage of doing it in 240p instead of 380p.

32X seems like it'd be a great emulation device for older consoles with a flashcart if someone bothered to homebrew up some emulators. I'm sure it's more than fast enough for NES, possibly even SNES.

Then again, the controllers aren't really ideal for emulating Nintendo consoles I guess.