Hideki Sato on the Sega Saturn (incredible new interview)

[Gryson]

Recently, an interview was conducted with Hideki Sato on his life and his time at Sega as part of an oral history research project documenting the game industry in Japan. The interview transcription is over 150 pages long. If you don’t know, Sato was in charge of Sega’s consumer R&D department for many years and took part in the development of many/all of Sega’s consoles. He later served as Sega’s president.

I’ve translated for Sega-16 a few interesting snippets of the final part of the interview that relate to the Sega Saturn.

The full transcripts can be found here: Interview Transcripts

Highlights
• More information on Sega of America’s desire to use the 68020 in the Saturn
• Changes made in the Saturn design in response to the PlayStation
• Difficulties that third party developers had with the Saturn, and the lack of early support from Sega
• Sony’s strong support for third party developers
• Sega intentionally limiting Saturn production due to being unwilling/unable to bear the losses they were taking on sold units
• The strong advantages Sony had in manufacturing cost and flexibility
• Ken Kutaragi telling Sato that Sega should become a third party to Sony

On the MC68020 and why the SH was chosen:

Motorola had the MC68020, the successor to the MC68000. It was a strong-selling 32-bit CISC microprocessor. Sega of America, who were developing their own 16-bit Genesis games, wanted to use the MC68020 in the Saturn. That would have allowed for essentially updated versions of the current types of game software, and the development libraries could easily be done. They wanted to go for forward compatibility.

However, from my viewpoint, this lacked the necessary “jump” in technology. I thought that it might be okay to move forward with such a continuation of the current technology, but all the same, I felt we needed to move in a new direction, to change things up. Compared with the 16-bit generation, we needed to move away from mask ROMs, from solid-state memory, which was too expensive. CD-ROMs had become cheap, but the technology was no longer new. The PC Engine had already been using it for years. We needed something more.

At the time, Hitachi happened to be developing the SH processor. After seeing the specs, I was impressed by its high performance. I decided to go with it, even though it was still in development (this was a very rash move for me). The SH is a RISC (Reduced Instruction) CPU, and at that time, NEC was also developing one, the V Series. I felt that Hitachi’s SH was good, so I went with that.

On how the Saturn design changed in response to the PlayStation:

The Saturn actually had just one CPU at the beginning. Then Sony appeared with its polygon-based PlayStation. When I was first designing the Saturn architecture, I was focused on sprite graphics, which had been the primary graphics up to that point.

So I decided to go with polygons (due to the PlayStation). However, there weren’t any people at Sega who knew how to develop such software. Of course, we had Yu Suzuki in the arcade department, but I couldn’t just drag him off to the console department. He was developing titles like Virtua Fighter and Virtua Racing. The expertise of all of the developers we had was in sprite graphics, so there seemed no choice but to go with sprites. Nevertheless, I knew we needed polygons. Using various tricks, adding a geometry engine and so on, I changed everything. In the end, just like the PlayStation, we had pseudo-polygons built on a sprite base. I felt no choice but to design a sprite-based architecture. Having said that, after some significant progress, pseudo-polygons did represent a “jump” in graphics in a certain way. There was a distinction of sorts. The processor was very powerful and could support 4,000, even 5,000 sprites, and I thought we could make the graphics work using a sprite engine after adding the Yamaha and such.

It seemed like we were finally nearing completion. Then, the final PlayStation was revealed. It supported 300,000 polygons. Well, that was ultimately a bunch of lies, but… When you compared the Saturn with the PlayStation, we were completely missing something. The response that I chose was to add another SH processor, so we ended up with two SH-2s. By chance, the SH supported two-way cascaded data transfer. You could add a second processor and connect them in a cascade and get multi-CPU performance. When you get to about the PlayStation 3, multi-processors had become common, but the Saturn was the first home console to use multi-processors. So I added a second SH-2, but I felt that the ‘impact’ was still weak. Well, the SH-2 is a 32-bit processor, and we had two of them, so we could call the Saturn a 64-bit machine. It’s a dirty way of getting to 64-bits. But we revealed the CD-ROM-based Saturn using 64-bits as our sales point.

On the difficulties of developing for the Saturn:

At the beginning, there was no compiler. You had to program the SH in assembly. The people at Sega were good at assembly. That’s all they had been using on the MC68000. C, C++ were too slow to use.

However, third parties struggled with programming the SH in assembly, and there were two of the CPUs along with a CD-ROM. We asked third parties to make games, but without development libraries, they couldn’t do anything. They’d take a week and barely even be able to get something to display on the screen, let alone be able to start making a game. Our third party support was awful. The hardware was incredibly difficult to use. However, if you worked with it a bit, you could get a ton of sprites, with scaling and rotation and so on.

On Sony’s support for third party developers:

Sony was good at supporting PlayStation third party developers. Why? Sony didn’t have a development department. They didn’t have a software department. What do you do if you don’t have a software department? You ask somebody else. Sony went to Namco, to Taito, to Konami. They said that they were putting together a game console called the PlayStation, and they invited these companies to develop games for it. Sony exerted all its efforts on supporting third parties and enhancing their collective powers. Sony CEO Norio Ohga himself went to talk to the third parties. From their perspective, it was a big deal for Ohga to come and ask this. From Namco’s viewpoint, if they put out Tekken, they could compete evenly with Sega’s Virtua Fighter.

The number one game in the PlayStation world was Ridge Racer. And Konami being Konami, they had their typical games. It’s obvious that the PlayStation had the better games. No matter how much effort Sega put in on its own, it wasn’t going to be enough.

So Sony went to Namco, Taito, Konami, and others, and they said here are the specs, and don’t worry, there aren’t two CPUs or anything difficult like that. They said the PlayStation will be easy to develop for, and here are all the development libraries we’ll put out. Sony had a very easy-to-use SDK (Software Development Kit). And Ohga himself was making these offers, and the third parties were told they could port all of their own titles, and so on. With all of that, it certainly seemed like the PlayStation was better.

On Sega’s losses associated with the Saturn and their response:

So we released the Saturn in 1994, and as I said before, there were two SH-2s. In addition, memory was expensive at this time, and we were using a large amount, so costs were very high. For each Saturn sold, we lost about 10,000 yen ($100). That’s how the hardware business works. But the goal was to recoup the losses from software royalties. If there are lots of third parties, lots of games sold, and we get 2,000 yen for each, it’s possible. However, if software sales are weak, and for each console sold, we’re ultimately losing 5,000 – 6,000 yen, what’s going to happen from the business perspective? We’re going to stop selling consoles. This later became a huge problem.

Every month, or even every week in Sega’s case, we had meetings to examine the current situation. Each department would report on where it stood in relation to its goals. So, imagine if the sales goal for the end-of-year sales war is, say, 3 billion yen, and the profit goal is 300 million yen—but wait, the profit is in the red. That profit is a very important factor, so what does the business side do? They decide that it’s not necessary to have sales of 3 billion yen. Instead, 2 billion yen will do. In other words, they stop selling 1 billion yen’s worth of hardware. That way, if each unit sold is losing 5,000 yen, and we extend that to 20,000 units, that’s 100 million yen lost. By stopping the sales of 20,000 units, in a way that becomes 100 million yen in profit. So they slammed on the brakes in terms of unit distribution. Even though there were people that wanted to buy the console, Sega didn’t want to sell it, because the more they sold the more they went into the red.

From the perspective of the third parties, they saw that Sega was curbing the sales of the Saturn. The more consoles there were, the more games would be sold. But if console sales were being limited, then this created a serious problem. As they say, poverty dulls the wit. This led to a negative feedback loop.

On Sony’s manufacturing advantages and Kutaragi’s invitation to become a third party:

To launch a new console, you really need 50-60 billion yen at the least. You have to sell those first million units. If your costs are 30,000 yen per unit, then that comes to 30 billion yen for 1 million units. And you have to design the hardware and create the electronics, make the molds and do the tooling, and this will soon use about 10 billion yen. And then you have to create the games and do advertising. You need about 500-600 people. Without all this, you can’t launch a home console. You can’t do it little by little. You really have to go all in.

Sony had annual sales of 3 trillion yen. They made their own CD-ROM drives. They had their own semiconductor factories. Once when I was talking with Ken Kutaragi [the creator of the PlayStation], he said “Hideki-chan”—he refers to me using the “chan” diminutive—“Hideki-chan, there’s no way you can beat me. Where are you buying your processors? From Hitachi. From Yamaha. What about your CD-ROM drives? You’re buying everything. By buying from Hitachi, Hitachi is profiting. You can’t make anything yourselves. We can make everything ourselves, including custom parts. We have our own factories.” Near Nakashinden, they had a huge factory where they made audio equipment that they were using for the PlayStation. Their cost structure was completely different.

“That’s the way it is, Hideki-chan,” Kutaragi told me. “So quit the hardware business. Why not just do software? We’ll give you favorable treatment.” He wanted us to go third party. We had been going for so long in the hardware business, for better or worse, and to go third party now? We had been half-heartedly successful in America once, and this made it impossible to quit the hardware business. Maybe if the Mega Drive, the Genesis, had been a failure, things would have been different. But we had a strange taste of success.

At that time, Sega’s brand image was incredible. When you powered on a Sega console, ‘SEGA’ would always appear first. Even if it was a third party game from Namco (or anybody else), Sega’s name always appeared first, followed by Namco’s. So anybody that had a Sega console, it didn’t matter what game they played, they would see Sega’s name. This helped plant the Sega brand in peoples’ minds. This was incredibly effective. To go from that to a Sony third party… Well, we had already started so it was too late.

I would have a polite dinner with Kutaragi about once every three months. He’d tell me that because we released a console last time, they would be the ones to do so this time. We are the same age, although he’s two or three months older. I would call him the polite “Kutaragi-san,” although sometimes I’d call him “Ken-chan.” Because I was two or three months younger, he’d say “Hideki-chan, please give up!”

So we released the Saturn, and in the end, it came down to software. It’s obvious, but what do consumers look forward to? They want fun games. And that’s where we failed.

There is, of course, much more content than this, but it is beyond me to translate everything. I've tried to focus on the good stuff.

I believe this is the most detailed description of the Saturn's fate that we've ever gotten from the Japanese side.

Thoughts?

[Leynos]

All the wrong choices. Should have worked cohesively with SOA to come to a middle ground on how the hardware should be designed.

[Team Andromeda]

Thought the SH2 was added at the last minute but then had some doubt so nice that was cleared up once and for all.. Also nice to hear that unlike the at rubbish that came from Tom; on how he was looking to work with Sony that he knew the Saturn spec was rubbish.. It turns out, that not only did SEGA not know about the PS, SOA was pushing for a lesser spec Saturn, with a lesser CPU. If Tom and his bunch at SOA had their way, the Saturn would have been even more outclassed for 3D Vs the PlayStation.


Little shocked to see SEGA were losing so much on each Saturn sold at the start.

The tools and lack of Sonic and transparent effects are for me the major failings. Thanks for the interview and I think it finally shows what a lair Tom really was.

[Barone]

So the mighty geniuses at SOA were willing to choose a puny 68020? Lol.

Thanks a lot for the translation.

[foxnoodles]

we had Yu Suzuki in the arcade department, but I couldn’t just drag him off to the console department

Wut? He couldn't grab a guy to help you out with the architecture of the main product? wut?

The processor was very powerful and could support 4,000, even 5,000 sprites, and I thought we could make the graphics work using a sprite engine after adding the Yamaha and such.


It seemed like we were finally nearing completion. Then, the PlayStation was revealed. It supported 300,000 polygons.

Right, and he had no idea...

At the beginning, there was no compiler.

The people at Sega were good at assembly. That’s all they had been using on the MC68000. C, C++ were too slow to use.

How? Is this Sega or some kind of FeiHao booth in China? Stuff like this doen't happen lol even in indie companies.. Well hire people god damn it and come up with everything needed.. that's how it's done?


p.s. Thanks for the translation this is kinda huge.

[Gryson]

Wut? He couldn't grab a guy to help you out with the architecture of the main product? wut?

That might be a fault in my translation. Based on previous stuff he and others have said, I think he means that Sega didn't have anybody at all who knew how to make polygon-based games (not anything to do with the architecture). There was a big concern that their developers were really good at sprite-based games, and they'd have serious issues if the system was entirely polygon-based. He meant that he couldn't just reassign Yu Suzuki to start developing console games, since he had his hands full with arcade game development.

Right, and he had no idea...

Well, something caused him to add the second SH-2 at the last minute. It seems likely that, as he says, the strength of the PS caught them off guard. What are you thinking?

How? Is this Sega or some kind of FeiHao booth in China? Stuff like this doen't happen lol even in indie companies.. Well hire people god damn it and come up with everything needed.. that's how it's done?

I guess that's what happens when you're using a brand new processor, and you add a second one at the last minute! It was likely very difficult to develop an SDK when your most experienced programmers are struggling to figure the hardware out (see Yuji Naka's comments on how nobody knew how to program the Saturn). And they seemed to put a priority on getting games out.

p.s. Thanks for the translation this is kinda huge.

You're welcome!

[foxnoodles]

That might be a fault in my translation. Based on previous stuff he and others have said, I think he means that Sega didn't have anybody at all who knew how to make polygon-based games (not anything to do with the architecture). There was a big concern that their developers were really good at sprite-based games, and they'd have serious issues if the system was entirely polygon-based. He meant that he couldn't just reassign Yu Suzuki to start developing console games, since he had his hands full with arcade game development.

Well, something caused him to add the second SH-2 at the last minute. It seems likely that, as he says, the strength of the PS caught them off guard. What are you thinking?

I think both topics lead to this guy being a retrograde or something, as he stated he had this "strange taste" of being a winner (on sprite based market) he never even mentioned that Arcade was a huge part of Sega's business and success back in the day and it was all about 3D for more than a decade, not to mention that Sega released Virtua Racing with SVP, which was a polygon processor, but he didn't even try to ask for help from people who actually do it? It sounds weird to say the least. Kinda seems that he didn't wanted anyone else to be involved in this for selfish reasons? Cuz the whole story sounds like a very one man oriented deal. If what he says is true, then he singlehandedly destroyed Sega. Also where was the Marketing dep head all this time lol?


As for the software, well again, He coulda hired "spec ops devs" from the original CPU manufacturer(which in 99% of the time provides you with at least a simple assembly translator for drivers or something like that and even a hardware demo, the one they use for testing), don't know, sounds fishy.
Also he's contradicting himself when he says that at on hand the goal was to get software loyalties and on the other hand he never did anything that would help to achieve it(no SDK, no SDK after 6 months, year, no bootable firmware updates that could be loaded from a CD, nothing)

[Gryson]

I think both topics lead to this guy being a retrograde or something, as he stated he had this "strange taste" of being a winner (on sprite based market) he never even mentioned that Arcade was a huge part of Sega's business and success back in the day and it was all about 3D for more than a decade, not to mention that Sega released Virtua Racing with SVP, which was a polygon processor, but he didn't even try to ask for help from people who actually do it? It sounds weird to say the least. Kinda seems that he didn't wanted anyone else to be involved in this for selfish reasons? Cuz the whole story sounds like a very one man oriented deal. If what he says is true, then he singlehandedly destroyed Sega. Also where was the Marketing dep head all this time lol?

Keep in mind the timing here. The initial development period occurred right when Model 1 was just out. See his other quotes from 1998 interviews:

We were also stuck on whether to focus game development on sprite-based games, or new 3D CG games. Sprite-based games were what Sega had done up till then, so we had a lot of built-up experience there, both in a personnel and technology sense; it seemed like a waste to just throw it all away. And Sega only had a few internal teams dabbling in CG design. We therefore decided to give the Saturn the ability to handle both kinds of games, with a robust sprite and CG engine. However, although we’d separated the two engines well enough in a hardware sense, creating games for the Saturn turned out to be a little difficult. The software development libraries were also insufficient, so third parties saw the Saturn as a difficult system to develop for. We sold 5 million systems in Japan, but we struggled in the overseas market.

When the Saturn was being designed, the video game industry was right in the middle of the transition from sprites to CG. In the arcade world, you could see this contrast between the System 32 boards, capable of displaying 300,000 sprites, and the Model 1 boards which ran Virtua Fighter and showed the future of polygons. In order to not lose all the assets and know-how we’d accumulated in previous years, we first thought about basing the Saturn on the System 32 boards, but we inevitably realized that it would be best to have polygon and CG capabilities too, so we included both in our design. It was done in the spirit of having the best of both worlds, but it also kind of felt like we were splitting the baby, and not doing justice to either. (laughs)

There were two candidates for the CPU. The first, which Sega of America was pushing for, was the 68020. It had good compatibility with the 68000 processor and would be easy to use, but its limitations were also clear. The other option was the RISC CPU: it seemed much more powerful, but for several reasons, the risk was also much higher (just as the name “RISC” implies!). As it had always been with Sega, we needed a home console that would be powerful enough to handle our arcade ports. That being the case, we took the risky-but-idealistic path and selected the RISC processor, the Hitachi SH2.

http://shmuplations.com/segahistory/

As for the software, well again, He coulda hired "spec ops devs" from the original CPU manufacturer(which in 99% of the time provides you with at least a simple assembly translator for drivers or something like that and even a hardware demo, the one they use for testing), don't know, sounds fishy.
Also he's contradicting himself when he says that at on hand the goal was to get software loyalties and on the other hand he never did anything that would help to achieve it(no SDK, no SDK after 6 months, year, no bootable firmware updates that could be loaded from a CD, nothing)

It was a brand new processor in a very complex architecture. I think he realizes it was a mistake to expect much from it early on, but it does seem like Sony had their back to a corner.

[Silanda]

I see the 68020 comment is being used to bash SoA again, but that's not really fair unless there's some context: we don't know when they were saying this. If the Super-H series hadn't been released at that time, and AFAIK that came on the market in '93, this could have been in '92. If it was that early, and the expected release date was sooner than it ended up being, going with a familiar architecture could have had some appeal. Still, it seems strange to me that they would have wanted the 68020 unless we're missing something; that chip had been available since '84 and the 68040 had become available in '91.

With hindsight it's easy to see that going with a low powered chip would have been a mistake, however, in '92 the Playstation was still compatible with the SNES, and AFAIK the 3DO hadn't even been demoed. What was expected of future consoles was still unclear.

EDIT: I'm also confused about his comment regarding the ability to render 4000-5000 sprites. Is he talking per frame? If so, since the Saturn uses sprites/quads instead of polys, wouldn't that work out at the equivalent of 300000 polygons per second at 60fps?

[Barone]

I see the 68020 comment is being used to bash SoA again, but that's not really fair unless there's some context:

I was being playful given the fact that SOA personnel interviews usually paint SOJ personnel as idiots and the same just happened but in the opposite direction this time.

Of course, we would need to know the context to understand where they were coming from, also assuming that everything stated by Hideki is true and accurate (which most people here always do for SOA-related interviews).

But SOA's handling of the 32X did went in the same direction Hideki indicates here: "That would have allowed for essentially updated versions of the current types of game software, and the development libraries could easily be done. They wanted to go for forward compatibility.".
And that also makes sense IMO with the 32X having no specific components for 3D.

[Silanda]

I was being playful given the fact that SOA personnel interviews usually paint SOJ personnel as idiots and the same just happened but in the opposite direction this time.

No problem. To be honest, I was responding more to TA's comment that paints Tom Kalinske as a liar without sufficient proof.

But SOA's handling of the 32X did went in the same direction Hideki indicates here: "That would have allowed for essentially updated versions of the current types of game software, and the development libraries could easily be done. They wanted to go for forward compatibility.".
And that also makes sense IMO with the 32X having no specific components for 3D.

That's true. It kind of makes me wonder how complete the picture we have of the Saturn's development history is, and when a successor to the Mega Drive was first proposed.

[Barone]

Probably a far from complete one.

It would be interesting to know what kind of console or add-on SOA was really pushing for.
Conjecturing a little bit but based on what we know, it seems to me that what SOA wanted was an upgraded Genesis with a 68020, with some extra RAM, upgraded palette and probably a just a simple audio solution for the garbled PCM playback which would run regular Genesis games just fine.

Something which could actually make a lot of sense for 1993, especially if they were targeting that Jaguar audience.

It would also be very interesting to know which games from the competition potentialy ignited these discussions. I mean, it's hard to buy that it was all just due to DKC or the Jaguar release.
There might be a lot more to it than what we know; developers' feedback or something else that made they go for another hardware upgrade after the somewhat recent release of the Sega CD.

[Gryson]

EDIT: I'm also confused about his comment regarding the ability to render 4000-5000 sprites. Is he talking per frame? If so, since the Saturn uses sprites/quads instead of polys, wouldn't that work out at the equivalent of 300000 polygons per second at 60fps?

He's speaking about things that happened over 25 years ago, so I would take any specific numbers to be vague approximations (in other words, I don't think he's checking any specs when giving the interview). But yes, I think your math is correct.

By the way, the thing I find most interesting here is that Sato never seems to consider the Saturn a success in Japan. He makes it clear that the Saturn was costing Sega heavily, and that it pushed third party devs away.

His broad analysis is that the Saturn failed because the complicated hardware, lack of dev support, and Sega's unwillingness to fully commit pushed third parties away. This ultimately resulted in a system that lacked a variety of strong games, a system that Sega was trying to prop up alone.

I find this analysis much more interesting than the typical "Saturn failed because of Kalinske's surprise launch at E3" or "Saturn failed because of Stolar's comments at E3" approach (e.g. this article). As Sato says, it all comes down to the games.

[Team Andromeda]

No problem. To be honest, I was responding more to TA's comment that paints Tom Kalinske as a liar without sufficient proof.

In various interviews Tom made out how he was meant to have been In talks with Sony over Hardware. Yet t we have from the main Hardware man in the whole of the SEGA group, the confirmation that Sony making a console was a susprise..

Tom also told Retro gamer that he and his SOA team thought the Saturn was under powered Hardware and he knew that at the time.. Looking over how Tom made out the Saturn was in fact more powerful, back in the day We now also learn SOA was pushing for a Saturn with a less powerful CPU; hardly a major jump and would have been closer to the NEC FX in terms of CPU.. That's before one moves on how SOA or SOJ didn't get on, if we listen to Tom, never backed up the main SOA tech guys.

One doesn't need or have to try hard, to show Tom for the spin King and lair he is.. Also Sat-San doesn't hit out at SOA, even admits the tools were rubbish and simply says SOA wanted the lesser CPU for very good and sound game development development reasons.

Yet again, the Japanese show their class and respect

[Gryson]

In various interviews Tom made out how he was meant to have been In talks with Sony over Hardware. Yet t we have from the main Hardware man in the whole of the SEGA group, the confirmation that Sony making a console was a susprise..

Why do you say this? I didn't intend that in my translation (that it was a surprise). Edit: Made translation clearer.

My understanding is that Sato was first designing a sprite-based system, and then he heard about the PlayStation and decided he needed to incorporate 3D, and then when the PlayStation was officially revealed (or some such), he decided he had to include a 2nd SH-2.

The interesting thing related to Kalinske's Sony-Sega tie up is that it probably wasn't the incredible deal Kalinske thought it was. Sato and Kutaragi knew each other and had discussed Sega going third party for Sony. Sato explains clearly why this would never have worked due to the huge success of the Genesis. Kalinske called this "the stupidest decision ever made in the history of business," but I doubt he knew what discussions had actually taken place.