The Saturn's multi-processors: Ahead of their time or just a horrible design?
BelPowerslave
to mention it but never seem to find the time.
The Saturn's multi-processor layout: Was it too far ahead of its time?
It seems like *every* PC nowadays has either dual processors or a
processor with dual cores. The Saturn had two main processors, yet when
we see a lot the software it's hard to believe that that's the best two
processors, working together could do. I mean, sure, we got stuff like
Rally, VF2, etc that obviously have all the processors doing their
absolute best, but that shit was really rare.
Back then, two processors was unheard of, even in a PC, so I'm sure the
tools to make good use of them weren't quite up to par. Seemed like
Sega, and a few others, were the only ones who really knew how to handle
it.
So, what do we think? Was Sega really thinking ahead when they placed
multiple processors in the Saturn, or was it just a horrible, horrible
design error that scared off all but the really good coders and the
assholes(Probe, mostly) who just simply ported their software rather
than recoding?
Bel
--
Whip Ass Gaming: http://www.whipassgaming.com/
"I am Andrew Ryan, and I'm here to ask you a question. Is a man not
entitled to the sweat of his brow? 'No!' says the man in Washington, 'It
belongs to the poor.' 'No!' says the man in the Vatican, 'It belongs to
God.' 'No!' says the man in Moscow, 'It belongs to everyone.' I rejected
those answers; instead, I chose something different."
- Andrew Ryan, BioShock
Kendrick Kerwin Chua
BelPowerslave <b...@whipassgaming.com> wrote:
<snip>
>So, what do we think? Was Sega really thinking ahead when they placed
>multiple processors in the Saturn, or was it just a horrible, horrible
>design error that scared off all but the really good coders and the
>assholes(Probe, mostly) who just simply ported their software rather
>than recoding?
I'm pretty sure that this was just an extension of the Sega CD design.
Slapping a CD drive on the side of the Genesis brought with it a second
Motorola 68000 processor. Sega always designs their current game hardware
with an eye on the previous design, and so I'm pretty sure this is why you
ended up with two SH-2 chips in the Saturn. I don't believe that the
design itself was an error, but I do think it was a mistake to leave the
complex programming tasks to the third parties instead of providing a
fully functional API.
-KKC, considering a GPS purchase.
--
-- Home of the amphibious hybrids! Buy your next car from | kendrick
Innsmouth Auto, on Highway 128 just before the old pier. | @ io.com
Where our goal is to sell you a 'Dagon' good automobile! |
Scott H
> I've been thinking about this for a long while now, and I keep meaning
> to mention it but never seem to find the time.
>
> The Saturn's multi-processor layout: Was it too far ahead of its time?
> It seems like *every* PC nowadays has either dual processors or a
> processor with dual cores. The Saturn had two main processors, yet when
> we see a lot the software it's hard to believe that that's the best two
> processors, working together could do. I mean, sure, we got stuff like
> Rally, VF2, etc that obviously have all the processors doing their
> absolute best, but that shit was really rare.
>
> Back then, two processors was unheard of, even in a PC, so I'm sure the
> tools to make good use of them weren't quite up to par. Seemed like
> Sega, and a few others, were the only ones who really knew how to handle
> it.
>
> So, what do we think? Was Sega really thinking ahead when they placed
> multiple processors in the Saturn, or was it just a horrible, horrible
> design error that scared off all but the really good coders and the
> assholes(Probe, mostly) who just simply ported their software rather
> than recoding?
>
> Bel
I've made the point several times, and finalized my wording in the new
Sega CD - 3DO - Jag - 32X section on Gamepilgrimage. Sega was both
looking forward and backward when they designed the Saturn. Multiple
CPU architecture was and is the future of graphic architecture design.
The PS2, Xbox 360, PS3, some of Sun Microsystem's cutting edge hardware
and all new PC CPU designs prove my first point for me. The dual System
16 boards that ran games like After Burner were, as far as I know,
Sega's first foray into the idea.
Furthermore, the lack of hard coded special effects and development kits
were actually the result of 3rd party requests. As we both know, many
3rd parties distinguished their Genesis games with custom special
effects. As I understand it, some developers actually told Sega that
they would not sign on for a new Sega system unless it allowed them to
code in assembly to more processors than the average developer could
possibly take advantage of. You see the effect in the games, big name
developers who actually dedicated resources to Saturn development
created games that blew the socks off the EAs and Akklaims.
Unfortunately for Sega and the Saturn, Sony's clout and marketing
temporarily made it "cool" in the industry to shell out software in less
than six months. Later PS1 games and all PS2 games were coded in
assembly again, but the shift lasted just long enough to make Saturn
developers have to spend more on average than dedicated PS1 developers.
--
Scott
Cafeman
originally had Saturn with a simpler chip design, mostly with 2D games
in mind. It was only after the shock of seeing what the Playstation
was going to do that Sega went back and quickly came up with the dual
SH-2 chip design which would give comparable 3D performance. They
didn't design an alpha channel into their 3D graphics architecture
however - this is why hardware 3D transparency wasn't as easy or cool
looking on Saturn as it was on PS1.
Still, they were going multi-chip from the start, so yes they were
still ahead of their time.
Scott H
> From what I've read from various sources over the years, Sega
> originally had Saturn with a simpler chip design, mostly with 2D games
> in mind. It was only after the shock of seeing what the Playstation
> was going to do that Sega went back and quickly came up with the dual
> SH-2 chip design which would give comparable 3D performance. They
> didn't design an alpha channel into their 3D graphics architecture
> however - this is why hardware 3D transparency wasn't as easy or cool
> looking on Saturn as it was on PS1.
As near as I can tell, this entire story is an industry rumor started by
Vic Ireland. There are no published specs for what the Saturn was going
to be before the PS1 specs were announced in mid 1993. There is a
statement floating around that the SOJ president at the time was pissed
that Sony's paper specs (the exaggerated 500k Texture mapped, 1 Million
flat shaded poly spec sheet) made the Saturn look worse on paper. Since
first year Saturn and PS1 games were in the 60-80k poly per second
range, this was only a marketing issue, not a spec issue. Even the 32X
has two SH-2s in it. I suspect that either the VDP1, the VDP2 or the
DSP was added after the PS1 specs were announced, but not more than one
of these.
>
> Still, they were going multi-chip from the start, so yes they were
> still ahead of their time.
>
--
Scott
BelPowerslave
> Vic Ireland.
Makes you wonder when he came up with that statement. I mean, if he came
up with it *after* Sega snubbed WD at the one E3 convention...
Bel
--
Whip Ass Gaming: http://www.whipassgaming.com/
"Live every week like it's Shark Week"
- Tracy Jordan, 30 Rock
Scott H
>> As near as I can tell, this entire story is an industry rumor started
>> by Vic Ireland.
>
> Makes you wonder when he came up with that statement. I mean, if he came
> up with it *after* Sega snubbed WD at the one E3 convention...
>
> Bel
The only sources I have seen that were negative towards the Saturn
before 1996 are all from EA press releases or Sony.
--
Scott
Cafeman
> Vic Ireland. >
I have never heard this before. It has been a few years but back in
the 90's and early 2000's I was more in tune with the SEGA stories and
information, so bear with me, things are vague. The only thing I
remember from Vic Ireland is how he got enraged (during the Bernie
Stolar Dreamcast ramp-up) when WD got a crappy booth at E3. That was
during the end of the Saturn era wasn't it? But I can remember
reading with fascination the story that I posted above concerning the
last-minute Saturn architecture changes that were caused by Sony's
super-powerful (then) Playstation. Yep, a Sega executive lambasted
his team for being beaten - I've read that story too. Not on the net,
but I remember from various articles in the printed mags -- mostly
Next Generation, also in Steven Kents book and interviews.
I seem to recall getting annoyed at Gamefan's pro-transparencies talk.
Sure, they also loved some Saturn titles but hardly a PSX review could
be written with the terms "glorious transparencies", it seemed. Like
the Transparencies were all that mattered. And of course there was the
debacle where one reviewer lowered Saturn's SF Alpha review because
the shadows weren't blue - he speculated it had something to do with
lack of transparencies!
Scott H
>> As near as I can tell, this entire story is an industry rumor started by
>> Vic Ireland. >
>
> I have never heard this before. It has been a few years but back in
> the 90's and early 2000's I was more in tune with the SEGA stories and
> information, so bear with me, things are vague. The only thing I
> remember from Vic Ireland is how he got enraged (during the Bernie
> Stolar Dreamcast ramp-up) when WD got a crappy booth at E3. That was
> during the end of the Saturn era wasn't it? But I can remember
> reading with fascination the story that I posted above concerning the
> last-minute Saturn architecture changes that were caused by Sony's
> super-powerful (then) Playstation. Yep, a Sega executive lambasted
> his team for being beaten - I've read that story too. Not on the net,
> but I remember from various articles in the printed mags -- mostly
> Next Generation, also in Steven Kents book and interviews.
Vic was very active in the gaming community up until about 2000. If you
watch closely, Steve Kent even uses direct quotes from Vic Ireland are
all around any comments about the Saturn. The only direct quotes I have
seen about the Saturn being deficient in any way come from Vic Ireland
and Bernie Stolar (then SOA President). I figure that most journalists
consider executives to be the best sources, because they're probably the
only sources they're going to bother with.
> I seem to recall getting annoyed at Gamefan's pro-transparencies talk.
> Sure, they also loved some Saturn titles but hardly a PSX review could
> be written with the terms "glorious transparencies", it seemed. Like
> the Transparencies were all that mattered. And of course there was the
> debacle where one reviewer lowered Saturn's SF Alpha review because
> the shadows weren't blue - he speculated it had something to do with
> lack of transparencies!
>
Yup, Nick Rox, right up there with Jeff Gerstmann.
--
Scott
Greg Gillis
> Cafeman wrote:
>
>> From what I've read from various sources over the years, Sega
>> originally had Saturn with a simpler chip design, mostly with 2D games
>> in mind. It was only after the shock of seeing what the Playstation
>> was going to do that Sega went back and quickly came up with the dual
>> SH-2 chip design which would give comparable 3D performance. They
>> didn't design an alpha channel into their 3D graphics architecture
>> however - this is why hardware 3D transparency wasn't as easy or cool
>> looking on Saturn as it was on PS1.
>
>
> As near as I can tell, this entire story is an industry rumor started by
> Vic Ireland. There are no published specs for what the Saturn was going
> to be before the PS1 specs were announced in mid 1993.
According to GameFan, as late as December of 1993 Sega still hadn't
fully committed to using two processors in Saturn:
"At this point the ... [Saturn] system is about 40% developed and Sega
hasn't decided between one 32-Bit CPU or two."
- Kei Kuboki, GameFan December 1993
It's hard to tell if they were thinking of adding a processor to their
current design (perhaps because of the PlayStation), or if they were
thinking of removing one (perhaps for cost reasons?).
There was also early talk of the Saturn potentially using a 4x CD-ROM
drive, so maybe Sega was just keeping the system specs as open-ended as
possible until the Saturn was closer to release.
> I suspect that either the VDP1, the VDP2 or the
> DSP was added after the PS1 specs were announced, but not more than one
> of these.
Next Generation hinted at a similar scenario in their first issue (late
1994):
"Sega has spent the last nine months or so playing catch-up with Sony
after a publisher-friend tipped Sega off about the power of PlayStation.
New specs and development tools only recently arrived with third
parties, superseding Sega's original description of the project. The
main difference between them is apparently the addition of more
dedicated processors taking work away from the two CPUs."
- Next Generation Premiere Issue
"These last minute improvements - while no doubt improving Saturn's
capabilities - may mean a problem with initial software support either
being thin on the ground, or not showing Saturn's true potential."
- Next Generation Premiere Issue
>> Still, they were going multi-chip from the start, so yes they were
>> still ahead of their time.
They were certainly considering a multi-chip system very early in the
Saturn's development, but whether or not that was the plan from the
start seems unclear...
--
Greg Gillis
Visit The Armadillo Games Home Page:
http://homepage.mac.com/greggillis/armadillogames
... featuring "Radiant Weirdness Zone", the "Fawn" RPG series, and
numerous armadillii...
Visit FORS YARD, A Chronological Retrospective of the Sega Genesis:
http://homepage.mac.com/greggillis
parallax-scroll
Sega was originally planning a 32-bit 'GigaDrive' console by late
1990 / early 1991, intended for release in the 1993 timeframe. It was
to be based on the System32 arcade board (which powered Rad Mobile,
Golden Axe Revenge of Death Adder and many others) a 2D sprite-based
board which was state-of-the-art for 2D visuals in 1991, blowing NEO-
GEO out of the water.
GigaDrive was to compete with
*NEC's Hudson-designed 32-bit Project Tetsujin / Ironman
(which began development in 1990, shown to the press in 1992, then
downgraded & changed into the PC-FX by 1994)
*Atari Panther, then Jaguar,
*Commodore Amiga CD32,
*3DO
*perhaps Nintendo's 32-bit incarnation of the SNES CD-ROM called
Nintendo Disc which was more powerful than the 16-bit Sony/Nintendo
Play Station version of the SNES-CD.
However GigaDrive (and later, even the Saturn) was certainly *not*
designed to compete with the powerfully 3D-capable PlayStation or
Nintendo 64.
GigaDrive evolved into Saturn with many changes during the course of
development. The addition of the 2nd SH-2 CPU and addition of a 2nd
VDP were some of the last upgrades in a long process of changes to
Sega's 32-bit console efforts.
Within a few weeks or months before the Saturn was released in Japan
in November 1994, some of SEGA Japan's management wanted to cancel
the Saturn altogether, and instead ask Martin-Marietta / Lockheed
Martin to design a fresh game console, or at least the 3D graphics
side of it, for release in 1996. However SEGA's strong ties to
Hitachi ruled that out, and Saturn went ahead.
IMHO Sega should have
*made the SegaCD more powerful for 1991/1992 with a stronger VDP to
support more sprites, colors, backgrounds. This was originally planned
IIRC but cut out of the final design. Yhe scaling & rotation ASIC
remained but it wasn't too impressive compared to even their mid-late
1980s Super-Scaler boards.
*not released a 32X at all, whatsoever (a more powerful SegaCD
would've taken care of 2D capabilities fine, and 32X was too weak to
be a good 3D system. Thus, the Sega MegaDrive/Genesis + MegaCD/
SegaCD would be the main Sega console formats until the mid 1990s.
reducing the amount of Genesis-based formats from 4 to 2. Instead of
(1) Genesis, (2) SegaCD, (3) 32X, (4)32X CD, there is only Genesis
and SegaCD. reduces consumer+developer confusion, mistrust, hate,
Sega's expenses, etc.
*scrapped GigaDrive / Saturn as some at Sega wanted.
*work with Lockheed on a 3D console to beat PS1, N64,
3Dfx Voodoo Graphics + other 3D accelerator PC cards, and edge out the
3DO M2 (assuming 3DO M2 launched).
A console with capabilities inbetween arcade MODEL2 and MODEL3 for
1996 (though not as powerful as Dreamcast) would've been entirely
possible at $299 given 1995-1996 semiconductor technology. This
console lasts for 5-6 years, and 4 years before the next-gen comes out
(with some overlap like PS1/PS2).
*Dreamcast never happens, at least not in the form that it did in
1998/1999.
*Sega works with Microsoft on a joint-platform (more or less Xbox1-
like) for year 2000 (not 2001) to counter PS2 and GameCube.
If Sega is still independent after that, Sega again works with
Microsoft on a new system for 2005-2006 with roughly the same if not
better capabilities than what became Xbox 360, perhaps with an
innovative control scheme to counter GameCube2 / Revolution/ Wii .
*wakes up from dream, back to reality :/
parallax-scroll
PS1 polygon paper specs, specs
transformed / calculated by GTE
(Geometry Transform Engine)
lines: 1.5 million
polygons: 500K
remember these are only the amount calculated, not pushed to the
screen.
polygons rendered/drawn to the screen by the graphics chip
(the important part)
360K flat shaded
180K texture mapped, gouraud shaded, with lighting
the 180K figure is the real, useful peak spec for PS1's polygon
capabilities. No game used more than that with texture-mapping, g-
shading and lighting. Early PS1 games like Ridge Racer were around
90K (about 1/3 of the Ridge Racer arcade machine).
Scott H
Your other posts are great, but do you have any sources for this one?
--
Scott
Scott H
Model 2 Specs
http://www.system16.com/hardware.php?id=713
Main CPU : Intel i960-KB @ 25 MHz 32bits RISC
Graphics Co-Processor : Fujitsu TGP MB86234 FPU 32bits 16M flops
Co-Processor Abilities : Floating decimal point operation function, Axis
rotation operation function, 3D matrix operation function
Sound CPU : 16bits 68000 @ 10Mhz
Sound chip : 2 x Custom 28 channel PCM chips, 1 for Music and 1 for
Effects (Can access up to 8meg sample rom *per chip*)
Sound Timing Chip : YM3834 @ 8MHz (only used for its timers)
Audio RAM : 540 Kilobytes (4 megabits)
Video resolution : 496x384 in 65536 colors
Geometry : 300,000 polygons/s. 900,000 vectors/s
Rendering : 1,200,000 pixels/s
Video : Shading Flat Shading, Perspective Texture, Micro Texture, Multi
Window, Diffuse Reflection, Specula Reflection.
It should be noted that the Hitachi SH-2 was apparently first used in
the Saturn and then the 32X.
Saturn Specs:
http://www.gamepilgrimage.com/Consolehistory.htm#Saturnandps1
CPU: Two 28.6 MHz, SH-2 32-bit RISC
CO-PROCESSORS: 20 Mhz SH-1
SCU (DMA and Control Processor)
Motorola 68EC000:
32 PCM Channels
8 FM Channels
Audio RAM: 512KB
VIDEO: VDP1, VDP2, DSP (geometry)
RAM: 2 MBytes
Video RAM: 1.5MBytes
RESOLUTION: Non-interlaced NTSC: 320X224, 320X240, 352X224, 352X240,
352X256, 640X224, 640X240, 640X256, 704X224, 704X240, 704X256
Interlaced: 320X448, 320X480, 352X448, 352X480, 640X448, 640X480,
704X448, 704X480
POLYGONS: 500,000 flat-shaded quads per second; 200,000, 8-bit (256
colors) texture-mapped (non-lighted)
CD RAM: 512 KB
Max Color Palette: 24-bit, or 16,777,216
Marketing performance: 50 MIPS (w/o DSP)
--
Scott
BelPowerslave
Dude, it's another one of Radeon's personalities...what do you think?
Bel
--
Whip Ass Gaming: http://www.whipassgaming.com/
"Ah no....not Toilet Duck again. You know what that does to you...you'll
be seeing the pink elephants again."
- Father Ted to Father Jack upon finding that he's drank yet another
bottle of Toilet Duck
Scott H
>> Your other posts are great, but do you have any sources for this one?
>
> Dude, it's another one of Radeon's personalities...what do you think?
>
> Bel
Why is it that trolls are the only ones that bother to make up pretend
conversations anymore?
--
Scott
BelPowerslave
> BelPowerslave wrote:
>>> Your other posts are great, but do you have any sources for this one?
>>
>> Dude, it's another one of Radeon's personalities...what do you think?
>>
>> Bel
>
> Why is it that trolls are the only ones that bother to make up pretend
> conversations anymore?
Well, you know how Radeon is, whenever he's feeling lonely and needs
some attention he crossposts shit all over the place.
Bel
--
Whip Ass Gaming: http://www.whipassgaming.com/
"I am Andrew Ryan, and I'm here to ask you a question. Is a man not
parallax-scroll
The stuff about GigaDrive & Saturn can be found all over the place.
A few sources:
http://www.eidolons-inn.net/tiki-index.php?page=SegaBase+Saturn
http://darkwatcher.home.att.net/console/saturn.htm
short history of Sega Dreamcast's birth:
http://archives.devshed.com/forums/hardware-136/short-history-of-sega-dreamcast-s-birth-1753417.html
It's also based on my memory from what I've read in other articles and
magazines.
I can't prove or source *every* single thing I've said, but overall,
it shouldn't be hard to understand Saturn (and 32X) evolved from
GigaDrive. The exact years are hard to pin down. Some articles
mention GigaDrive / Saturn development started in 1992.
Well, Saturn did, but GigaDrive was mentioned as far back as late very
early 1991 in EGM (thus with magazine lead times, that's late 1990)
and it wasn't just some EGM-only rumor.
The things where I said " Sega should have" is just my personal
opinion / ideas / wishlist, so there aren't sources for that :)
There are many things Sega did R&D wise that we will probably never
know.
Of particular interest to me is the time from 1994 to 1997 when Sega
had discussions with the Real3D division of Lockheed Martin, at least
twice, for a replacement / successor to Saturn, or even for an upgrade
for Saturn.
The success of a console platform starts with the hardware. It doesn't
end there, the games become the most important thing, but, without
solid hardware, there's no base for good software. The 32X and Saturn
were both hardware mistakes, in the end.
Scott H
>
> The stuff about GigaDrive & Saturn can be found all over the place.
> A few sources:
>
> http://www.eidolons-inn.net/tiki-index.php?page=SegaBase+Saturn
> http://darkwatcher.home.att.net/console/saturn.htm
> short history of Sega Dreamcast's birth:
> http://archives.devshed.com/forums/hardware-136/short-history-of-sega-dreamcast-s-birth-1753417.html
>
> It's also based on my memory from what I've read in other articles and
> magazines.
> I can't prove or source *every* single thing I've said, but overall,
> it shouldn't be hard to understand Saturn (and 32X) evolved from
> GigaDrive. The exact years are hard to pin down. Some articles
> mention GigaDrive / Saturn development started in 1992.
> Well, Saturn did, but GigaDrive was mentioned as far back as late very
> early 1991 in EGM (thus with magazine lead times, that's late 1990)
> and it wasn't just some EGM-only rumor.
I was specifically going after the claim that the Saturn was only going
to be a 2D system with modest 3D capabilities. How Sega, who had always
thrived off of its arcade ports, could go 2D only with the Model 1 and 2
out in the arcades is a question these theorists never bother to answer.
> The things where I said " Sega should have" is just my personal
> opinion / ideas / wishlist, so there aren't sources for that :)
> There are many things Sega did R&D wise that we will probably never
> know.
> Of particular interest to me is the time from 1994 to 1997 when Sega
> had discussions with the Real3D division of Lockheed Martin, at least
> twice, for a replacement / successor to Saturn, or even for an upgrade
> for Saturn.
> The success of a console platform starts with the hardware. It doesn't
> end there, the games become the most important thing, but, without
> solid hardware, there's no base for good software. The 32X and Saturn
> were both hardware mistakes, in the end.
Yeah, that would be the popular opinion which has no facts to back it
up. My opinion is stated on my webpage, but I'll summarize here. It is
fun and popular to make up reasons why someone or something failed and
ignore problems with someone or something that succeeded. The 32X and
Saturn were different approaches to different ideas, as was the PS1 and
the N64 and the Model 1 + 2 boards. One or two complaints about
hardware complexity does not make the 32X and Saturn a "bad design".
The games speak for the hardware more than pundits should be allowed to.
--
Scott
Scott H
> I thought popping these in here for comparison purposes would be helpful.
Then I saw some interesting similarities, but not with the Saturn
hardware.
> Model 2 Specs
> http://www.system16.com/hardware.php?id=713
> Main CPU : Intel i960-KB @ 25 MHz 32bits RISC
32X Specs:
http://www.gamepilgrimage.com/Consolehistory.htm#32x
CPU: Two 23 MHz, 32-bit SH-2 RISC
Optional overlay of Genesis graphics:
Model 2
> Graphics Co-Processor : Fujitsu TGP MB86234 FPU 32bits 16M flops
> Co-Processor Abilities : Floating decimal point operation function, Axis
> rotation operation function, 3D matrix operation function
32X
CO-PROCESSORS:
7.6 MHz 68000 (12Mhz with Sega CD)
3.58 MHz Z80 (Sound or Graphics)
TI 76489 (PSG)
> Sound CPU : 16bits 68000 @ 10Mhz
> Sound chip : 2 x Custom 28 channel PCM chips, 1 for Music and 1 for
> Effects (Can access up to 8meg sample rom *per chip*)
> Sound Timing Chip : YM3834 @ 8MHz (only used for its timers)
> Audio RAM : 540 Kilobytes (4 megabits)
32X:
Yamaha YM 2612 (FM):
10 Channels + 2 Digital channels (32X)
> Video resolution : 496x384 in 65536 colors
> Geometry : 300,000 polygons/s. 900,000 vectors/s
> Rendering : 1,200,000 pixels/s
> Video : Shading Flat Shading, Perspective Texture, Micro Texture, Multi
> Window, Diffuse Reflection, Specula Reflection.
32X:
RESOLUTION: 320x224
RAM: 512 KB plus Genesis or Sega CD RAM
POLYGONS:
25,000 texture mapped per second, 50,000 theoretical (box specs)
Max Colors On Screen: 32,768
Color Palette: 32,768
Marketing performance: 23 MIPS per SH-2
Polygon performance aside, it's interesting how similar in "complexity"
the 32X makes the Genesis to that of the wondrously designed Model 2
board. One would think they were designed by the same company.
Philipp Klaus Krause
> I've been thinking about this for a long while now, and I keep meaning
> to mention it but never seem to find the time.
>
> The Saturn's multi-processor layout: Was it too far ahead of its time?
> It seems like *every* PC nowadays has either dual processors or a
> processor with dual cores. The Saturn had two main processors, yet when
> we see a lot the software it's hard to believe that that's the best two
> processors, working together could do. I mean, sure, we got stuff like
> Rally, VF2, etc that obviously have all the processors doing their
> absolute best, but that shit was really rare.
>
> Back then, two processors was unheard of, even in a PC, so I'm sure the
> tools to make good use of them weren't quite up to par. Seemed like
> Sega, and a few others, were the only ones who really knew how to handle
> it.
>
> So, what do we think? Was Sega really thinking ahead when they placed
> multiple processors in the Saturn, or was it just a horrible, horrible
> design error that scared off all but the really good coders and the
> assholes(Probe, mostly) who just simply ported their software rather
> than recoding?
>
> Bel
Well, even though tools then were not what they are today I don't think
the two main processors are a big problem. Symmetric multi-processing
was well-researched then and now and Sega could have provided (maybe it
did, I don't know) a SDK that shows how to use them (a minimal
multitasking or multithreading OS would have done).
However looking at the rest of the hardware I see more problems. Just
look at this board: http://en.wikipedia.org/wiki/Image:SATURNBOARD.JPG
There's lots of custom chips, expensive to produce, difficult to
program. The problem is not the main processors it's the other stuff:
Two different VPDs, the extra SH-1 for the CD-ROM, the 68k meant for
sound processing, etc. This is mirrored by lots of different types of
memory, accessed in different ways.
Philipp
parallax-scroll
> parallax-scroll wrote:
>
> > The stuff about GigaDrive & Saturn can be found all over the place.
> > A few sources:
>
> >http://www.eidolons-inn.net/tiki-index.php?page=SegaBase+Saturn
> >http://darkwatcher.home.att.net/console/saturn.htm
> > short history of Sega Dreamcast's birth:
>
> > It's also based on my memory from what I've read in other articles and
> > magazines.
> > I can't prove or source *every* single thing I've said, but overall,
> > it shouldn't be hard to understand Saturn (and 32X) evolved from
> > GigaDrive. The exact years are hard to pin down. Some articles
> > mention GigaDrive / Saturn development started in 1992.
> > Well, Saturn did, but GigaDrive was mentioned as far back as late very
> > early 1991 in EGM (thus with magazine lead times, that's late 1990)
> > and it wasn't just some EGM-only rumor.
>
> I was specifically going after the claim that the Saturn was only going
> to be a 2D system with modest 3D capabilities. How Sega, who had always
> thrived off of its arcade ports, could go 2D only with the Model 1 and 2
> out in the arcades is a question these theorists never bother to answer.
>
Well (and people often overlook this) you really should understand
that back in late 1990, early 1991 when Sega started developing
GigaDrive, there WERE NO Model1 & Model2 3D-boards in arcades.
Model1 did not arrive until late 1992. In 1991 Sega had various low-
end to highend 16-bit Super-Scaler 2D boards and the new 32-bit System
32 board which was still only 2D. Model1 was not out until late
1992 in Virtua Racing, and it was concidered experimental, and while
somewhat impressive, not totally ground breaking given that Namco had
already had Polyonizer / System 21 out since the late 80s/early 90s in
various forms.
Sega's real first successful (and impressive) 3D effort was Model2, a
board completed in 1993, powering games by early 1994 (Daytona).
However by then, the Saturn had just about been completed, going
through its final last-minute upgrades -It was coming out in late 1994
in Japan. Too late to go with a true, fully-3D console unless Sega
delayed Saturn until 1995 for yet further upgrades that may or may
not have worked. -- Or scrapped Saturn altogether for a new 3D machine
which woudn't be possible until 1996 (unless Sega had started on a 3D
console earlier from the beginning, in 1992-1993 like Sony & Nintendo
did).
I think I said in my previous post, some of Sega Japan's management
wanted to cancel Saturn before the Japanese launch and have Lockheed
make an all new machine. The Nov 1995 Next Generation issue mentions
that:
http://img122.imageshack.us/img122/258/saturn2lmc2crop1033x13522rk.jpg
but That idea was rejected, sadly, and so were later (1995-1996)
proposals for Lockheed to design the GPU for Saturn 2. That's at
least twice Sega & Lockheed walk away from each other in partnerships
for the home market.
So the problem was, Saturn, and its forerunner, GigaDrive, were never
intended to be 3D from the ground up. While PS-X / PS1 / PlayStation
and Project Reality / Ultra 64 / N64, were.
> > The things where I said " Sega should have" is just my personal
> > opinion / ideas / wishlist, so there aren't sources for that :)
> > There are many things Sega did R&D wise that we will probably never
> > know.
> > Of particular interest to me is the time from 1994 to 1997 when Sega
> > had discussions with the Real3D division of Lockheed Martin, at least
> > twice, for a replacement / successor to Saturn, or even for an upgrade
> > for Saturn.
> > The success of a console platform starts with the hardware. It doesn't
> > end there, the games become the most important thing, but, without
> > solid hardware, there's no base for good software. The 32X and Saturn
> > were both hardware mistakes, in the end.
>
> Yeah, that would be the popular opinion which has no facts to back it
> up. My opinion is stated on my webpage, but I'll summarize here. It is
> fun and popular to make up reasons why someone or something failed and
> ignore problems with someone or something that succeeded. The 32X and
> Saturn were different approaches to different ideas, as was the PS1 and
> the N64 and the Model 1 + 2 boards. One or two complaints about
> hardware complexity does not make the 32X and Saturn a "bad design".
> The games speak for the hardware more than pundits should be allowed to.
>
> --
> Scott
>
> http://www.gamepilgrimage.com
32X and Saturn could've been used more for 2D games. They were both
very powerful for that purpose. They were lousy 3D machines, even in
the hands of the best coders like AM2, who pulled off some incredible
stuff with the hardware mistakes Sega made, as far as 3D.
BelPowerslave
> the two main processors are a big problem. Symmetric multi-processing
> was well-researched then and now and Sega could have provided (maybe it
> did, I don't know) a SDK that shows how to use them (a minimal
> multitasking or multithreading OS would have done).
This is what I think as well. I mean, from what I've read of how
Hyperthreading works, it seems like it's so very easy to implement. If
Sega could have just gotten proper tools to the developers, so they
could have made use of both processors, I think we'd of seen a much
higher quality of 3rd party titles.
> However looking at the rest of the hardware I see more problems. Just
> look at this board: http://en.wikipedia.org/wiki/Image:SATURNBOARD.JPG
> There's lots of custom chips, expensive to produce, difficult to
> program. The problem is not the main processors it's the other stuff:
> Two different VPDs, the extra SH-1 for the CD-ROM, the 68k meant for
> sound processing, etc. This is mirrored by lots of different types of
> memory, accessed in different ways.
Yeah, I think that the other processors were the issue, not so much the
two main processors. Still, when you think about stuff like SFIII that
supposedly even used the Sound Processor to push graphics, it boggles
the mind to think what the games would have looked like if Sega had dev.
kits that seamlessly let developers use them all properly.
Bel
--
Whip Ass Gaming: http://www.whipassgaming.com/
"Microsoft Xbox 360:
Features: Custom-made PowerPC CPU designed by IBM with 3 separate
processor cores nicknamed Curly, Larry, and Moe for their efficiency
and performance."
- Console Wars 2006, SomethingAwful
Benjamin
<snerp>
> Yeah, I think that the other processors were the issue, not so much the
> two main processors. Still, when you think about stuff like SFIII that
> supposedly even used the Sound Processor to push graphics, it boggles
> the mind to think what the games would have looked like if Sega had dev.
> kits that seamlessly let developers use them all properly.
>
> Bel
Well, part of that is the problem with being a first party developer
for your own company's hardware. Though the company wants its third
parties to be successful selling on its platform, the company's first
party developers are competing with those third party ones for a slice
of the software sales. I don't believe (though I could be wrong) that
Nintendo or Sega had truly assisted its third party developers in the
past beyond the bare minimum, so it'd be natural for a similar
relationship to continue on Saturn.
Benjamin