Fascinating Wiki Talk post about N64 chipset

[swaaye]

I was perusing the latest addition to the N64 talk page over at Wikipedia when I noticed it was a fascinating post about an interview with Tom Kalinske, former CEO of Sega. Apparently they also looked at the hardware that ended up in N64.

Link

 

[KOF]

This has been known for many years to any Sega fans. (thanks, SegaBase) but yeah, it wasn't until recently that Kalinske has provided more details about the fiasco.

 

[function]

Apparently they also looked at the hardware that ended up in N64.

What? Nintendo using dated technology? Never!

A big 'congratulations' to Sega of Japan for wisely going with the cost effective and developer friendly Saturn hardware instead.

 

[Fox5]

What? Nintendo using dated technology? Never!

A big 'congratulations' to Sega of Japan for wisely going with the cost effective and developer friendly Saturn hardware instead.

Yeah, the n64 chipset was turned down because the die size was too big and yields too poor? So you go and make a console with a dozen simpler chips that likely caused a system that was just as expensive if not more so than then the n64? Though I guess it's possible the n64 chipset saw some upgrades by the time it came out from when sega saw it, maybe it originally was going to use the same cpu as the playstation or a lower clock speed.

 

[swaaye]

What? Nintendo using dated technology? Never!

A big 'congratulations' to Sega of Japan for wisely going with the cost effective and developer friendly Saturn hardware instead.

Saturn was anything but easy to develop for from what I've read. The poorly implemented twin CPUs were a nightmare all by themselves. Yu Suzuki admitted this in an early interview regarding some game ports. They even quoted that Next Gen mag article in the Saturn article on Wikipedia. And this is why there are no decent Saturn emulators. N64 emulation is fairly decent on the other hand.

 

[Deepak]

I bet function was being sarcastic with his last sentence.

 

[swaaye]

Yeah I bet you're right lol. For some reason I was incapable of detecting sarcasm earlier today. Hmmm.

 

[Kutu-jiao]

The team that worked on the N64 weren't they called ArtX a spinoff from SGI?

It was because of this this company that ATi sprung up from being a normal graphics company into an agressive one.

 

[Reznor007]

Saturn was anything but easy to develop for from what I've read. The poorly implemented twin CPUs were a nightmare all by themselves. Yu Suzuki admitted this in an early interview regarding some game ports. They even quoted that Next Gen mag article in the Saturn article on Wikipedia. And this is why there are no decent Saturn emulators. N64 emulation is fairly decent on the other hand.

Actually almost all N64 emulation is a hack. The coprocessor is simulated, and not truly emulated. MAME has the most accurate emulation, and of course, is the slowest. MAME's Saturn support(there was an arcade version of Saturn called STV) is pretty decent. Most of the arcade games work pretty well, and many are full speed on a fast system.

 

[OtakingGX]

The team that worked on the N64 weren't they called ArtX a spinoff from SGI?

It was because of this this company that ATi sprung up from being a normal graphics company into an agressive one.

That was Gamecube. It was still SGI that Nintendo was working with on the N64. Then ArtX grew out of SGI, and was acquired by ATi around the release of the Gamecube. That team within ATi went on to design R300 (Radeon 9700 Pro) which was ATi's comeback chip, of sorts.

 

[LunchBox]

Quite an interesting read.
Thanks for posting!

 

[Megadrive1988]

if Sega had been smart, they would've talked GE Aerospace / Martin Marietta into making a graphics chip for Saturn. the Saturn should've been delayed until 1996. GE/MM could've had a chip that surpassed Model2 while offering a decent percentage of Model3 power (like 1/3 to 1/2). would've kicked the crap out of Nintendo 64's RCP and even the 3DO M2's Bulldog chip. a single CPU could've been used. like a PPC 603 from Hitachi. or even something less.

 

[OtakingGX]

if Sega had been smart, they would've talked GE Aerospace / Martin Marietta into making a graphics chip for Saturn. the Saturn should've been delayed until 1996. GE/MM could've had a chip that surpassed Model2 while offering a decent percentage of Model3 power (like 1/3 to 1/2). would've kicked the crap out of Nintendo 64's RCP and even the 3DO M2's Bulldog chip. a single CPU could've been used. like a PPC 603 from Hitachi. or even something less.

But government contractors are rarely cost sensitive. Saturn would have cost $10,000 with that proposed graphics chip inside.

 

[crystall]

A single CPU could've been used. like a PPC 603 from Hitachi. or even something less.

I never heard about Hitaching fabbing PowerPC CPUs, that said the 603 was a much bigger and more complex chip than the SH2 and was released much later.

 

[swaaye]

There was also the little-known PPC 602 used for M2. It was sorta like the R4300i I think, a "consumer oriented" crippled PPC. M2 used a pair.

 

[Squeak]

But government contractors are rarely cost sensitive. Saturn would have cost $10,000 with that proposed graphics chip inside.

They did make the rather excellent Intel740 aka Real3D Starfighter, which was rather GS'ish in it's simplicity and RAM setup, but with the addition of excellent MIP mapping.

 

[swaaye]

They did make the rather excellent Intel740 aka Real3D Starfighter, which was rather GS'ish in it's simplicity and RAM setup, but with the addition of excellent MIP mapping.

Ya except they built the thing to be totally reliant on icky AGP texturing. Surely Intel's forced direction there to drag AGP into the limelight.

 

[Squeak]

No one was ever forced to rely on AGP texturing, so we actually don't know how well it really would have worked if someone put their mind to it. If the textures are 4bit and MIP mapped, you need surprisingly few kilobytes per frame to get a good result (esp. back then.)