This might be the most complex hardware architecture of a home console ever.
The thing had two custom RISC chips which could be used as CPU (one with additional GPU capabilities and the other with DSP) plus a 68000 devs were not supposed to use.
You know what's also funny? I've never heard anybody complaining about the Saturn being buggy.
It's legendarily hard to code for, but seems like it was at least pretty solid.
Even the Genesis was sort of "buggy." I think there was one particular design choice that crippled digital sound playback. Also the shadow/highlight functionality is kind of weird, not sure if "buggy" is the right word, but weird.
There are lots of quirks with the hardware which point to Atari interfering with the development - the 68K was never supposed to be there (and an 020 would have uncrippled the bus by allowing it to run at full speed, see the arcade board), not using the dual RAM buses, having an object processor (Flare majored on DSP and Blitter, the object processor looks like a 5200/7800/Amiga/Panther throwback mandated by Atari) necessitated having a 2-chip solution where 1-chip would have been faster to develop and better.
Having said that, the Jag VR looked amazing for the time.
When asked if they were weary about developing on PS2, due to its reputation of it being tough to code for, Sega devs famously laughed and replied that they already mastered the Saturn - how much harder could it get?
But yeah, for a flagship title they should have went for a port with the official SDK.
The end result though was… demonstrating the Windows CE overhead.
The Cell was very much not that, the SPEs were unnecessarily difficult to use for a console and ultimately a dead end though they made for great supercomputing elements before GPGPU really took off (some people built supercomputers out of PS3s clusters as that was literally the cheapest way to get cells).
The Cell supposedly cost 400 millions to develop, a cost largely borne by Sony. MS got IBM to retune the PPE for Xenon, but they'd likely have made do with an other IBM core as their base had that not existed.
>I wish we had a button that was like ‘Turn On PS4 Mode’, but no,” Druckmann said. "We expected it to be hell, and it was hell. Just getting an image onscreen, even an inferior one with the shadows broken, lighting broken and with it crashing every 30 seconds...that took a long time. These engineers are some of the best in the industry' and they optimized the game so much for the PS3’s SPUs specifically. It was optimized on a binary level, but after shifting those things over, you have to go back to the high level, make sure the systems are intact, and optimize it again. "I can’t describe how difficult a task that is. And once it’s running well, you’re running the [versions] side by side to make sure you didn’t screw something up in the process, like physics being slightly off, which throws the game off, or lighting being shifted and all of a sudden it’s a drastically different look. That’s not improved any more; that’s different. We want to stay faithful while being better.”
The 7th gen appropriately "lasted" 7 years, "the early years" was a ton of it.
> Naughty Dog, a key contributor to the PS3 graphics libraries
ND is a first party studio, that means access to information and efforts to push the platform most developers can't justify, especially as third parties would usually need cross-platform support: the PS3 was no PS2.
> so much that porting PS3 games to the PS4 was "hell"
Which proves the point: the PS3's Cell was an expensive, unnecessarily complicated, and divergent dead end.
https://web.uri.edu/gravity/ps3/
But I guess the cluster overhead would reduce performance to less than a 4090 in real applications.
But from what I can find, the Cell also sucked at FP64, with a rate of 1/10 for a total of 15 GFLOPS (https://en.wikipedia.org/wiki/PlayStation_3_technical_specif... , second paragraph). The 400 PS3 cluster would be 6 TFLOPS or 5x RTX 4090.