It's 64-bits. Do the math! So let's do some math. How many bits are these consoles?

true, but like with just about all console peripherals outside of input devices, it completely bombed. Jaguar was dead at that point anyway.

I don't know... most addons for consoles didn't bomb (I guess), because they were addons, but because there just were no real incentives to upgrade. I mean, look at the Mega CD or 32x. Most of the games released there were either REALLY bad (FMV games), slight upgrades of normal Mega Drive games or they simply came too late.

In a generation, where PC gamers need to upgrade their system regularly, an addon for a console could be succesful, as long as there are enough games supporting it.

But the Jaguars addon was just plainly too little too late for saving this stilborn console.
 
Arwin said:
Not to mention that, if I remember correctly, the 68000 was really, really good too. Was it about around the 68040 that it started falling behind the x86 offerings, or was it sooner? I can't remember ...

It was around the 68040. The 80486 was able to ramp in clock much higher than the 68040, and the super-scalar Pentium sealed the deal. The 68060 was able to actually outperform the Pentium in many cases, however it was much too late, and by then Motorola was more heavily committed to PowerPC and relegated 68K to the embedded space (where it has done exceptionally well).

jonabbey said:
It was around then, yeah. The 68000 series never went super-scalar like x86 did with the Pentium Pro.

Actually that's not true, as the 68060 was a super-scalar design. Also Intel technically took x86 super-scalar with the Pentium, not the Pentium Pro.

SimonF said:
Wasn't the saying at the time something like "The 68000 is a 32-bit processor masquerading as a 16-bit processor while the 2(?)86 is an 8-bit pretending to be 16-bit"?

So the expression goes...
 
swaaye said:
An interesting thing about N64 is that apparently even though it has a 64-bit CPU it ran mostly 32-bit code.
Wasn't that "only" 32bit code? Was there even an option to go 64bit?
On that note, from what I can see, the Series number no longer links directly to bitness with MIPS CPUs. PSP R4000 cores for instance are completely limited to 32bit.
Which made it irritating at times - eg. PS2 also used 32bit ABI (as do both 360 and PS3), but while that meant writting "32bit"-code from memory/pointer perspective, we still get to use all the 64/128bit arithmetic and wide registers. No such joy on PSP, making porting codebases more icky.

On that note, I don't think there has been any console in existance yet that Didn't use 32bit ABI(there was one idiotic attempt which was luckily shutdown by developer outrage while SDK was still in early stages), so for all the hopla about higher-bit CPUs since 95, none of them actually used it for more then pseudo-SIMD wider arithmetic and well, wider data bus to memory.
 
Last edited by a moderator:
Actually that's not true, as the 68060 was a super-scalar design. Also Intel technically took x86 super-scalar with the Pentium, not the Pentium Pro.

Right, I looked it up after my previous post. The Pentium Pro was the first out of order x86 processor, not the first super-scalar. None of the 68k series went out of order.
 
Actually, no, then went on to live as ColdFire (of which there's probably dozens in your house and car).
 
The PS2 was marketed as "128-bit", was it not actually it's ability to do 2 seperate 64-bit "threads" or something like that while the FPU actually could accept 128 bit code?
 
Mobius1aic said:
The PS2 was marketed as "128-bit"
Because the CPU actually had 128bit general purpose registers, that could actually do (some)128bit arithmetic, on a 128bit bus.
It's the only console that does this, so the 128bit monicker is as fitting as any, to point out the CPU was and still is unique today in consumer space.
 
Because the CPU actually had 128bit general purpose registers, that could actually do (some)128bit arithmetic, on a 128bit bus.
It's the only console that does this, so the 128bit monicker is as fitting as any, to point out the CPU was and still is unique today in consumer space.

128bit scalar, as opposed to 128/256/512 vector? What would the use of that be?
 
It's just bitwise logical ops and some load/store/shift ops (which are operand size/structure agnostic).
 
For some reason I remembered that it could do simple add too - but I just looked it up and it doesn't (not in single operation anyway).

jonabbey said:
128bit scalar, as opposed to 128/256/512 vector? What would the use of that be?
It was more a sideeffect then anything. They extended GP-register file to 128bit to support integer SIMD extensions(so vector stuff was also there), which also came handy working with VUs.
Of particular note, direct data transfer between register files is really nice(even more so if cpu has crippling memory access gotchas like 360/PS3).
And obviously, more register space is always helpful (too bad compilers don't do much with it though).
 
Back
Top