Great Moments in Microprocessor History

An anonymous reader writes "The microprocessor changed the world: how did we get from the first 4-bit models in the 1970s to today's 64-bit multicore monsters? This article covers the history of the micro from the vacuum tube to today's dual-core multithreaded madnes."

Comments

Re:lightning rods

[kryogen1x]

Calculus was invented independently by two different people in two different places. That would have been better.

Waiting for the next great leap

[Junks Jerzey]

The jump from the 6502 to the 68000 (a scant four years apart), was a huge one. Ditto for many of the x86 generations. But performance has leveled off quite obviously in the last few years. The difference between a 3GHz P4 and a 3.6GHz P4 is fairly small, as both tend to be memory bound for real-world applications. And at the same time the power consumption for the 3.6GHz has increased more than the performance.

So what's going to be the next big leap for desktops and notebooks? 64-bit processors are here, yes, but all else remaining the same these run *slower* than 32-bit processors, because the cache effects of 64-bit pointers more than offsets the ability to do 64-bit integer math (note that the x86 FPU has been 80-bit since its inception). Dual core is nice...but it's only a win for multithreaded applications or when you're running multiple applications at the same time. Even then, the effect of multiple threads sharing a cache can result in lower performance than many people expect.

Surely someone is going to set the PC world on its ear with a massive performance leap that doesn't require 1000 watt power supplies?

Re:Waiting for the next great leap

[ChatHuant]

The jump from the 6502 to the 68000 (a scant four years apart), was a huge one.

Let's not forget the wonderful Motorola 6809 (introduced in 1977, two years before the 68000). At the time I was doing some work in PDP-11 assembly, and switching to 8080 assembly language was a nightmare of special use registers and un-orthogonal statements. Even the Z80 (though much better), suffered because of the need for compatibility with the horrible 8080. The 6809 was beautifully clean in comparison.

Re:Waiting for the next great leap

[Angstroem]

The jump from the 6502 to the 68000 (a scant four years apart), was a huge one. Ditto for many of the x86 generations. But performance has leveled off quite obviously in the last few years. The difference between a 3GHz P4 and a 3.6GHz P4 is fairly small, as both tend to be memory bound for real-world applications. And at the same time the power consumption for the 3.6GHz has increased more than the performance.

Amen, brother.

That's what I'm seeing for quite some time especially on the PC market. ZX81 to C64 was a quantum leap, C64 to Amiga was another (this time even in processing power). And then the "PC revolution" when Motorola stopped further development of the 68k line and left the field for 486 and Pentium.

Personally, I'd say the last "felt" revolution was the DEC Alpha and maybe the introduction of DSP-like instruction set enhancements like MMX, 3dnow!, Altivec and the likes.

Nowadays, "revolution" takes place on the GPU side, not so much on the CPU which suffers from the effects you're mentioning. But also on GPU level we're slowly approaching an end, i.e. will be hitting the memory barrier (not to mention heat and power consumption).

What's coming next? My guess is that the next *big* hits come from neural networking and quantum computing, but not from traditional general purpose computing... That branch has pretty much come to an evolutionary end -- also on physical level because with current clock speeds the travel speed of electrons indeed does matter (ask the Pentium-4 designers...)

Instead, you will see a constant rise of specialized systems (system-on-chips, SoCs) and partly reconfigurable systems. Going away from overdesigned general purpose machines towards application-specific, efficient designs. Reconfiguring, self-aware systems which can either be manually adapted to given tasks or will self-optimize towards a (set of) given application(s).

In other words: the PC revolution is over and you will not see a massive performance leap there. Not unless someone finds a cheap way to overcome the memory bottleneck -- and even then the performance leap would not be *that* dramatic given the current results for cache hit ratio, and branch (and other) prediction accuracy.

Re:Waiting for the next great leap

[Entropius]

It depends.

P4's are power hogs. They (AFAIK) have no processor power management capability, and will thus chug along at 3.4GHz even when you're playing Zork. I'd imagine that they use at least 100 W (for just the processor). The folks at http://www.techreport.com/onearticle.x/7417 did a test: an idling P4 system without monitor uses 150 W, and under load the complete system sucked down 230.

Another website gives the power use of just the processor (P4 EE 3.4 GHz) under full load at right under 200 W.

Athlon 64's, even the desktop models, use basically the same technology that's been in laptop processors for a while. (Note that this WAS NOT turned on in the Ath64 tests in the link above.) They can underclock themselves (down to 800 MHz, typically), lower the core voltage, and use other tricks to decrease power consumption when under light load. This dramatically decreases power consumption without hurting performance. The sources I've seen cite desktop Ath64 power use at something near 70/100 W (idle/loaded), but the idle numbers seem too high: my laptop Ath64 uses around 20 W for the WHOLE SYSTEM (idling), and from what I hear the desktop A64's are basically the same as the mobiles, so the desktop version should be able to idle at under 70 W.

Desktop video cards (for gaming) can use a hundred watts or more.

So, in short, a desktop machine can use 100 W or less (Athlon 64 at idle), to as high as maybe 400 W (P4 system playing Doom 3). All this is without a monitor: my guess is that a 15" LCD uses no more than 15-20W, and a CRT may use 100W or more.

Re:Waiting for the next great leap

[evilviper]

64-bit processors are here, yes, but all else remaining the same these run *slower* than 32-bit processors

But all else HASN'T remained the same. The AMD64s have more registers, built-in memory controller, and plenty of other improvements that make it significantly faster than 32-bit x86 processors. In addition, Cool-n-Quiet (really more about the motherboard than processor) which reduces heat, something you seem superfically concerned with.

The difference between a 3GHz P4 and a 3.6GHz P4 is fairly small, as both tend to be memory bound for real-world applications.

Guess what, the 64-bit processor you are discounting happens to solve this issue you are complaining about with 32-bit chips...

Re:Performance

[Billly Gates]

Was your dataset memory limited?

I know the G5's are probably at least 15-30 times faster than the 68040's in the Quadra's but if it took 3 days on such a beast it would still take a few minutes to perhaps an hour on a G5. This is pure cpu time.

The rest may be because your huge data now sits in the ram vs sitting in the hard drive.

Not Enough Research

[fyngyrz]

This article gave extremely short shrift to one of the, if not the, most powerful 8-bit micros ever made; the motorola 6809. This was a revolutionary -- not evolutionary -- micro. Multiple stacks, highly orthogonal, broad and powerful indexing capabilities... while it had roots in the 6800, it no more resembled the 6800 than does a Lamborghini Countach resemble a Volkswagon Golf.

Re:886

[runderwo]

The 286 added protected mode, though not as nice as 386+. You could access up to 16MB of memory in 80286 Protected Mode. oh, baby!

It was worse than that; the 286 provided no mechanism to switch out of protected mode. Windows-286 was a dog because it depended on so many DOS interrupt routines which had to be executed in real mode (since the drivers had not been written as native VXDs yet). The only reason this was possible at all was because a facility was provided to reset the CPU through the keyboard controller. Every time the software needed to switch the CPU into real mode, it had to be completely reset through the keyboard controller. You can read all about the A20 mechanism here.

The 386 finally added the ability for the control program to switch back to real mode. With that it carried a bug that allowed the user to set a segment limit of 4GB while in protected mode and then quickly switch back to real mode, giving the user access to a 4GB address space in real mode (where only a 1MB address space should be available). Many games and demos circa 1992-1993 exploited this "Unreal-mode" feature like Ultima 7 and Zone 66, and were known to be the nastiest, most incompatible programs ever to exist, never getting along with any memory manager or multitasking operating system. If only game programmers had used something sane like DPMI back then!

IIRC the 486 added something related to cache, I forget if it was on-chip cache

Yes, it added an 8KB on-chip instruction cache - to the dismay of many legacy programs which used precisely timed tight loops for program timing.