farewell to PowerPC - hello to performance and cool running
I still wax nostalgic over the Motorola MC68000 of the original Macs. To me it is the best CISC CPU ever devised and a lot of my identification with Apple Macintosh was bonded on its choice of that CPU (which I delved into assembly programming quite frequently).
My first professional job was programming the 68000 Mac where I poped its case and slapped a logic analyzer on the 68000 CPU to assist debugging by trapping on address accesses. Great times indeed!
Then some years later I recall the transition to PowerPC when working at Aldus and getting PageMaker to run on Apple's choice of a new RISC CPU. The instruction set was new and interesting - and code ran faster.
One thing, given the segmented addressing scheme ancestry of x86 and its paucity of registers, pretty much any chip could beat x86 in terms of superior instruction set that a geek would tend to prefer. Any geeks that "enjoyed" x86 at the instruction set level were the masochist - and likely worked at Microsoft. (Though I still managed to write some very worthwhile assembly code on x86 that boosted application performance.)
Yet here we are now with the new Mac Pros on Woodcrest Xeon x86 CPUs. The difference in architecture of this new Mac vs its PowerPC predecessor says a great deal about how far behind the curve PowerPC has fallen. There's simply no contest by any criteria of measurement. Is rather sad that PowerPC innovation - in the areas that count for modern desktop and laptop design - is now out of the race. Even the XBox 360 has a fundamental heating issue due to its PowerPC use (otherwise the XBox 360 is a very lovable piece of hardware from a geek perspective).
At the end of the day I prefer my Macs to be able to stay on the cutting edge of performance and power consumption reduction. It's been years since I've bothered to piddle with programming a CPU at the instruction set level (three, actually). The ugliness and inelegance of the x86 instruction set? I'll just deal with it the same way I do my morning breakfast sausage - enjoy the taste but don't ever bother to visit a sausage factory.
My first professional job was programming the 68000 Mac where I poped its case and slapped a logic analyzer on the 68000 CPU to assist debugging by trapping on address accesses. Great times indeed!
Then some years later I recall the transition to PowerPC when working at Aldus and getting PageMaker to run on Apple's choice of a new RISC CPU. The instruction set was new and interesting - and code ran faster.
One thing, given the segmented addressing scheme ancestry of x86 and its paucity of registers, pretty much any chip could beat x86 in terms of superior instruction set that a geek would tend to prefer. Any geeks that "enjoyed" x86 at the instruction set level were the masochist - and likely worked at Microsoft. (Though I still managed to write some very worthwhile assembly code on x86 that boosted application performance.)
Yet here we are now with the new Mac Pros on Woodcrest Xeon x86 CPUs. The difference in architecture of this new Mac vs its PowerPC predecessor says a great deal about how far behind the curve PowerPC has fallen. There's simply no contest by any criteria of measurement. Is rather sad that PowerPC innovation - in the areas that count for modern desktop and laptop design - is now out of the race. Even the XBox 360 has a fundamental heating issue due to its PowerPC use (otherwise the XBox 360 is a very lovable piece of hardware from a geek perspective).
At the end of the day I prefer my Macs to be able to stay on the cutting edge of performance and power consumption reduction. It's been years since I've bothered to piddle with programming a CPU at the instruction set level (three, actually). The ugliness and inelegance of the x86 instruction set? I'll just deal with it the same way I do my morning breakfast sausage - enjoy the taste but don't ever bother to visit a sausage factory.
posted by rogerv at
7:31 AM
0 Comments:
Post a Comment
<< Home