[Kosh]

Way back in the day, when IBM was designing their first PC, why did they go with the Intel CPU instead of other stuff that was available at the time (like Zilog, Motorola, etc)?

[Fury]

Google knows. Click the very first search result for your answer.

[Kosh]

Ok, so what is your opinion?

[headdie]

it looks like the chip was chosen for flexibility i mean being able to access anything over a 1/2 meg of memory at the time was like having 8 gig now and i believe even then Intel was the market leader in certainly America

[Fury]

Ok, so what is your opinion?

I think the e-article explained it rather well. I have no opinion beyond that. I would be far more interested in how to break x86 (CISC) stranglehold of the CPU market that 8086 began, and have RISC CPUs enter desktop and notebook (netbooks don't count) markets.

[Flipside]

I had an 'argument' with one of my University lecturers about the feasibility of RISC chipsets in modern computers, he held that RISC was dead and the way forward was multi-tasking, I held that combining RISC with multi-tasking would produce a far better result.

[Fury]

What does CISC/RISC have anything to do with multi-tasking? Multi-tasking is not dependent on CPU instruction set.

[Spicious]

That context switching wastes heaps of cycles with x86s?

[Ghostavo]

CISC and RISC implementations have begun to blur for some years now.

CISC CPUs have used RISC introduced improvements and vice versa.

[Fury]

Architectural improvements perhaps. But it doesn't chance the fact that x86 instruction set is bloated beyond hope and they can't even drop legacy support or potentially break applications.

[Ghostavo]

And what current processor do you suggest as a replacement?

The so called RISC processors nowadays either aren't really RISC anymore or have some other annoyance that makes them not worthwhile for the desktop.

[Fury]

That is because there is no viable market for RISC on desktops. Even if they produced a desktop RISC CPU, they couldn't get any market share because OSes and applications typically only support x86. Almost all exceptions are in the linux and unix land where sources are freely available and can be recompiled. That's why there's no room for anything but x86.

[Ghostavo]

I just found an interesting article.

The irony of it was that I found it while reading a totally unrelated TVTropes page.

[Admiral LSD]

IBM chose the 8086/8088 largely because it was cheap, but also because they had a licence from Intel to build them in-house. The Motorola 68000 would have been a better choice, but IBM were building the PC down to a price. The 68K was an absolutely brilliant piece of forward thinking. Even though it was only packaged as a 16-bit part, the core itself was essentially 32-bit. When properly implemented, programs written for a 68000 could be transferred to the 68020 (which was 32-bit inside and out) and be 32-bit without modification. And this was in 1979, Intel wouldn't release its 32-bit 80386 for another 6 years and it would be at least another 10 after that before 32-bit computing entered the mainstream (Yes, I'm aware that there was Linux, OS/2 and NeXT, among others, offering 32-bit on x86 years before that, but it wasn't really until Windows 95 that 32-bit was thrust into the public consciousness). The problem with 68000 though, was virtually everyone took it at face value and implemented it as a 16-bit processor and ended up having all kinds of migration difficulties when 68020. Even Apple, who arguably had the most successful line of 68K-based machines out of everyone who used it, grappled with 32 bit cleanliness issues for years after they released the first 68020-based Macintosh IIs.