Commodore was the worst thing that happend to the Amiga

[Hammer]

The core of 68K PC market was flawed i.e. Motorola was unable to keep pace with the other CPU vendors in the desktop market. Screwing (cut&paste) up the ISA standards are one of the main factors why the customers don’t stick with a particular platform. Motorola use to have PC vendors such as HP, DEC and SUN building 68K based workstations.  

Around 1992 time period, Motorola should have made 030/881 like 486DX like CPU to compete with Intel's bottom end 486DX i.e. making the solution cheaper with one chip packaging instead of two chip packaging. Secondly, one shouldn’t underestimate FPU’s contribution and influence on 1992 era 3D games.

[Hammer]

HopperJF wrote:
Commodore did great until around 1991 when Irving Gould got there and AAA was cancelled

The A1200 would have kicked ass as an 030 based 8Meg AAA machine for £399

This only delays the dying 68k core....

I rather see higher clocked EC020/881/effective chucky gfx/some fastram combination for early 3D textured games target.

[Hammer]

I think thats what the 68040 was supposed to be, a 68K equivalent of the 486DX, if only they had implemented the full 68882 into it's FPU rather than a subset.

Not in price during the late 1992/1993 hence my proposed integrated 020/881 with relatively high clock speeds. 040 was late and relatively expensive.

[Hammer]

The PowerPC was invented for Apple, they got 68K compatibility by doing it in software

It doesn’t make PC platform vendor’s job easy i.e. the OS porting issues is now on PC platform vendor not from the CPU platform vendor. Look at post-68K dead body count from such a migration move.

Secondly, Apple's 68k compatibility development remains only with Apple wouldn't be shared with the other PC platform vendors. This is a time waster due to 68k compatibility development duplications.

But if we move it up the chain (i.e. from Motorola’s POV), then Motorola’s 68k customers can benefit from easy mass core migration.

Thirdly, software emulators moves the front-end decoder from the CPU side to the memory side. Refer to Transmeta’s flawed compatibility solution.  

Intel also did a RISC design (the i860) but it never took off either.

This RISC core experience leads to "Pentium Pro"’s development i.e. Intel corrected their mistake**.
Secondly, i860's success wasn’t pivotal for Intel at that time i.e. it was basically an experiment. Intel is known to parallel their R&D i.e. refer to "only the paranoid survives" doctrine.

Btw, Intel i860 (N10) was used by many as a graphics accelerator.

No, they did their own in house RISC designs and the 68K couldn't keep up.

IF a particular RISC core suck then use another one.

As DEC's Alpha experience illustrates (e.g. "Why The Fastest Chip Didn't Win," BusinessWeek, April 28, 1997), it doesn’t needs to be the number 1 preforming CPU. It only needs to sufficiently increase the performance with legacy software being used for gluing it's customers to the existing software solutions.

Remember that the Amiga was never about high CPU speeds,

It’s a short term fix for boosting math performance for early 3D textured mapped games.  

Note that the 68882/68881 *is* a “co-processor” with specific job i.e. math related. 68882/68881 is not much different to Amiga’s custom co-processors since it takes over some of general processor’s jobs.

 everything intensive was done by the chip-set, had they of survived that strategy most likely would have continued.

Well, CBM haven’t solidified their 3D centric solution at that time.

[Hammer]

Doing it in hardware would of probably have proven as successful as Intel's attempt to do x86 compatibility in the Itanium

Remember, Intel botched the front-end of Pentium Pro (first member of P6 family) e.g. 200Mhz Pentium Pro runs like 133Mhz Pentium Classic in X86-16 software. Intel again botched it's X86 front end with Itanium (EPIC).

OOO features in Itanium was relatively crap compared to Pentium M/Pentium IV. Itanium’s focus was in-order execution while rest of OOO front end jobs was moved to software (in the memory domain).

AMD (J. Sanders III) was quoted to be saying that Intel’s Itanium was a good idea but poorly executed. Refer to AMD 29050, K5 and it’s similarity with the Itanium design. Note that, early X86 decoders with RISC cores actually preformed poorly compared to Pentium Classic. The X86 CPU vendors persisted with the design and fixed the problems.