Aaron Spink (aaronspink.delete@this.notearthlink.net) on August 11, 2014 1:02 am wrote:
[snip]
> is ARM64 nicer to work with than x86? Sure, but at the end of the day, that
> doesn't mean a whole lot. The original Itanium ISA was basically Intel Alpha
> and would of been a great ISA, then they had to get HP involved.

How did you come to know what IA-64 was going to be before HP got involved? Were you part of the Digital to Intel personnel transfer (I forgot when that happened)? Just moved your shingle by yourself? Know people who were able to reveal such details?

I may very well be wrong, but I did not think that there was much public disclosure of what IA-64 was going to be (I also assume Itanium was a branding choice that came after HP became involved).

> ISA means much less than many assume it to mean.

I wonder if this may be a bike shedding effect. It doesn't take much computer architecture knowledge to understand the principles of ISA design, so many more people are likely to have an opinion about what color to paint the bike shed the ISA should look like. If something is comprehensible enough that one can have a reasonable opinion, then (by understandable albeit flawed thinking) it must be important. (Not unlike choosing the color to paint a bike shed, there are complex considerations that may not be recognized even when multiple people present reasonable cases for different choices based on different reasons. A diversity of opinion may even increase one's confidence in being right, perhaps "reasoning" that if most people are wrong and one naturally feels one is more likely to be right then one is actually almost certainly right [e.g., with 10 opinions, the chance of one being right is 10%, one feels one is at least 50% likely to be right, so one must actually be 90% likely to be right].)

Even thinking about higher level microarchitecture does not seem to be especially inaccessible compared to actually doing hardware design. While the very uninformed may make microarchitectural suggestions that would make readers here smirk, even a more informed outsider may be puzzled by design decisions made by professionals, partially because even the context of (much less the reasoning for) the decisions are not public. (The L2 cache design for Bulldozer puzzles me, but I know that I am extremely ignorant about layout constraints (and about design reuse issues and design schedule issues and ...) and that the designers were not stupid. If they were idiots, the apparent [obvious and seeming] mistake would be more understandable.)