[Info-vax] [OT] Portable operating systems, was: Re: PowerX Roadmap -

[Johnny Billquist]

On 2016-09-20 21:27, Bob Butler wrote:
> On 2016-09-20, GreyCloud <mist at cumulus.com> wrote:
>> On 09/20/16 04:48, Johnny Billquist wrote:
>>> On 2016-09-19 21:51, GreyCloud wrote:
>>>> An interesting point you just brought up. AMD right now isn't exactly
>>>> getting rich these days selling their current processors. Assuming one
>>>> can, approach AMD with this particular question and see what is there.
>>>> If AMD did it correctly, this may give AMD some hope of increasing sales
>>>> in this particular arena.
>>>
>>> Nope. Totally pointless. People (comapnies) in general will not write
>>> any code or OS that would only work specifically on AMD processors. If
>>> AMD haven't already done the same as Intel, they will. Anything else
>>> just don't make sense.
>
> Intel couldn't get 64 bit done for their ablomination. AMD did that. So it's
> not really 100% correct to say nobody would code to AMD. Basically all the
> 64 bit code on Intel x86_64 is coded to AMD. I understand what you meant but
> I still think the argument is valid. If AMD supported enough features for
> another OS to run on it and not on the competition whilst not breaking their
> "Intel" support at the same time it could be worthwhile in terms of
> marketing and probably actual money. But designing and fabbing new chips is
> pretty costly.

The 64bit AMD extensions are not really comparable to what we're talking 
about here. And it only really became a success as Intel also decided to 
adopt it. Intel, at the time, was pushing for a different 64bit 
solution. So a 64bit solution was on the horizon, no matter what. And 
the people had to choose between Itanium, and the AMD x86-64. Forced to 
make a choice, the x86-64 was a choice of staying with what they had. It 
would be more fare to compare the suggestions here to the Itanium. Ie. 
let's make something incompatible...

So I would say it is in no way comparable to the suggestion that AMD 
should introduce an incompatible MMU to what Intel is pushing.

>>> And you really do not need 4 modes. I have said that for years around
>>> here. Seems like VSI understood, but a lot of people still seem to want
>>> to hang on to this like a religion.
>>>
>>
>> Not totally true on the modes.  Why did Data Generals machines have 8
>> modes?
>
> Must have been Intel-envy. If 3 or 4 are good 8 has to be at least twice as
> good right? Some people view complexity as a necessary evil. Healthy people
> in the engineering business view complexity as evil period. You can tell from
> Intel's abominations there were and are some sick puppies "designing" their
> chips. I hadn't heard that about Data General before but I'm sad I did. 8
> modes to run a crappy monitor program, a few serial lines and BASIC seems like
> overkill but maybe that's just me. I think they probably could have used one
> mode and not even needed all of that.

:-)

>> I know they went out of business, but why 8?
>
> So they wouldn't die of complexity first? I don't know. They were clunkers.
> I don't think they could compete with DEC at all in the mini arena and
> that's all there was for both companies. Probably DEC not needing all those
> 8 modes gave DEC enough design latitude to stomp DG into computer history.

The 32-bit Eclipse was complex. It did not improve it.

> A lot of what we have today in the Inteliverse is old, torn, moldy baggage
> that stinks and stinks and never goes away. I don't think any company has
> enough money and integrity and sense to straighten that out.

Right. Which is why we also will not see anything incompatible with 
what's on the market now. Sad but true.

	Johnny