[Info-vax] Reimplementing VMS, was: Re: HP adds OpenVMS Mature Product Support beyond the end of Standard Support

[Stephen Hoffman]

On 2014-02-04 16:33:30 +0000, John Reagan said:

>> Whether GEM was modified for Poulson, I don't know.  AFAIK, there 
>> haven't been any new compilers released.
> 
> Not to my knowledge.
> ...
> However, Poulson actually added a handful of interesting instructions 
> that provide some forms of integer multiplication and integer division 
> without using floating point.  Since those are new, you'd would use the 
> /ARCH qualifier to enable those instructions and without any emulation, 
> such images would only run on Poulson. However, the question is moot.

Yeah, it was more whether there were any advantages that could be made 
available via compiler changes with any related code-generation work 
having been (possibly, hypothetically) started, particularly given that 
Poulson was marketed as the most significant shift of the architecture 
to date.   For those that haven't seen it: 
<http://www.realworldtech.com/poulson/>

As for introducing dependencies on processor generations or instruction 
subsets, John's well aware of how the compiler can generate code that 
can "guard"processor-specific code paths, as that mechanism was 
implemented with the GEM instruction scheduling used on Alpha — I'd be 
shocked if that same approach weren't possible within Itanium.   Just a 
SMOP, particularly when it's not me doing the programming.  :-)   I'd 
hesitate to consider some sort of trapping here if that's possible, 
based on the overhead of unaligned references — this trapping was 
implemented on VAX for VVIEF, and with the instruction extension 
emulation support for the byte-word, count extension and multimedia 
instruction extensions that is present in OpenVMS Alpha V7.1 and later.

The lack of Poulson support within VMS does (also) make this whole 
discussion moot, as well.  There's little point in updating the code 
generation.

-- 
Pure Personal Opinion | HoffmanLabs LLC