[Info-vax] Three boot camp sessions on YouTube

[Johnny Billquist]

On 2016-10-12 15:39, Bob Koehler wrote:
> In article <e5edf4ce-52e0-4622-a2ad-dd16a71f37ac at googlegroups.com>, Neil Rieck <n.rieck at sympatico.ca> writes:
>>
>> For me, the big difference between Alpha and Itanium in those videos was th=
>> e register count (32 vs. 128).  Having 128 registers may have seemed like a=
>>  good idea when they were designing that chip, but we all know that an OS i=
>> s going to require some/all those registers to be periodically saved then r=
>> estored (well, not all on every interrupt). I wonder if Intel would have im=
>> plemented 128 had they known that dynamic memory technology would keep impr=
>> oving as it has.=20
>
>    It's been quite a few years since some chips started employing register
>    sets, where swapping context doesn't mean dumping registers to RAM, just
>    switching sets inside the CPU.  Don't know if any of the Intel
>    offerings do this, but if not, then it's probably onlhy a matter of
>    time.

The problem with this is that this is a limited resource normally much 
smaller than the need, which means that it don't scale.

The concept existed already in the stone age (well, sortof). For 
example, the PDP-11 have two identical register sets, and which one is 
used is decided by a bit in the PSW. Of course, since in most generic 
OSes, you have more than two processes, you cannot really use this in 
any useful way in the OS.

And the same goes if your CPU have 16 or 256 register sets. You just 
cannot really use this, since you might very well have more than 256 
processes you switch between, so you still need to save the context out 
somewhere else.

	Johnny

-- 
Johnny Billquist                  || "I'm on a bus
                                   ||  on a psychedelic trip
email: bqt at softjar.se             ||  Reading murder books
pdp is alive!                     ||  tryin' to stay hip" - B. Idol