[Info-vax] OpenVMS I64 V8.1 "Evaluation Release"?

[Johnny Billquist]

On 2012-03-22 19.33, Nomen Nescio wrote:
> Johnny Billquist<bqt at softjar.se>  wrote:
>
>>> I don't think that's true. What about z/Arch? What about POWER?
>>
>> z9 EC supports a maximum of 512 GB of physical memory. That is a long
>> way from 64 bits...
>> z9 BC supports a maximum of 64 GB. Even less than the EC.
>
> I am not talking about what they support, I'm talking about whether the CPU
> can address that much physical memory. I may be wrong though, after thinking
> about it and hearing the points about chip design where every pin is either
> used or not manufactured. I do know there is a 64 bit virtual address space
> on System Z and there is no per address space limit of 4G unlike many PC OS.

Correct. The virtual address space is 64 bits.
There are *not* 64 address pins out from the CPU.
The 4GB address space limit is also virtual address space, and that is 
32 bits.
Even though some PC CPUs actually have more than 32 address pins on the 
CPU, the address limit is still 32 bits for a process.

It is very important to keep the physical address space, and the virtual 
address space apart, which I've pointed out a number of times now. And 
just because your CPU do not have, nor can address anything close to 64 
bits of physical memory does not make 64 bit virtual addresses meaningless.

But your comment (at the top of this post) was a response to my claim 
that no actual CPU today can have anywhere near 2^64 bytes of physical 
memory. I might possibly have expressed myself sloppily, as it might be 
that the page address registers in the CPU design could conceivably form 
a 64 bit address, but there is no CPU today that actually have the pins 
required to use that. However, many MMUs (if not all) actually place 
more restrictions on the physical addresses as well, so that they don't 
actually form 64 bit physical addresses.

	Johnny