>>>>I asked that 3/4 gig question a while ago. The answer was that Vista reports 3 gigs but uses more
>>>
>>>IOW, it gives you 3 to use and keeps the rest to itself?
>>>
>>>Does anyone remember UMB, EMS, XMS? Sounds so familiar :).
>>
>>Yes, I remember those. And trying to keep the difference between extended memory and expanded memory straight. And the 1 meg memory barrier being broken (with workarounds like the above). Those were the days!
>
>Well they are back. The cheaper hardware eventually gets to hurt the software. Just like it was cheaper back then to have only 20 bit wide address bus, and to have devices map into memory space instead of having a space (read: circuitry) of their own, so it is today. If your machine can't use the 4th meg, why is it so? It's already in use, at least some of it, by all sorts of untouchables - your video card, your window opener, your can opener, your can't opener and whatnot - all those memory-mapped devices, again. Plus that the OS has this VAXlike habit of eating the memory space from address ffffffff downwards, while userspace starts from 00000000 and upwards, with the first 64K taken for interrupts and whatnots - again memory specific hardware, not all memory addresses are equal.
>
>It seems the history of our motherboards is thickly seeded with such cut-ear-to-patch-butt measures which gave someone a two month market advantage at the time, while hurting everybody in the long run...

As long as you are 32 Bit the memroy has to be mapped somewhere - so why not fronm the top ?
Introducing paged memory (in the sense of PAE, not the 4KB slices between RAM and swap disk)
while working under 4GB would have introduced an unnecessary slowdown for a dozen years (since NT 3.5).

PAE might have been made easier - I think most of the benefits of 64 Bit could have been reached
with a nice PAE implementation, but sodtware needs sometimes bogus arguments to keep the bucks rolling.

regards

thomas