Well thought I'd share my results as a data point:

Got a Lenovo (aka IBM) T61 laptop from circa late 2007.  It's the Santa Rosa 
mobile platform, specifically an Intel Core 2 Duo Merom 2nd gen processor 
(T7500 @ 2.2Ghz).  I don't know what the exact chipset is (anyone know an easy 
way to find that with an app...?) but in any case the chipset it what usually 
came with that chip.


So I installed Win7 Ent 64-bit (and remember I only have 4GB TOTAL memory 
installed) and on the System screen, the machine went from saying 3.2GB usable 
out of the 4GB installed (which was with Win7 32-bit), to just saying a plain 
4GB RAM installed with nothing missing...

 

So it looks like even though I only have 4GB physical memory, the MMIO devices 
don't need to reserve any of my 4GB of memory b/c they can map above it (I 
assume up to 8GB in my case).  So based on the KB article, I assume if I 
upgraded to 8GB of RAM, it would show 7.2GB of RAM b/c then the devices would 
still need to map and since 8GB is their max, it would take away from my 
physical memory...I'll see if I can get 8GB of RAM for it and test this out! ;)

 

BINO


 
> Date: Sat, 1 May 2010 15:51:37 -0700
> From: maccr...@gmail.com
> To: hardware@hardwaregroup.com
> Subject: Re: [H] Win7 Ent 32-bit vs 64-bit?
> 
> Differing amounts of "memory hole" from differing configurations would be my 
> guess 
> before blaming a bug.
> 
> Agreed, with driver support (on new devices at least) finally happening there 
> is no 
> reason to use x32.
> 
> On 5/1/2010 3:16 PM, Bryan Seitz wrote:
> > Seems to be a bug or a chipset thing if it's different on different 
> > systems, all with 4G of ram.
> > Either way there's no reason to not use a 64 bit os in 2010.
> >
> > On Sat, May 01, 2010 at 10:43:14AM -0700, maccrawj wrote:
> >> Sorry, your point/counterpoint is? Think I'm missing something here.
> >>
> >>
> >> On 5/1/2010 7:28 AM, Bryan Seitz wrote:
> >>> Well no, I've seen systems with 4G of memory show:
> >>>
> >>> 2.5G
> >>> 2.8G
> >>> 3.5G
> >>>
> >>> with /PAE
> >>>
> >>> :)
> <snip>
                                          

Reply via email to