Anyone Using PatchPae2 For More Than 4GB with a 32 Bit OS?

Background on this Patch to allow more than 4GB in 32 bit Windows:
https://www.raymond.cc/blog/make-windows-7-and-vista-32-bit-x86-support-more-than-4gb-memory/

A very good and very sharp friend of mine was the North East Architecture Specialist for
Intel years ago and he told me probably more that 10 years ago that the claim from
Microsoft that you need a 64 bit OS to have more than 4GB of RAM was total BS. He
said if you want proof MS server versions of the 32 Bit OS's are just face lifts of the
mainstream OS and they allow for 64 GB of RAM. Here it is right from MS, X86 is the
32 bit version - Windows server 2008 is based on Win7 and 2003 on Vista, they both
allow for 64GB of RAM:
Memory Limits for Windows and Windows Server Releases (Windows)

It is the Processor Address Extension (PAE) feature that allows for more RAM:
Physical Address Extension - Wikipedia, the free encyclopedia

I prefer a 32 bit OS because these versions allow continued use of older programs and
I just think that they make more sense than 64 bit. My instinct,is that 64 bit has to be
less efficient and wasteful of memory bandwidth.
Anyway, there's been a patch out there to make a standard 32 Bit install of Vista,
Win 7 or 8 lift the 4GB memory limit, there is an older patch for XP.
Most people report that when the patch works systems with 4GB go from reporting:
4 GB installed (3.5 GB Usable)
to just:
4 GB installed
the lack of the limitation shows that it is in the mode with no marketing limit on RAM usage.
Others with 6 or 8 GB in their systems see the full amount after the patch.

I've tried installing the patch manually and with 2 scripts on my CF-51 MK3 with Win7, and I
never get the proper result, it always stays the same even with the patched kernel. I only
have 4 GB so the elimination of (3.5GB usable) is the indicator.
Anyone tried this - tips?

By the way, the patch often doesn't work with Nvidia video cards because their
drivers are not properly written. You have to work hard to make a driver
incompatible but unskilled people seem to be good at these things. I've seen
so many bugs in Nvidia software that I try to avoid their products anyway.

Here is a video demonstrating how to do it with a script:
Enable more RAM on 32-bit Windows 7 (PatchPae2, Batch) | ibrowsy videos | ibrowsy.com

I've tried the batch file linked below that video on two systems and it works as far
as installing the patch goes. It made no visible change on the CF-51 MK3. I also tried
it on a Dell Inspiron 1545 Win7 and it worked lifting the restriction on the 4GB installed
but then it crashed while watching a video. There are several reports of the Intel drivers
now causing problems also. Intel should know how to properly write drivers and I would
not doubt (for a second) that it is deliberate because they are in cahoots with MS or to
sell more new systems and 64 Bit processors. They say if you fall back to the XP driver
in Win7 and 8 it does work but you loose the functionality in the newer driver. It will
probably work for people running XP but you have to find an older version of the patch for
XP, it is on the same site I believe.

It would be kind of cool to have 8GB in a CF-51 with a 32 Bit OS if we can figure this
out. I have a CF-51 MK3LL coming soon that I think has XP installed - I'll try it on
that system and see if it works.

I wish someone would debug and patch the modern drivers - wishful thinking.

There are versions of 32 Bit Linux and the various flavors that also support more than 4GB.

 

No point in posting screen shots of the CF-51 MK3 with this patch since it did not seem to work.
It was with Win 7 32 bit, 4GB memory installed (3.5GB usable) in both cases.

Here are screen shots for the Dell Inspiron 1545 with and without the patch:



I don't have any 4GB memory modules to try 6 or 8GB.

 

Interesting

 

Seems the chipset used in the CF-51 MK3 only supports 2X2GB modules max. I've read
also that it does not support certain memory remapping capabilities but have not confirmed
this. It might explain why I'm unable to lift the 3.5 GB limit.
I usually run 3GB in this system and can't think of any situation where it is a limitation but
if I put in 4 GB it would be nice to have all of it available.
I'm going to research the possible memory mapping issue with this chipset.

 

It's not a secret.
This Wikipedia link is better
3 GB barrier - Wikipedia, the free encyclopedia

So you could have more problems with your hardware/software with PAE, than just using 64-bit OS.

Something for example? I'm on 64-bit since CF-19 MK2 and never had problems with software (except of drivers, but it's a problem even with newer 32-bit OS).

Very strange assumption.

Another one

All that looks like conspiracy theory )

P.S. You can use part of invisible memory for caching
http://www.romexsoftware.com/en-us/primo-cache/index.html
or ramdisk
http://www.jensscheffler.de/using-gavotte-ramdisk-in-windows-7
if you want.

 

Not a conspiracy theory it is marketing and greed.
I worked in CPU design and it would take pages and pages to explain here
what I'm talking about. I contributed to the design for a custom 64 bit super computer for the government
and also a small clone of the IBM mainframe 370 architecture. These are cases where I actually worked
on the design's architecture and detailed schematic level implementation.
I've also done board level designs for fault tolerant single board computers for the military.
My view is that if 32 Bit with PAE was supported it would be fine for 99.99% of the users out there,
very few actually need a 64 bit CPU. Granted the 32 bit processors have had 64 bit busses for many
years and this is a very old trick to get better performance from relatively slow memory and I/O busses.
The 16 Bit DEC 11/70 had 32 bit busses and a 32 bit ALU so that it could do double word operations
with a single pass through the ALU.

Example of a program that won't run on 64 bit is CALSOD, an old DOS program. It runs on
some win7 32 bit systems in a virtual 86 session or in DOSbox, I've not tried it yet on a 64 bit system.

Don't trust everything you read on Wikipedia, this link that you suggest has errors as I see it:
http://en.wikipedia.org/wiki/3_GB_barrier
The PCI section suggests that memory will necessarily be wasted in any 32 Bit system with a PCI bus.
My view is that for example if you map the PCI bus to a physical address say above 32GB then at least
the first 32GB of space can be used without any holes or issues. People ARE running 8GB 32 bit
systems, with PCI busses and with all of the RAM available. Servers with PCI buses provide for up to
64GB of contiguous memory.

 

99.99% of the server users, I think
Under the server environment it's normal to use professional, not amateur, software. In this case chances to encounter incompatibility are much lower. And for amateur OS and users I can't see any difference in price (if we talk about marketing) between 32/64-bit CPU and 32/64-bit OS. So why not to use 64-bit

You have to try
Look, I'm playing King's Bounty ) Win8 64-bit, DOSbox. No dances with tambourine.
http://1drv.ms/1pfsD5z

 

I'm not here to try and change YOUR thinking, that was not the intent of my statement
but if you want to believe that then go ahead. The incompatibility is in system level
software only. User level does not have access to features that cause incompatibility.
I have the 64 Bit version on a few systems and I'm not interested in trying anything else.
The marketing is to get people to pay for yet another copy of the OS, not talking about
the difference in price.
I prefer not to use DOSbox but that's good that it works for you for that application.

 

Ok, sorry, I didn't know that was a letter to UN, OSCE and software/driver vendors )

 

I researched this in depth back in 2009 and I do remember trying it, and just noticed that
this 32 bit Vista desktop that I'm typing on has 4GB with no restriction shown. I believe
that I used 32 Bit RamPatch shown here on this webpage:
https://www.raymond.cc/blog/make-windows-7-and-vista-32-bit-x86-support-more-than-4gb-memory/
I thought, that it just made registry changes but now looking again I think that it also patches
the kernel. It has been running fine with integrated Intel Graphics but I don't remember if I
had to do anything to the drivers to make it work - I don't think so.

Another good reference:
http://www.remkoweijnen.nl/blog/2009/06/23/patch-vistas-kernel-to-address-more-than-4-gb-of-memory/

 

On October 15 2014 Microsoft's Windows Update referenced here https://technet.microsoft.com/en-us/library/security/2949927 interfered with the PatchPae2 process in some way, so I uninstalled it. It took me all day to figure out what the problem was. All I had had to do was boot into the non-patched Windows 7 operating system (with the RAM limitation) but I didn't know that for all of yesterday (posting this on October 16th). Whew! I do wonder what risk I'm taking by having uninstalled that update, KB2949927. There is technical discussion on it at the link.