Windows 7 32bit Can See All 4 GB RAM

[wondim]

There are enough topics under this same topic but sadly all are closed, so I needed to create another topic to talk about what I got. Windows 7 or Vista can support memory above 3 GB.

Here is the article that explains the topic in great detail.

http://www.geoffchappell.com/viewer.htm?doc=notes/windows/license/memory.htm

Feedback on the article result can be found here and the author himself has commented on the discussion.

http://www.raymond.cc/blog/archives/2009/08/19/make-windows-7-and-vista-32-bit-x86-support-more-than-4gb-memory/

And my test result also shows a good result. I am now seeing all my 4GB RAM with my 32bit Windows 7 professional.

Here is the proof.

Please, it would be great if this topic is left open for further discussion.

[nitroshift]

Windows Vista SP2 and Windows 7 32 bit can see all 4GB of RAM out of the box, the problem is that due to the 32-bit architecture they can NOT use that amount. The method used to calculate the amount of RAM that a specific architecture can use is 2³² for 32-bit operating systems and 2⁶⁴ for 64-bit operating systems. Further reading here : http://www.msfn.org/board/32bit-windows-no...am-t130000.html.

[dencorso]

But do also read: [32Bit or x86] Use more memory above 3.25GB, Its possible now!. HTH.

[wondim]

Windows Vista SP2 and Windows 7 32 bit can see all 4GB of RAM out of the box, the problem is that due to the 32-bit architecture they can NOT use that amount. The method used to calculate the amount of RAM that a specific architecture can use is 2³² for 32-bit operating systems and 2⁶⁴ for 64-bit operating systems. Further reading here : http://www.msfn.org/board/32bit-windows-no...am-t130000.html.

I think the best thing to do is to test the patch and check the if your ram can read anything above 3.2 GB RAM. And I did as you saw using Vmware and the result shows just the opposite of what Microsoft is saying. 32bit applications can use RAM above 3.2 GB. See my screen shot and read the article that I shared on the first post for detail.

The link that you shared is a pinned one but closed. I have seen it before I posted this one but I got an opposite result of what it said.

But do also read: [32Bit or x86] Use more memory above 3.25GB, Its possible now!. HTH.

I searched for similar on the search page but all were locked but I should have taken more time to search. Here is one interesting topic where geoffchappell posted and got banned.

http://www.msfn.org/board/licensed-memory-...ta-t133516.html

Any way, I am not as expert as anyone here, but just test it as I did using VMware or Virtual Box or another memory hungry software.

Edited December 30, 2009 by wondim

[nitroshift]

Geoffchapell was banned for a reason. And because similar topics in the past have stirred quite a few spirits, I am going to close this one too before it gets too late.

[cluberti]

You *can* use the kernel to address memory over the 4GB boundary (2^32) from an x86 OS, but you still have the limitation that the CPU (when in x86 mode) can only ever actually address 4GB of address space at any one time, no matter where you set your base address. Ultimately, you can get a 32bit Windows OS to do a lot of things with the physical RAM over 2^32, but it's really a lot of smoke and mirrors (good smoke and mirrors, mind you, but it's still not native) and you're still never actually using a RAM window larger than 4GB (the CPU has to move that window up and down, over and over, to make it work). Geoff is a smart guy, and his test will work good enough for most desktop applications, I would think. However, it's worth mentioning that it's not supported at all by the vendor (Microsoft), you have to put your system into test signing mode to get it to work (bypassing kernel driver signing security, amongst other things), and you're still only ever actually addressing 4GB of memory address space (in a sliding window, of course) at any one time. If you want to do all that just to avoid installing an x64 OS, you're free to do so. However, given that we've had 2 versions (3, if you count XP x64) of Windows client that can run and natively do under x64 what the test x86 kernel is doing and taking a perf hit to do so (small, but it's there), there's no real reason anymore to run x86 and perform a bit of hackery like this unless you *absolutely, unequivocally have something that needs to run on x86 and won't run on x64*, and that's gonna be pretty rare.

It is worth closing though, just due to the nature of the flame fest that always ensues - closing the thread. If you want to discuss it with Geoff, he has a forum and is more than willing to chat with you about it in technical detail, but it's been beaten to death and back again here.