6G of ram on 32bit OS? reserve 2G for built-in graphics, other 4G for OS?

giantbucket

Dignified
BANNED
ok, here's a question: laptop has E1-2100 processor which has built-in graphics. running a 32bit version of Win8 (not even 8.1, at least not yet). currently has a single 2G ram stick. i can add a second 2G stick and have 4G which i think is the limit for the OS.

BUT - what if i add a 4G stick for a total of 6G and within BIOS allocate (up to) 2G for the onboard graphics? that leaves at least a full 4G for the system. does that even work?

or will the memory that i allocate to the built-in graphics HAVE to come out of the 4G limit that's part of the 32bit OS itself?
 
the graphics memory pool comes from the maximum allocated memory of the OS. so 4GB's maximum will apply regardless of how much RAM your computer has between all its parts.

so 6GB's installed, OS supports 4GB. that additional 2GB's is completely worthless.
 
You might find this of interest: http://www.techrepublic.com/blog/windows-and-office/override-the-4gb-memory-barrier-on-32-bit-windows-81-systems/ . The extra 2GB is not necessarily useless. 32-bit Server versions of Windows have always been able to access more than 4GB, but you need a patch to make it happen on consumer versions. Needless to say, make sure you have good backups before trying this at home.
 
32bit windows can only allocate up to the 4gb of memory. You can have 4 gb, 8gb, 32 gb and all windows will ever see and be able to use (for both system and graphics) as 4gb.

If you want to use more then 4gb you are going to have to install a 64 bit version of windows. No there is no upgrade option, it is a full reinstall of Windows and Programs
 
sounds like i might make use of a ramdisk free-software if i happen to grab a 4G stick to add to said laptop. it's not my own laptop, so i can't be a geek and self-justify installing 8.1x64 or anything else like that. just that the current 2G is quite undersized.
 
IF WINDOWS DOES NOT KNOW THE EXTRA RAM IS THERE, NO SOFTWARE ON WINDOWS CAN USE THE EXTRA MEMORY.

3 people have told you now that 32 bit windows can not address more then 4gb of memory. You can not even allocated it to integrated graphics, anything over 4gb is useless on 32bit windows.
 


so, are the various ramdisk apps completely fake in that case?

i may just throw in a 2G if that's the case, it's cheaper anyways.
 
Ramdisk is not fake, it sounds better on paper then in real world use, but it is not fake.

Lets think this through. Ramdisk is a software that you install on windows that makes a "hard drive" out of your memory. Your 32 bit windows can only see and address 4gb of memory, thus anything beyond 4gb it cant use, thus you cant make a ramdisk out of memory that windows cant access.

Thus yes, 2x2gb is the best option right now.
 

This is untrue. The link that I gave earlier, and Server versions of Windows, show that it is possible to utilize more than 4GB of RAM on a 32-bit version of Windows. This is also common on Linux. It's all down to PAE ( http://en.wikipedia.org/wiki/Physical_Address_Extension ). That 3 people have wrongly stated that there is an absolute 4GB limit only shows how widespread a misconception it is.

What is true is that on a 32-bit system no single process can access more than 4GB (less when you allow for kernel usage).
 


which is also an incorrect statement as only one person has stated that there is a 4GB hard limit. the others stated support limits. the first was a hardware limit, and the other an OS incurred limit that is present on all consumer level 32bit versions of windows.



now that RAM disk software's interesting, if what its claiming is true then he could make usage of the unallocated space.
 
I counted 3 people saying that 32-bit Windows could only address 4GB of RAM. I believe that I have given authoritative references showing that that statement is untrue. Argue all you like, the simple fact is that - with suitable software - 32-bit versions of Windows can make use of more than 4GB of RAM. And I believe that is the question we are discussing.

It's hardly surprising to see the opposite argued as it is a widely misunderstood topic. For a definitive explanation I can recommend the Intel Programmer's Manuals. They are a bible to those of us who play at writing hobby operating systems. In particular you should read the chapters relating to the paging mechanism in x86 processors.
 
wow this thread got real interesting and controversial real fast!

and, um, i made a mistake. turns out her (friend's) laptop actually had 64bit, so i just added a 4G and it's much better now. HOWEVER, my dad's laptop does run 7x32 (confirmed!) and only has a paltry 1G of DDR2 ram. only one slot in total, so if i can SOMEHOW score an 8G stick, that might be worth some ramdisk effort.
 
Unfortunately, I would strongly suspect that there is a limit to the size of RAM that laptop will support. This is not related to the processor but to the hardware design. Many older laptops will only accept 1 or 2 GB modules. With a typical two slots this would limit you to 4 GB; larger modules just wouldn't work.
 
PAE can work, but not all software is coded to handle it which is why I tell people to stick to 4GBs or less. Server software tends to be coded REALLY well, while consumer software/drivers aren't. If one of them gets loaded above 4GBs, PAE can make you life a PITA really fast. Worth a shot if you want to play with it, but with x64 out now there really isn't any reason to bother.
 
The only software that uses PAE is the operating system (and drivers). The user programs are unaware of it as they are just presented with virtual memory addresses by the OS and are totally unaware of the underlying paging mechanism.

An operating system either supports PAE or it doesn't. Windows server does out of the box, comsumer versions of Windows need a small patch to do so. The RAM drive software mentioned is a driver so, in essence, part of the OS. There's no magic about PAE; it's exactly the same mechanism that allows 64-bit OSs to address huge amounts of memory. The difference is that a 32-bit processor will be presented with a 32-bit virtual address whereas a 64-bit one will get a 64-bit virtual address. In both cases (with current Intel and AMD processors) they use a 48-bit physical address. This is all clearly explained in the Intel (or AMD) Programmer's Manuals. I recommend that you study them.

The idea that server versions of Windows are "coded really well" but that consumer versions aren't is ridiculous. The kernels are essentially exactly the same code.