Total and Available Virtual Memory

What is the relationship between Total and Available Virtual Memory?

 

What is the distinction between "Total Virtual Memory" and "Page File size "?

 

Before posting my question, I consulted Webopedia, but it does not clarify my question, which is prompted by the fact that my "Total Virtual Memory" is more than double the size of the "Page File size ".

 

At My Computer>System Properties>Advanced>Performance Options…>Virtual Memory, the "Total paging file size for all drives: 995 MB ".

How can "Total Virtual Memory" exceed this figure?

Isn't "Total Virtual Memory" the space taken by the Paging file?

 

Nope - it's Pagefile + RAM

 

OK. If "Total Virtual Memory" = PageFile + RAM, then what accounts for the following resuts?

Total Virtual Memory = 2,035 MB (taken from Start>Programs>System Tools>System Information)

Page File = 995 MB (taken from My Computer>System Properties>Advanced>Performance Options…>Virtual Memory)

RAM = 524 MB (taken from Start>Programs>System Tools>System Information)

But 2,035 does NOT equal 995 + 523!

 

Acording to rambler, above, and http://osr5doc.ca.sco.com:457/PERFORM/virtual_mem.html,
the sum of physical memory and swap space defines the total virtual memory that is available to the system. Yet, I repeat,
with my Win2k PC, Total Virtual Memory = 2,035 MB (taken from Start>Programs>System Tools>System Information)

Page File = 995 MB (taken from My Computer>System Properties>Advanced>Performance Options…>Virtual Memory)

RAM = 524 MB (taken from Start>Programs>System Tools>System Information)

But 2,035 does NOT equal 995 + 523!

So what's going on?

 

Page File = 995 MB

This is the initial size. The total virtual memory referred to in System Information adds the maximum size value which is probably set to around 1512 MB in your case.

 

Nope - my definition still stands. Windows treats RAM and pagefile as a single address space, albeit with slow and fast components.

However, my PC has 512Mb RAM and a fixed max=min=384Mb (=393,216K) pagefile. System Info reports this as:

Total Physical Memory 522,736 KB
Available Physical Memory 263,960 KB
Total Virtual Memory 1,405,780 KB
Available Virtual Memory 928,052 KB
Page File Space 883,044 KB
Page File C:\pagefile.sys

Something amiss here surely? Total and Available RAM are correct (if a small amount of shared RAM is included) - Task Manager agrees with THESE figures, and has a figure of 883,044K for "commit charge limit ". MS defines this as the "Total Virtual Memory ", and is the sum of RAM and pagefile - "the maximum size your virtual memory can be without making changes to its configuration ". This implies that the upper limit on pagefile(s) is taken into account in the calculation - pretty well what Steve is saying

However, in my case, RAM + max. pagefile = 522,736 + 393,216 = 915,952K and there seems no way to massage these figures into what either Sysinfo OR Task Manager show.

Until I read this thread I was quite happy to accept what Task Manager showed as "tablets of stone ", now I'm not quite so sure - but where does Sysinfo get it's figures from? Notice it shows "page file space ", which is pretty unequivocal, as the same figure that Task Manager calls "Commit Charge Limit "!

I did wonder a while ago why AIDA32 showed my "total swap space" - meaning pagefile(s) - as 862Mb when I've got a fixed 384Mb pagefile, but just wrote it off as a bug. I now see that 883,044K/1024 = 862.34Mb. - obviously the same source Sysinfo uses. I just popped up Norton System Information - the third opinion seems to be that I've got "511Mb total physical memory" - fine, but "1548Mb total Windows memory" which it defines as "physical memory plus virtual memory provided by the Windows swap file ".

I think I'll go and have a lie down for a while.....

 

A+ Memory answer

virtual memory is the amount of ram that the operating system sets aside for runn for software applications. physical is total amount you have in your pc. The maximum amount of virtual for windows 2000 Pro- DataBase server 4095 Xp is 4096.

 

For the record, my Page File size is FIXED at 995 MB.
So the mystery persists. I hope rambler can shed some more light when he gets back up after taking time to lie down.

At least, I've discovered from rambler's contribution that I'm not alone. Now it would be good for those of you who have taken the time to post here to check your systems and share your findings with the rest of us.

 

Can't shed any more illumination on the mystery of the figures reported in W2K (I'm feeling refreshed now, thank you NormanS.) I'd like to add that I've been working with Virtual Storage O/S since the late '70s early '80s, when IBM introduced them on large mainframes, so I understand the concepts involved.

In those days, RAM (we called it "Main Memory ") was VERY expensive. The technique of using small fast processor caches had only just been developed, and memory had to be as fast as possible for the processor to utilise its power. 9nS cycle time was common, compared with todays RAM speeds of 6nS for PC2700 RAM - for those days, that was FAST. Cost was rapidly coming down as larger chips were designed, but could be of the order of $1/2 million for 16Mb. With processor boxes costing upwards of $5m, the aim was to utilise processor and RAM at 100% for as much of the time as possible.

Design technicians realised that large amounts of this costly RAM were not accessed by the processor for minutes or even hours, and so the concept of Virtual Storage was born. Main memory was supplemented by large pagefiles spread across the disk subsystem for I/O efficiency. Programmers could now use a virtual address space of say 2Mb, knowing that typically, only 200 or 300K would need to be in RAM (the "working set "). The O/S took care of the rest, monitoring the "age" of each 4K page (time since accessed), and paging out the oldest as programs demanded more memory than available in RAM. If a program needed to access a page that was now on a pagefile (a "page-fault "), the O/S memory manager would schedule a page-in, paging out an "old" page first if necessary - a "swap ", hence the alternative term swapfile.

Since pages were (and are) a fixed 4K, the paging I/O could be optimised, with page requests chained together and streamed through the disk channels and controllers.

The overall concept has changed little, except that RAM is now relatively cheap, and it's not uncommon for PCs to have all working pages in RAM, with free RAM available and only some key O/S pages stored in the pagefile(s).

I just thought you might be interested, and I'd also add that since those early days, very little has changed in principle in either mainframes or PCs. Hardware has got faster and much cheaper, but works in very much the same way.

I'll continue to research the conflicting and apparently erroneous figures reported by Sysinfo, but don't hold out much hope - I've found very little from MS on this,