Page 1 of 1

running out of virtual memory

Posted: Sat Aug 13, 2005 8:11 pm
by ReadyMan
I play a game called Rise of Nations that eats up my memory like crazy. Whenever I finish a game that takes 30 or more, trying to activate anything takes forever, and the only way I can get things to load properly again is to reboot.
Is there a way to clear the memory w/o rebooting?

thx!

-RM

--I have 512mb of Rambus on a P4 2.26

Posted: Sat Aug 13, 2005 8:28 pm
by DCrazy
What OS are you running?

Posted: Sat Aug 13, 2005 8:43 pm
by ReadyMan
win xp home

Posted: Sat Aug 13, 2005 9:01 pm
by Krom
Your gonna hate me for this, but the only real solution is to buy more memory. ;) And rambus is still expensive last time I looked.

Posted: Sat Aug 13, 2005 9:04 pm
by Matrix
Just build that new computer u have been planing on getting for like the past year now :P

Posted: Sat Aug 13, 2005 9:35 pm
by ReadyMan
Xciter--Thanks for the tip....and Its set to 384 mb...
Will increasing it help?

Matrix-- :) sir, yes sir! I'm working extra to try and get a few extra $ for that system (and waiting for the 7800GT to show up to lower prices--which should be pretty quick)

Krom--I hate to drop any more $ into this system as it becomes my wife's office machine as soon as I get he new one...

Posted: Sat Aug 13, 2005 9:54 pm
by Matrix
I agree with xciter, u need a larger page file, I say no less then 1GB.

And nice choice on the 7800GT, I've been thinking about pitching my AGP MSI and going to a DFI w/ a PCIe 7800GT

Posted: Sat Aug 13, 2005 10:07 pm
by Jeff250
You don't need to clear your memory. That's why things are running slow to begin with. After the game used it all and quit there's nothing left stored in it and it all has to be reloaded. :P But yes, follow above suggestions.

Posted: Sat Aug 13, 2005 10:48 pm
by ReadyMan
there are two settings:
Initial size:384
Maximum size:768

what should I set these both to?

(sorry for the dumb question)

Posted: Sat Aug 13, 2005 10:48 pm
by CDN_Merlin
setting your pagefile to 1 size is best as it will never get fragmented. I'd set it to zero, reboot, defrag, then reset it to 768 min 768 max. This will make sure the entire file is in one continuous spot.

Posted: Sat Aug 13, 2005 11:08 pm
by fliptw

Posted: Sun Aug 14, 2005 12:49 am
by Ferno

Posted: Sun Aug 14, 2005 1:20 am
by ReadyMan

Posted: Sun Aug 14, 2005 1:41 am
by Matrix

Posted: Sun Aug 14, 2005 9:59 am
by Krom
384 MB page file on a 512 MB system, yikes that was a bit small, for 512 I'd set it to 768-1536. On my 1 GB system I have it set to 1536-3072.

Posted: Sun Aug 14, 2005 1:02 pm
by fliptw
Xciter wrote:You don't need to lock the page file!!! It's one continuos file which will only grow if you reach it's minimum and if set up right it's not going to grow. Read the part of that FAQ about Dynamic sized page files.

locking it and you lose the ability of it it grow if needed...
1. Space is cheap.
2. Windows is paranoid about loosing data. if it will grow the pagefile, windows will dump the data to disk immediatly... anything past the min size will be fragemented, further degrading performance during the pagefile resizing(in writes of 4KB each). Windows is the modern only operating system that allows that - and its a bad hack that exists that still around do to people bitching about win95 not running on 386s with 4 megs of memory and a 30 Meg HD. If the option wasn't there to begin with, nobody would be saying its a good thing to do.

Posted: Sun Aug 14, 2005 1:59 pm
by BUBBALOU
Krom wrote: I have it set to 1536-3072.
X2

Posted: Sun Aug 14, 2005 2:21 pm
by fliptw
Xciter wrote:If on the other hand the system needs more pagefile space and it can't, because you've disabled expansion, some program, or perhaps the OS, will outright fail. This is far, far worse than the very mild and occasional performance hits you take in the "expansion enabled but not used often" scenario.
That is correct behavior. If for some reason the OS needs more than the maximum amount of pagefile specified, I surely hope it would fail.

But your probably looking at this thru the filter of your super fast scsi drives and SMP systems, where your choice of hardware minimizes the performance impact of such pagefile expansion... and not witnessed it on a machine with a low-end 5400 rpm ide drive with a process that was also thrashing the HD at the same time it was adding more pagefile. Locking the page file in that instance would've reduces the CPU overhead concering the pagefile expansion - it just becomes writes to the pagefile, rather than writes to the file table and the pagefile.

But Im thinking you treating the swap file as an extension of real ram, which is incorrect.

Posted: Sun Aug 14, 2005 2:49 pm
by DCrazy
Xciter wrote:Even if the pagefile fragments, it's going to be fragmented in MUCH larger chunks then Windows can write to it in (64k)


Won't Windows only write to the pagefile in 4K blocks, considering that's the size of a page?

Regardless, locking the pagefile seems foolhardy unless you notice constant HDD thrashing in a memory-intensive program (like a game or something). Then locking the pagefile would give a good estimate as to at what point it gets full, useful for increasing the minimum size of the pagefile. Why would anyone want to disable the safety buffer that Windows gives you, especially since it warns you anyway when it's expanding the pagefile? What's worse, no process has priority in memory allocation over any other process, meaning that Explorer.exe and MemoryHungryProcess.exe will compete and eventually both die due to the lack of resources.

Posted: Sun Aug 14, 2005 3:13 pm
by Krom
At most I've seen my page file get into two fragments and that was easily fixed since I have multiple drives. When it did get fragmented, it was on the order of several hundred MB split off, not 4 or 64k.

Posted: Sun Aug 14, 2005 7:23 pm
by Top Gun
Pardon my complete technical ignorance, but what exactly is a pagefile? I'm asking because, after running FS2_Open, my system (P4 3.0, 512 RAM, Radeon X300 64MB) runs like a dog, usually until I restart it. I was wondering if fiddling with the pagefile size would help to alleviate that problem or whether it's just caused by the limitations of my hardware. If it would help, where exactly would I look to fiddle around with it?