[Home] [Groups] - Message: [Prev in Group] [Next in Group]

nu.kanga.list.mud-dev

10639: Re: [MUD-Dev] Text Parsing

[Full Header] [Plain Text]
From: Petri Virkkula <pvirkkul@iki.fi>
Newsgroups: nu.kanga.list.mud-dev
Date: Fri, 11 Jun 1999 06:35:20 +0300 (EEST)
References: [1] [2] <-newest
Organization: Kanga.Nu
>>>>> "Matthew" == Matthew Mihaly <diablo@best.com> writes:

Matthew> Hmm! So TOP isn't particularly accurate then? I was wondering

	It is accurately showing how much a process is _directly_
	using memory. In my example there was 443M missing. My
	_guess_ was that it was mostly used by a disk cache, thus
	indirectly by processes, and as our driver is the only active
	process, I think it is reasonable to say that the driver uses
	most of the disk cache and then also most the "missing"
	memory. 

Matthew> about this myself, becuase as malloc creeps up due to a
Matthew> memory leak, performance slows, even though it shows no
Matthew> virtual ram being used. I will take your recommendation and
Matthew> upgrade to 256 megs of RAM immediately, thanks!

	Even when there is no virtual RAM is not being used the extra
	RAM helps keep disk cache hit percentage high, in our case
	read cache hit percentage is 99%, write cache hit percentage
	is 92%. The "sar", "iostat" and "vmstat" utilities are your
	friends here!


	Petri



_______________________________________________
MUD-Dev maillist  -  MUD-Dev@kanga.nu
http://www.kanga.nu/lists/listinfo/mud-dev