For what its worth if Windows operates anything like linux when it comes to measuring CPU usage then it is quite possible to have a very busy process which doesn't register. Its very rare to so this on windows I suspect due to the high frequency of the kernel. You can check back in the archive for reasons on this but its a common thing to see linux 2.4 based kernel machines show wildly inaccurate readings in top etc as the info provided by the kernel has a course granularity.
The fault boils down to the fact that usage is calculated in terms of full timeslices. However if a process's usage pattern is such that it yields before using a full timeslice then it will appear to use no CPU. Its not a bug in the app or in the kernel is just an optimisation that has a rare but nasty side effect. Linux 2.6 kernels suffer less from the issue due to their higher HZ rate ( 1000HZ 2.4 was 100HZ ) and Windows similarly.
Note: Also multiple CPU's has a dramatic effect under windows as it calculates in % of the total CPU available so 5% on a dual CPU machine = 10% of real CPU and if the box is a P4 ( hyperthreading ) u'll have 4 CPU's so that's really 20%. So in effect anything under 4% on a dual cpu hyperthreaded machine is hard to spot.
Just some background info people may be interested in :)
N.B. I cant comment on the CPU graphs in HL itself as we run console mode so dont even see them.
A good tool to check these things with is process explorer from www.sysinternals.com it will graph the cpu total and per process.
Steve / K ----- Original Message ----- From: "Ian mu" <[EMAIL PROTECTED]> To: <[email protected]> Sent: Saturday, March 19, 2005 2:06 AM Subject: Re: [hlds] AMD 64 GOOD or BAD?
Hehe ta Steve, I do apologise to others I can get a bit tetchy at times as I have done to you in the past :) (I certainly appreciate 100% any information provided even if seems misleading as helps build a clearer picture), but 0% full 24 slot servers gets a bit silly under anyones comprehension as far as I'm concerned (apart from starting/idle games/maps with people on).
Btw I just want to confirm I don't necessarily disagree with what people "see" in terms of graphs etc (i.e I'm not calling anyone a lier from their info..I've seen it on different Linux kernels etc to know how things can "appear" different), just in terms of what seems "reasonable". 24 slot action (sounds like a porn movie) at 0% just doesn't make any logical sense or you would be running hundreds of them, even though I've seen similar myself. Run a certain amount and you will get complaints. You "know" its not right in what its representing, so "why" do people believe it, yet "don't" run that many?
Simply put, comes a point of what something says, and what you get play wise doesn't add up. It's up to you what it is.
Btw I'm not having a go at Team Pfeffer, its actually at least good to see someone comment on their fps values etc which I think is good information and more than many (I run some at 50, some at 500 for what its worth, and have said get more complaints on 500 than 50 so I agree on the stability side...just I don't think thats "all" by any means, but if I was an expert I would offer more advice). More information is good, and sorry if it felt like I was having a go or anything, just I don't believe the report as opposed to you if that makes sense?
Ian
On Sat, 19 Mar 2005 00:34:36 -0000, Steven Hartland <[EMAIL PROTECTED]> wrote:
lol well said Ian I couldn't be arsed to say anything myself seen it too many times before and its getting tiresome.
Steve / K ----- Original Message ----- From: "Ian mu" <[EMAIL PROTECTED]>
> Tbh if you use 0% cpu on a 24 slot full single cpu server, my first > instinct as someone I consider relative sane (I may be wrong, > naturally first sign of madness!) would be to check the reporting. If > anyone thinks they use 0% cpu for a FULL 24 slot server on a single > cpu server (or even duals) (not talking about empty ones, they are > irrelevant, some people don't really seem to get this), they are > simply an idiot pure and simple. Do you confess to not knowing wtf you > are talking about? > > If you use 0% cpu with a full 24 slot server, why not load it with 100 > 24 slot servers on it? Simply put because its complete rubbish :). So > either you believe it or not. People don't because they don't load > them like that, even with that reporting. > > If you get 0% cpu on a full 24 slot server full, in action (not on a > restart), either fix your machine, or fix your braincells. Either will > do. I couldn't care less, but people stop talking like complete idiots > as server hosters. Be aware as said previously of misreporting as the > idiots seem to ignore, which we are starting to see who are the > clueless people hosting servers and those who aren't.. > > > On Fri, 18 Mar 2005 12:21:35 -0800, Team Pfeffer <[EMAIL PROTECTED]> wrote: >> That is correct. That screen shot was during dust2, so that was a easy >> map for the server. The tiny spikes of 3% showed the start of a round >> during the buy time. We dont run any booster so the server fps stays >> at 64fps, with a few drops during the round start. A map change will >> use anywhere from 10% to 50% of the cpu for a instant. >> >> I can open the task manager and the performance tool in windows, but >> the result will be the same, one game without a ping booster will >> hardly load a amd64 system under windows beta server (RC2). It runs >> smoothly, so we don't try to tweak it.
================================================ This e.mail is private and confidential between Multiplay (UK) Ltd. and the person or entity to whom it is addressed. In the event of misdirection, the recipient is prohibited from using, copying, printing or otherwise disseminating it or any information contained in it.
In the event of misdirection, illegible or incomplete transmission please telephone (023) 8024 3137 or return the E.mail to [EMAIL PROTECTED]
_______________________________________________ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds
_______________________________________________ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds
================================================ This e.mail is private and confidential between Multiplay (UK) Ltd. and the person or entity to whom it is addressed. In the event of misdirection, the recipient is prohibited from using, copying, printing or otherwise disseminating it or any information contained in it.
In the event of misdirection, illegible or incomplete transmission please telephone (023) 8024 3137 or return the E.mail to [EMAIL PROTECTED]
_______________________________________________ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds
