You can disable themes within the display properties. So again disabling the service does not help at all.
Printable View
You can disable themes within the display properties. So again disabling the service does not help at all.
You have absolutely no concept of how the Windows operating system functions.Quote:
Originally Posted by KoolDrew
Windows is divided into operating layers, as this chart shows:
http://upload.wikimedia.org/wikipedi...chitecture.PNG
As you can see, the services are an integral subsystem of the Executive layer, found on the User mode layer.
While the Win32 enviornment and its subsystems POSIX and OS/2 are also a subsystem of the Executive layer and are found on the User mode layer, the applications run on a layer BELOW this, as the application branches from the enviornment subsystems show.
Each layer has a different prioritization, with the highest priority being the kernel. As far as memory management is concerned, it's entirely determined by prioritization. Since the services are of higher priority than the applications, there is never an instance where a service will reduce its footprint just to meet the requirements of X application.
Futhermore, SERVICES DON'T HAVE MEMORY MANAGEMENT WHATSOEVER!
Services are given a set minimum footprint that they rapidly expand far beyond, and once they expand to that, they don't shrink in usage.
You say that services that aren't in use don't use memory?
The easiest way to prove this is to set your theme to Windows classic and enable the Themes service.
Guess what?
The themes service, when unused, has a footprint of 200-500KB.
Lol, that's exactly what I'm studying ATM for my exams within two weeks :toast:Quote:
Originally Posted by teqguy
Anyway, for the individuals that don't believe me on my word:
after my exams I will post some benchmark scores: one with standard WinXP install, one with same install with standard tweaks (themes off, process killed etc) and one with my WinLite install.
Stay tuned :toast:
teqguy, instead of me arguing with you I went and asked someone who happens to be a kernal/driver developer anbd this is what he had to say.
Quote:
Well, I don't know where he got his diagram or his ideas... but he has several problems. His diagram is wrong in a few important details, too. It looks like an attempt to present the same info that's in our standard architecture diagram, but modified to avoid copyright issues... unfortunately not completely accurate. For example, while the "security subsystem" (by which I guess he means the lsass.exe process) is indeed 'integral" to the system, the workstation and server "subsystems' are not - they are just win32 services, part of svchost, and you can stop them if you want to.
> As you can see, the services are an integral subsystem of the Executive layer, found on the User
> mode layer.
This comes from a confusion in terminology. Alas MS has in many cases used the same word for several different things. "Services" is one example.
The "executive services" are not the "service processes" we are talking about stopping. "Executive services" are simply APIs exposed by the exec for call from other components, and in particular for call from user mode. They do not have processes associated with them and cannot be stopped - they aren't "running" in the sense that a process or a thread runs. Threads just call them when they want to, and they execute in the context of the calling thread like any other function or procedure.
Nor are they "found on the user mode layer," they are in kernel address space (though they are mostly pageable - their code is part of the system working set).
The services we are talking about stopping are, by contrast, processes running in user mode. They are peers to application processes.
> While the Win32 enviornment and its subsystems POSIX and OS/2 are also a subsystem of the
> Executive layer and are found on the User mode layer, the applications run on a layer BELOW this,
> as the application branches from the enviornment subsystems show.
The arrows on the diagram may show that but it isn't reality. First off the OS/2 subsystem hasn't been around since Win2K, and you only get Posix if you download "Windows services for Unix" from MS. But anyway... although a Win32 application process (or service process) does depend on the Win32 subsystem process (csrss) for a few things, for almost all OS API calls, it interacts directly with the executive. An app or service process is a peer in terms of thread scheduling and memory management to the Win32 subsystem process.
> Each layer has a different prioritization, with the highest priority being the kernel. As far as memory management is
> concerned, it's entirely determined by prioritization.
Nonsense. There is no such concept as "memory prioritization," and the layers do not imply priorities. "Priority" in Windows applies to threads and processes, and is important for scheduling; the pager and working set manager take no notice of it. And some service processes run at LOWER priorities than some apps. Not that that matters for memory allocation.
What DOES matter for memory allocation is a process's page fault rate. If you're incurring a high fault rate your working set upper limit may be expanded, so you can hold more pages of RAM at one time, so as to reduce the fault rate. If you're incurring a very low fault rate, and the overall fault rate is high or the system is short on free RAM, your working set upper limit may be reduced, so as to free RAM for other processes to use.
And guess what... this even applies to the pageable code and data in kernel mode. A high fault rate of a process -- even a low-priority process! -- may cause the memory manager to shrink the operating system's working set, causing file cache and pageable k-mode code and data to be pushed out of RAM in favor of the process.
> Incidently there is
> never an instance where a service will reduce its footprint just to meet the requirements of X application.
> Futhermore, SERVICES DON'T HAVE MEMORY MANAGEMENT WHATSOEVER!
> Services are given a set minimum footprint that they rapidly expand far beyond, and once they expand to that, they don't
> shrink in usage.
So I guess when my svchost's are taking a collective 40 MB or so, and then I run a big "hog" of an app and I see the svchost's have been cut down to 20 MB, I'm just hallucinating. Too many drugs, I guess.
> The themes service, when unused, has a footprint of 200-500KB
Oh gosh, half a megabyte. I wonder if I can stand it? Yes, this 200-500KB is the minimum working set size of ANY process, service or otherwise.
This guy, whoever he is, needs to attend one of our "Windows Internals" seminars. Where'd you run into him, anyway?
If you want to enter a third party into this discussion, let him do it on his own terms.
You are the one that conjured up the nonsense that a service that's not in use does not have a footprint, not he. So, nothing you could possibly copy and paste could hold a grain of salt around here, let alone the respect of people with fewer posts than yourself.
Furthermore, your friend doesn't have a clue what he's talking about either.
+ The Windows kernel does not directly address subsystems or services... this was one of the improvements over NT that came with 2000.
+ OS/2 and POSIX are prevalent in Windows XP out of the box, obviously your friend has never heard of the WoW(Windows on Windows) 16bit subsystem.
+ If his accumulative footprint for SVCHOST is at 40MB, he's dealing with a virus.
+ Despite the amount of memory used, the bottom line is that services that go unused should logically be disabled.
+ Applied science shows that Windows XP operates better with fewer services enabled, so there is nothing you or your immature friend could conjure up to dispute that.
If I add my System, user, Local Service and Network Service svchosts, I get 50MB, and no-way that I have a virus :)Quote:
Originally Posted by teqguy
Btw, interesting discussion now, but as teqguy said, plz invite your friend here so he can participate directly and not via you.
IMPORTANT: plz no whining, flaming and so-on. This is a forum, all different kinds of ppl, respect eachother ok? Thx :toast:
Oh, before I forgot: youre friend did now everything exept answering the main question in this thread: does the disabling of services affect the performance? (Ok I know the answer from reality, but I'm curious about his answer :))
There's no possible way those services could accumulate that much memory under normal circumstances without outside interaction.Quote:
Originally Posted by ZMarre
You've either purposely enabled all of the services, have a virus, or are suffering from Windows' infamous memory leak.
In any case, most of those services shouldn't be enabled all the time or at all, let alone have ever been included by default in the first place.
This is on a regular install, with lots of processes running (of which 7 svchosts). Quite normal if you ask me. And absolutely no viri.Quote:
Originally Posted by teqguy
Nah, Windows is made to satisfy ALL customers in the whole world u know...Quote:
Originally Posted by teqguy
Explain Fast User Switching, Background Intelligent Transfer, Smart Card, Remote Registry, Upload Manager, Web Client, UPNP, Windows Time, QoS RSVP, Error Reporting Service, NetBios Helper Service, and Messenger Service.Quote:
Originally Posted by ZMarre
None of those appeal to a vast majority, and some of them are even exploitable.
Microsoft couldve at least left these set to Manual, but instead most of them are set to Automatic.
explain how superPI times are quicker then with services disabled hmmmm? make me a graph for that one! :slap:
Uh, they probably aren't... but SuperPI doesn't function like a typical application, so just because that application isn't affected, doesn't mean other applications won't be affected.
Besides, if you don't need them running, there's really no point in having them enabled, despite whether or not they affect performance.
Quote:
Originally Posted by teqguy
OOOOOHHHHH! hey dude! can you give me the link then for how the placebo affect affects computer programs please! maybe if i get my computer to THINK its faster......... :slap:
dude, ive seen it happen myself, disabling services speeds up a computer,
If you're just going to be sarcastic and condescending, don't bother posting.
The amount of services running has a direct effect on the amount of resources used; it's pure ignorance to think otherwise.
If you have resources devoted to services, that directly translates to less resources available for applications... and while you might not notice that in the applications you use, the applications I use could appreciate from having additional memory available to them.
wait wait wait, you went from saying disabling services has NO effect from disabling services to saying disabling services has a POSITIVE effect.Quote:
Originally Posted by teqguy
dude your hilarious, anybody else think this is rather entertaining? guess you an al gore wannabe?
Dude, if someone has to reconsider hisself, I think it would be you... I see three posts of you here, and two of them are just below-level. Seriously.Quote:
Originally Posted by WeStSiDePLaYa
And he says that Windows in his whole content speeds up (it does! And if you don't believe me, well hell go outside and get a life); but maybe SuperPI doesn't because it only needs marginal use of RAM.
And if you would have read the whole thread, you would have seen that I will post all sorts of benchmarks men can think of, with full install, full install with services disabled, and nLite'd install.
BTW, didn't bachus_anonym post allready benchmarks? :confused: Seems like they disappeared? :poke:
Take it easy guys... That goes to WeStSiDePLaYa, Zmarre, teqguy...
I do not want to see anyone jumping at eachother or flaming, OK?
Anyway... I decided to run some SuperPi 1M & 16M calculations, to see how much difference it really makes as far as performance is concerned. Tests are done on my Internet PC with WinXP Home SP2. Startup items were disabled using msconfig. Benchmarks were run right after OS started and screenshots were taken in exactly same same order, after last run was completed (4th for 1M and 2nd for 16M).
So here we go:
-----------------------------------------------
SuperPi 1M
WindowsXP Home Default services started / no startup
1st run - 37.634
2nd run - 37.624
3rd run - 37.635
4th run - 37.624
WindowsXP Home - only 3 essential services started / no startup
1st run - 37.574
2nd run - 37.574
3rd run - 37.574
4th run - 37.564
As you can see, 1M does't really benefit from disabling services, but surely it freed 34-44MB of RAM space.
-----------------------------------------------
SuperPi 16M
WindowsXP Home Default services started / no startup
1st run - 15:17.178
2nd run - 15:14.916
WindowsXP Home - only 3 essential services started / no startup
1st run - 15:13.463
2nd run - 15:13.664
Again, 16M does't get too much of a boost from disabling services, slightly more than 1.5s.
Well, honestly I was a bit surprised by te results. When I get a chance I will repeat those tests on my Benchmarking PC, that has Windows Server 2003 installed as I do not really bench with WinXP. As far as I remember, on Server2003 I get ~4s difference in SuperPi 16M coming from default services to only 3 services left. I will post results as soon as I get a new CPU.
Warned for flaming - conrad.maranan
In syntehtic benchmarks, yes I do admit you ma see an increase in scores, but when it comes to real world performance such as gaming there will be no benefit from disabling services.Quote:
but there is proof that it does make a difference.
And gaming is the end all to computing?
Try using several applications that actually demand resources and you'll see a difference.
I was using gaming as an example of a real world test.Quote:
And gaming is the end all to computing?
placebo effect ;) Unless you have ~128MB of RAM there will be NO real world performance benefitQuote:
Try using several applications that actually demand resources and you'll see a difference.
buddy, i may not use proper puncution or spelling, but at least i have whats called reading comprehension, if you go and read my posts it says I ALSO THINKI DISABLING SERVICES IS A GOOD IDEA.Quote:
Originally Posted by teqguy
Quote:
Originally Posted by WeStSiDePLaYa
hmmmm...... seems to me that my very first post is stating that superpi times are faster with services disabled.
Freeing up memory can't possibly have anything less than a positive effect, because you're increasing the amount of available memory for other applications.Quote:
Originally Posted by KoolDrew
This is by no means a placebo effect... and it's ignorant to think otherwise when the logic is presented to you on a silver platter.
The bottom line is that there is absolutely no reason to have unused services running, so despite whether or not you or anyone else notices an inherent benefit from disabling them, you can't possibly sit there and contest that they should remain enabled simply for the sake of justifying your point.
It's obvious that you don't have and have never used applications where 512MB is near the minimum requirements, and 1GB is closer to the recommended requirements.
Try playing around with XSI, Premiere, CCE, or any professional software and tell me if you can still function with a memory insufficiency.
Uh... then what was that whole "placebo effect" statement you made in one of your posts?Quote:
buddy, i may not use proper puncution or spelling, but at least i have whats called reading comprehension, if you go and read my posts it says I ALSO THINKI DISABLING SERVICES IS A GOOD IDEA.
Quote:
OOOOOHHHHH! hey dude! can you give me the link then for how the placebo affect affects computer programs please! maybe if i get my computer to THINK its faster.........
Hell, how many times I still have to say it??
If I tell you that the difference between a full 595MB WinXP install and an install of 95MB is so huge that I almost fell of my chair, THEN IT IS LIKE THAT damned
I did it, you didn't so WHY THE HELL DO YOU DOUBT MY EXPERIENCE?
(not to one particular person, but to all who keeps saying that it can't be)
And if you don't call Windows-navigating real-life, then WHAT for godsake men can call real-life? :poke:
To proof that I'm no way under placebo-effect: I standard underclock my PC's (my PIV 2.4GHz is at 1.8GHz ATM for example) because I don't notice a single performancedrop between 2.4GHz and 1.8GHz.
If ricky and pedro say that by changing your ACPI uniprocessor driver into multiprocessordriver you gain something with SuperPI, everybody just nods and praises them. But when I say that with a uber-lite Windows everything speeds up, almost everybody screams like 'that can't be true, you're lying, whatever'... *sigh*
ZMarre, don't worry about it.
If you look at every single post KoolDrew and Westsideplaya have made, you'll see that all of their posts are single sentences that lack any real insight or depth to them.
So, if you're worried about anyone actually taking them seriously, I find it very unlikely.
I mean, come on, they're arguing against logic and tangible proof. :rolleyes: