OS (SOFTWARE) tweaking on high-end machines

K

kleox64

[H]ard|Gawd
Joined
Sep 21, 2004
Messages
1,467
Is there any point to doing this now we have 3.8GHz+ P4's and 2.6GHz+ AMD's.
This is for a games machine running at high res with AF and AA here and thier.

Aside from pagefile, some services does tweaking anything else really help framerates in HL2, DOOM3, FARCRY etc...

I took a look at tweakxp and just too many setting to contemplate.

kleo
 
Is there any point to doing this now we have 3.8GHz+ P4's and 2.6GHz+ AMD's.
Click to expand...
Some tweaks, sure, but in general I find most are simply not the panacea to a "bloated OS" they claim they are.

Tweaking services gains jack squat, and if you have *enough* RAM disabling the PF won't help much, if any. Other than that, testing, testing, testing... Try a tweak and tell us what it does.

I'm starting to detest tweak guids, too much FUD or ignorance in them...

 
Your best bet would be to use nLite and strip out all the excess junk in XP before you install it.

Doesn't it make more sense to install a lighter version of XP that doesn't even have the gunk instead of installing XP then having to go in and pick it all out?

Go check out nLite sometime and read through the forums. It's a simply amazing piece of software for what it can do.

Besides that, having the most RAM you can afford/put into the machine is the biggest factor in most PCs. Faster hard drives, faster CPUs, faster video cards... all of those take a back seat to the most RAM you can possibly have. Tweaking the services obviously will have an effect on the overall performance of the machine also.

For people that have 1GB of RAM or more, using a RAMDisk for temporary files (all kinds, including the Temporary Internet Files cache for IE or whichever browser you use) can speed up your PC like nothing else. I know people with P4 3.4 EE CPUs/AMD Ahtlon64s and I spend 20 minutes showing them how their "supposedly" fast machine has been slow all along.

In 20 minutes their machines literally go from "fast" to "ludicrous speed" because of using a RAMDisk for temporary files. Even notorious RAM hogs like Photoshop get a pretty serious boost in performance from using a RAMDisk for temporary file storage. Note that although you can assign a lot of RAM for Photoshop to use a "temporary workspace," that's still not the same as using a RAMDisk that Windows sees as the temporary file space.

Oh, and don't think about using a RAMDisk for your pagefile: it won't work, and even if by some miracle you actually get Windows to recognize a RAMDisk for a pagefile, you'll hurt performance in the long run.

There doesn't seem to be an all-inclusive tweaks page on the Internet the last time I looked. All the information that comes in handy is always scattered across the entire Net and you spend an inordinate amount of time trying to bring it all together to make your PC as fast as it can be running the OS you choose.

Good luck... :)

Paul
 
I use xplite to remove most the windows bloat and does make it feel more responsive, whether it has an effect on in game framerate is another matter.

I should of added machines with 1GB minimum of RAM for this thread.

That RAMDISK for internet files sounds interesting, ill give it a go.

As for services less services will have an an effect on overall system performace but i dont think its enough to actually effect framerate, my 2 cents.

kleo
 
As for nlite, looks real dodgy to me, as if its gona give me so many problems it wont be funny.
 
It's a very solid program, actually.

Nuhi, the creator of nLite, is constantly improving it and adding new features to get the most out of a custom small footprint XP installation.

I know that in today's world of 300GB hard drives, most people would say, "What do I care about a few Help files taking up a few megabytes of space on the drive, I have 250GB to spare," but a true tweakhound only laughs at such comments because it does matter if you want the ultimate performance from a box.

Especially one you built yourself.

For me, I work strictly with laptops (I'm on an IBM ThinkPad A21m with a P3 800 and 256MB RAM, working on a 20GB drive) so space is at a premium. And because of the older processor and the PC66 memory, doing everything I can possibly do to make it faster is worth it.

Using nLite, I turned my SP2 install CD (671MB) into a custom XP install CD roughly 157MB in size, and the rest of the space on that CD (filling out the 700MB) is comprised of my most commonly used applications including Photoshop, Acrobat 7, Office 2003 (another stripped down version I created for a simple Word/Excel install) and about 30 other small utilities.

Not everyone goes this far, but I do.

:D

Paul
 
Anything you can do with XPLite still just leaves the code itself on the drive. So...

My choice is to never install that craptacular code bloat onto my hard drive in the first place.

:)

To each his own, however.

Paul
 
ok so its on the drive but windows isnt using it and doesnt have access to the files.
 
br0adband's post is a perfect example of bad information.

Want PF in RAM, disable the PF. Voilla. You obviously don't understand the point of the PF if you want to create a RAM disk to place the PF on. The PF is essently a substitute for RAM... Anyways, disabling it ain't going to help much though, only in very specific cases from my findings.

Tweaking services doesn't help because of paging... Unused processes will get paged to disk and free RAM, so why stop the services in the first place? It's been shown before on in-game tests that disabling services doesn't help with performance... It will decrease boot time a bit, but what good is that?

br0adband, have you ever tested these tweaks? What benefit does disabling services give you?

edit: why do I think this has been discussed to death before... Oh, that's right it's because it has been. Search is your friend.

 
I've been tweaking Windows since most people were in diapers, so yes, these tweaks are tested thousands of times across thousands of machines in my 25+ year career.

Please note I didn't suggest making a pagefile on a RAMDisk, I said if you happened to get it working because some people that use a RAMDisk automatically assume, "Hey, if RAM is faster than a hard drive, and a RAMDisk is faster than a hard drive, logically putting my pagefile onto a RAMDisk will boost my performance right into orbit!"

Luckily, Windows is smart enough to not allow you to put a pagefile on a non-physical drive, so those people that try the above example (you all know who you are) already know it won't work.

Tweaking services can always help to some degree, but no, it won't magically make your PC 50% faster at Half-Life 2 because you disabled a few unneeded ones. Never said that, never even implied it.

Disabling services (especially on older machines) has benefits of increased reliability and stability in the long run. It's a simple fact: the less you have going on, the less chance you have of something going wrong.

It's not bad information, none of what I said is, it's all factual.

As for the search feature, that's a given.

As for why it's been discussed to death before, that's simple: new people join this and other forums daily, and just using a blanket statement like "Oh for fuck's sake, just use the Search feature" isn't a very interesting or rewarding way to help people. It's a blowoff that definitely doesn't make the place more helpful and "nice" to new users.

If I feel like answering the same question 100 times in 100 different threads, I'll do it.

Paul
 
Just to add, for anyone interested in more information about the RAMDisk I'm speaking of, you can find it here:

SuperSpeed RAMDisk

and here's a FAQ page with more information here:

SuperSpeed RAMDisk FAQ

and...

Since I am always happy to admit when I'm mistaken, it turns out that the next version of RAMDisk (version 8) does allow you to mount a pagefile on a RAMDisk. Will wonders never cease.

Once it's released, I'll do testing on it over several configurations and see if actually putting the pagefile on a RAMDisk will help with performance.

Paul
 
Not that I'm advocating a pagefile in RAM, but in some instances the VMM won't deal with things properly if no pagefile whatsoever exists. Putting a pagefile in RAM, while redundant at the conceptual level, could conceivably provide a workaround to a situation where the VMM simply insists upon having a pagefile.

That's about the only reason I can see why you'd ever use a RAM based pagefile, and that's theorectical at best.

And as we all know, opinions vary widely on tweaks. Opinions also vary widely on those offering advice on such.
 
Yep.

One major application that requires a pagefile is Photoshop. Go figure.

:D

It'll be interesting to see what putting a pagefile on a RAMDisk has in regards to performance with Photoshop and some other RAM hog software.

We'll see.

Paul
 
Putting the pagefile on a ramdrive is just asking for trouble. Many reasons but I will start with the number one reason right here:
ADDRESS SPACE

Address space is already at a premium and you want to take some away to speed up your machine? I guarantee your machine will run slower if you do that.

If you want to remove components from windows, there is a way to do it without relying on third party hacks.
http://msdn.microsoft.com/embedded/windowsxpembedded/default.aspx

Microsoft's supported way of removing components from windows.

I think everybody knows my opinion on tweaking, which is leave it at the factory defaults, unless you have a good reason to change it. ;)
 
If everyone ran stock installations of Windows, it would be like everyone owning Yugos. No spark, no creativity, useless dull and boring.

Of course, not everyone has a dual monster boxen like you do, Ranma_Sao. :D

People do this stuff because they've always done it, just like people have always tried to drive faster, run faster, go farther, do things no one else can or will do.

Nothing wrong with experimentation. Besides, your machine isn't as fast as it could be, even with those specs.

Paul
 
my solution to sawp space speed is quite simple... address the swap space to a drive other than your boot and app disks/ i have a disk that this machine boots from and a app/games raid array(0), and a seconed array(0) that right now has nothing but swap space on it (it is set up so i can dump files to is incase one of my other drives takes a shit, just turn off the swap and start dumping to it ) going form a singeil disk set up to a 2 disk setup and addressing the swap to the seconed disk is simply amazeing , almost like being on a 0 stripe, then going to a 0 stripe with a 0 stripe swap... talking about fast... granted its no replacement form ram, but haveing the extra space for security was worth it.. (i've had to use it in the last year... )
(/e)
oh.. and if your running a single disk setup create a partition at the outside of the disk about 2gb in size format it with a 64kb block size (ntfs) and set the swap to that while access times may go up a littel the data though put will be alot faster once it finds it (great for games like HL2 where it starts swapping images back out of ram )

thore
 
Having the pagefile (whatever size you have) on a second drive on a second controller is always the best solution, but yours, thore, is the cream of the crop. Getting the pagefile set up on a RAID would increase performance just that much more, but of course not everyone has RAID.

Cool...

Paul
 
hey... we'er talking high end here... while my processor and gpu are lagging ATM, getting a littel proformance out of it by addind 2 80$ HD's is worth it.

thore
 
br0adband said:
If everyone ran stock installations of Windows, it would be like everyone owning Yugos. No spark, no creativity, useless dull and boring.

Of course, not everyone has a dual monster boxen like you do, Ranma_Sao. :D

People do this stuff because they've always done it, just like people have always tried to drive faster, run faster, go farther, do things no one else can or will do.

Nothing wrong with experimentation. Besides, your machine isn't as fast as it could be, even with those specs.

Paul
Click to expand...
It could be faster I'm sure, but there's something to be said coming home from trying to break software at work, and just having it run solid. ;)

Again, I don't really care what people do to their own boxes, just advising best practices.

And thanks for the comments about my computer, just got it upgraded for my birthday, was a dual 1.2 with GeForce Ti4600. The ram was a gift from a person I did work for, he bought 4 1 GB registered sticks not realizing his mobo wouldn't take em. Decided it was easier to give them to me then try and return them on the internet. ;)
 
Decided it was easier to give them to me then try and return them on the internet.
Click to expand...

So who decided this, him or you with a lot of persuasion?

HAHAHAHA

There are times when I'm working on this damned PC and I get so sick of tweaking, tuning, etc. I gave it up a long time ago because I just got that feeling... most people know what I'm talking about.

In the movie "Real Genius" near the middle there is a scene where a bunch of the college kids are sitting in the day room area of the dorm, bodies scattered everywhere in chairs, on the couch, the floor, etc. Books everywhere, it's a total mess.

One kid sitting at the table puts his pencil down and then starts screaming, louder and louder. He rises from the table, grabs his hair and pulls it, starts screaming even louder. Then he jets out of the room, running down the hall outside into the courtyard still screaming his lungs out.

One kid just gets off the floor and takes his place at the table.

That is the feeling I'm talking about. That you just can't do anything else with the god damned hardware and yet you can't stop yourself.

Bleh... I need a new hobby. Besides, the fiancee is a geek too so that's not gonna work well once we get married.

hehehe (Hope she doesn't read this)

Paul
 
I use tunexp, make sure my msconfig startup files are relatively clean, and I personally do serious trimming of services, imo at least indexing and system restore need to be removed. Tunexp does quite a bit for me, and I like it overall.
 
You cant just disable the PF, windows pages unused processes to free up more RAM so I agree with phoenix as well. Also as people said alot of programs such as photoshop require a swap file(i think adobe call it the scratch disk) but you still need enough physical memory, a pagefile is no substitute to REAL memory. As for the RAM disk issue the 4GB barrier doesnt give enough free RAM for it to be useful, if you want a RAM disk get a proper RAM drive.
 
br0adband said:
I've been tweaking Windows since most people were in diapers, so yes, these tweaks are tested thousands of times across thousands of machines in my 25+ year career.
Click to expand...
Ohh, and e-penis comeptition!!! :rolleyes:

FYI, windows hasn't been out for 25 years.

Please note I didn't suggest making a pagefile on a RAMDisk, I said if you happened to get it working because some people that use a RAMDisk automatically assume, "Hey, if RAM is faster than a hard drive, and a RAMDisk is faster than a hard drive, logically putting my pagefile onto a RAMDisk will boost my performance right into orbit!"
Click to expand...
I never said you advocated it, I just said if you wanted the same effect (all paging in physical RAM) you simply need to disable the HDDs PF. If you understood why the PF exists you'd realize trying to put it on a RAM disk would only create MORE overhead then simply not allowing caching to the HDD in the first place. The very fact you suggest this tells me you don't grasp paging at all.

Tweaking services can always help to some degree, but no, it won't magically make your PC 50% faster at Half-Life 2 because you disabled a few unneeded ones. Never said that, never even implied it.
Click to expand...
OK, I'll bite. What does tweaking services do outside limit the functionality of your machine? Again you don't understand paging, which is why you think this is a "good" tweak. You and your 25 years of experience must be able to provide some graph or something to show some increased performance. Don't tell me you have been spending all this time "tweaking" and don't know the results...

Disabling services (especially on older machines) has benefits of increased reliability and stability in the long run. It's a simple fact: the less you have going on, the less chance you have of something going wrong.
Click to expand...
Negative!!! You have less running so you have *increased* chances of failure. Dependant services won't start, new functionality fails, errors ensue. Proof is in the search. Search "quackviper" and see numerous threads where disabled services=errors. Again, what benefit do you see?

As for why it's been discussed to death before, that's simple: new people join this and other forums daily, and just using a blanket statement like "Oh for fuck's sake, just use the Search feature" isn't a very interesting or rewarding way to help people. It's a blowoff that definitely doesn't make the place more helpful and "nice" to new users.

If I feel like answering the same question 100 times in 100 different threads, I'll do it.

Paul
Click to expand...
If you have something new to add by all means let's discuss it. I'd just encourage searching on this topic so you have a fair understanding of things like why disabling services is bad advice.

 
It sounds to me like br0adband knows what he's talking about and I appreciate his advice. I'm going to try nlite and RAMdisk before I crap all over the guy.
 
br0adband said:
Just to add, for anyone interested in more information about the RAMDisk I'm speaking of, you can find it here:

SuperSpeed RAMDisk

and here's a FAQ page with more information here:

SuperSpeed RAMDisk FAQ

and...

Since I am always happy to admit when I'm mistaken, it turns out that the next version of RAMDisk (version 8) does allow you to mount a pagefile on a RAMDisk. Will wonders never cease.

Once it's released, I'll do testing on it over several configurations and see if actually putting the pagefile on a RAMDisk will help with performance.

Paul
Click to expand...

Excellent info br0adband. Thanks for the links.


Phoenix86 said:
Ohh, and e-penis comeptition!!! :rolleyes:..........
Click to expand...
It seems there are not any threads with the word "pagefile" or "services" in this whole forum, that you don't try to belittle or crap on differing opinions. Chill, man. You got an opinion and experience, and br0adband has an opinion and experience.
 
Badger_sly said:
Excellent info br0adband. Thanks for the links.

It seems there are not any threads with the word "pagefile" or "services" in this whole forum, that you don't try to belittle or crap on differing opinions. Chill, man. You got an opinion and experience, and br0adband has an opinion and experience.
Click to expand...
I don't go around waving my experience as a flag to bow to. His comments about using windows when "most of us" were in diapers and how 25+ years exp makes his the autohority are simply irrelevant, inflamatory, and in several people's case very incorrect. Case in point, I really started learning PCs in the DOS era, pre-windows. Anyways, I'm just curious to see rather have a pissing contest or actually discuss some details. We'll see where it goes from there.

rcolbert, thanks for your insight. That'll further this thread leaps and bounds.

 
Phoenix86 - you and your two compadres simply seem to have taken the wrong voodoo class and are interjecting fallacious thoughts regarding services throughout this forum. It is clear to me that the three of you do not understand what a service is, nor do you understand what virtual memory and paging is and isn't, nor do you understand the distinction between the two (virtual memory and the pagefile.)

The problem is that you (collectively) are making personal attacks against people who are trying without success to impart a small modicum (redundant) of moderation into your assertions.

Lay off on people who want to adjust services. It does make a whole lot of sense in some cases. I have provided a wealth of information directly from Microsoft on the subject and it seems to have been ignored. It really seems like the message you're sending is coming from someone with experience in both time and depth that is relatively minor in comparison to sources that would be considered an authority on the subject. What are your credentials that make your advice on par with the MS insiders? (not speaking of myself) I'm talking about those who provide extensive research and advice to the contrary and publish such advice on Microsoft.com, or in Gartner studies, or in whitepapers and books, etc.

Bottom line is you sound like a 20-something who thinks he knows everything, or you sound like the "one" IT guy for a small company who acts like a real snot around everyone.

I wouldn't argue except I'm afraid of the impact that your bad advice has.

I recall listening to several assertions by instructors back in the 80's and early 90's about things that were strikingly similar in their detachment from the way things really work. I'll never forget those remarks because they formed faulty premises that took years for me to uncover as untrue. All I'm asking is that you either qualify your statements with appropriate degree, or you stop making very broad technical generalizations and arguing them as absolute truth. Sometimes is not always. And many is not all.
 
rcolbert said:
Phoenix86 - you and your two compadres simply seem to have taken the wrong voodoo class and are interjecting fallacious thoughts regarding services throughout this forum. It is clear to me that the three of you do not understand what a service is, nor do you understand what virtual memory and paging is and isn't, nor do you understand the distinction between the two (virtual memory and the pagefile.)
Click to expand...
*sigh* You still haven't responded to
this. You can toss accuations or voodoo all you want, put up or shut up. Also, since your new, you don't know that a lot of what I'm saying has been agreed on by many members of the forums, not 1 or 2 people...

The problem is that you (collectively) are making personal attacks against people who are trying without success to impart a small modicum (redundant) of moderation into your assertions.
Click to expand...
I don't make personal attacks, please show me where I have and I'll correct it. That's not my style.

Lay off on people who want to adjust services. It does make a whole lot of sense in some cases. I have provided a wealth of information directly from Microsoft on the subject and it seems to have been ignored. It really seems like the message you're sending is coming from someone with experience in both time and depth that is relatively minor in comparison to sources that would be considered an authority on the subject. What are your credentials that make your advice on par with the MS insiders? (not speaking of myself) I'm talking about those who provide extensive research and advice to the contrary and publish such advice on Microsoft.com, or in Gartner studies, or in whitepapers and books, etc.
Click to expand...
Provided a wealth of what? Go read that thread again, you were asked for specific information and have yet to respond. Also take not of Ranma_Sao's comments. He's a developer with MS on the Desktop team I believe. He knows his stuff.

Bottom line is you sound like a 20-something who thinks he knows everything, or you sound like the "one" IT guy for a small company who acts like a real snot around everyone.

I wouldn't argue except I'm afraid of the impact that your bad advice has.

I recall listening to several assertions by instructors back in the 80's and early 90's about things that were strikingly similar in their detachment from the way things really work. I'll never forget those remarks because they formed faulty premises that took years for me to uncover as untrue. All I'm asking is that you either qualify your statements with appropriate degree, or you stop making very broad technical generalizations and arguing them as absolute truth. Sometimes is not always. And many is not all.
Click to expand...
Funny, I ask you for evidence in another thread, you ignore the request and now you want me to justify something. Cute. Anyways, what would you like me to show you?

BTW, your taking personal shots here, not me. Keep it to the techincal details and stop the ad hominems, mmkay?

What bad advice am I giving? Pray tell...

 
Why oh why is this coming up again?

1) First, use the search feature, this has been discussed ad nauseum.
2) Argue if you want with Phoenix86, but your probably going to come out on the losing end.
3) Ask Ranma Sao what he does for a living....and then see if your still going to argue with him.
4) FOR THE LOVE OF GOD: Can we stop the damn e-wang competitions???? I worked in my current job for 4.5 years and I can run circles around people who have been there for 25+. None of that means anything at all....not the degree someone has, not the amount of time somone has done a job, etc. This isn't a union job, where years of service = knowledge and ability.
 
Well, I guess since I don't have my own "sticky" about pagefiles around here I can't know anything and that means you, because you do have one, makes you the HardForum "Expert on Pagefiles and Virtual Memory emeritus."

Bleh. No wonder this place gets the rep it does. Give some people some information and it turns into a "my dick is bigger than your dick and my PC is faster too" contest.

Whatever.

The information I provided is accurate, and easily verifiable if you're the kind of person that likes to do such things. I've installed Windows operating systems over 150,000 times in my years, so I'd like to think I know something about how they work.

As for the fallacy of "creating more overhead by putting a pagefile on a RAMDisk," well... consider that RAM is a few hundred thousand times faster than a physical hard drive. Consider that even a massive performance hit on a physical hard drive would barely be noticeable when translated to a RAMDisk, and you start to see why I'm getting interested in testing that ability of the new RAMDisk software that allows me to put a pagefile on a RAMDisk.

Let's say you have 2GB of physical RAM. You create a 512MB (identical minimum/maximum) pagefile on the hard drive just for kicks and do some testing on it. You fire up some applications, possibly some benchmarking suites like BAPCO's SysMark utility, the *almost* industry standard for such testing. You get your specs and your test results and you're done.

Now imagine using a RAMDisk program that will chunk off 512MB of your physical RAM and turn it into a usable RAM-powered pagefile. At this point, the usable OS code is in RAM, the applications you're running are in RAM and the Virtual Memory subsystem is in RAM, and the hard drive would only be touched to load new code or data files. You redo the testing procedures you did for the first configuration and when all is said and done, you compare the results.

Can you really say that in the second configuration using the RAMDisk as a pagefile (when it's possible, meaning when the newer version of the RAMDisk software is released), that this system would perform slower than the first configuration?

Really? You believe that? I find that hard to believe, but...

I test operating systems and software not only as a hobby, but to make a few bucks now and then. I'm not the casual hobbyist, not by any person's definition of it.

SuperSpeed has done thousands of test situations in their years of putting out caching and RAMDisk software. They're recognized as an industry leader in the system performance area and eeking out every single ounce of power from a system that can possibly be eeked out (I think that makes sense, just not sure about "eek" being a respectable word hehe). I trust their software since I've used it for years in not only corporate situations but on my own personal hardware. It works, simple.

But I digress. I'd hate to think I'm perpetuating the "my dick" thing. Best to let sleeping dogs lie where they may.

As for the BlackViper thing, he knows his stuff backwards and forwards. Limiting functionality? Well lemme see here, I don't need Remote Registry, no one needs Messenger which is finally disabled by default, Secondary Logon is pointless to most people, Windows Time - I have a watch, thanks, lots of stuff in services that are pointless and effectively useless little applications consuming CPU time and resources.

Listen, to each his own. I don't know everything, but I do know this much: neither do you.

Paul
 
Phoenix - We can argue all you want about the contents of the services.msc thread which was deleted. There was a whole lot of information in there, and much of it relevant to the discussion at hand. As I recall, it was folks arguing your position (not you) who were making all the attacks and likely the reason the thread is gone.

(regarding our other disagreement) It's a simple premise really.

First, since a service is a program, and can be pretty much any program, how can you quantify the impact of that program running as being negligible with sweeping statements. All services are as individual as the applications in your start menu. I just don't follow the logic there. Hell, I could run leakyapp.exe as a service if I so chose. Being a service doesn't make a program any more or less resource intensive. A service is defined by its function, not by the fact that it's being run as a service. I have yet to hear any valid contradiction to that assertion.

Second, any running process is potential surface area for attack. Period. How is it that this is even refutable? Microsoft has a stated security approach to reduce surface area by reducing the number of running services and processes to the minimum required. Why would service controls be included in security templates if services were in no way related to the security of a system?

Third, any running process must consume some small amount of system resource at a machine level, even if it's just an extra few cycles for the kernel to check a condition related to the process. Without any CPU cycles dedicated to it, a running process would be effectively dead an never run another cycle again until the process was killed and restarted. This is extremely minor, but the way you describe a dormant process as consuming zero CPU cycles give people a fundamental misconception about how a computer works. You are describing activity at an abstraction level higher up, and I am describing activity at a lower level. What is important here is that people who are learning these things for the first time understand that any program waiting for input must have a mechanism to periodically check for the input. The input could be a connection attempt, an interval of time has passed, a control request to shutdown has occured, or a million other things. I have never stated that this has an appreciable impact on performance. What I am saying is that it is a disservice to the folks in this forum to describe computer behavior in that way. If you ask an application developer, a systems engineer, and a chip designer the same question, you'll likely get three entirely different points of view. What I am saying is that your point of view seems to equate to the application developer view of things. There are two fundamentally different viewpoints, and all I have ever requested was that more qualified statements be made.

As to an appeal to authority, I have plenty of resources available to me that I consider authoritative. I have personally known the author of a very successful series of books from Microsoft Press for many years, and in fact a contributing author of serveral of those books is sitting about 15 feet from where I am at this moment. The cubicle across the aisle from where I sit is a hotel cube for Microsoft employees, and I have access to various MS technical folks day in and day out. I have spent a good deal of time in Redmond for various briefings and discussions. Although I will make no claim to authority myself, I can demonstrate easily that I have access to folks who are in many cases the premier authority on any given subject.

As to personal attacks, I apologize, The precedent was set by others with a similar viewpoint to yours, and I failed to make the distinction. Although I must note your sarcasm is alive and well.

As to being new here, well, whatever. My main interest here has been as a hobbyist, not an IT professional. In fact, I don't consider this forum to be a source of professional discourse. Truth be told, when I'm here I'm basically just goofing off. It's not as if I decided to start working in IT the same day I signed up for hardforum. Would it help if I whipped out my e-whang?
 
br0adband said:
Well, I guess since I don't have my own "sticky" about pagefiles around here I can't know anything and that means you, because you do have one, makes you the HardForum "Expert on Pagefiles and Virtual Memory emeritus."

Bleh. No wonder this place gets the rep it does. Give some people some information and it turns into a "my dick is bigger than your dick and my PC is faster too" contest.

Whatever.

The information I provided is accurate, and easily verifiable if you're the kind of person that likes to do such things. I've installed Windows operating systems over 150,000 times in my years, so I'd like to think I know something about how they work.
Click to expand...
How about this. Stick to the technical discussion and leave your "I got 25+ years exp" at the door. I don't wave my experience around, no need for you to either. No matter what your expereince you (and I) can be wrong.

As for the fallacy of "creating more overhead by putting a pagefile on a RAMDisk," well... consider that RAM is a few hundred thousand times faster than a physical hard drive. Consider that even a massive performance hit on a physical hard drive would barely be noticeable when translated to a RAMDisk, and you start to see why I'm getting interested in testing that ability of the new RAMDisk software that allows me to put a pagefile on a RAMDisk.
Click to expand...
Yes, I see the benefits of what you want to do. However, your missing some key concepts as they relate to XP, which *isn't* W9X. The first thing your assuming is that paging to disk happens a lot. Well, if you have enough RAM to try to put your PF on a RAMDisk, you aren't paging to *disk* much at all. So if you were to move this paging i/o from disk to RAM, great, but your system will only benefit *IF* your paging to disk a lot, and I have test this quite a bit. As long as you system as RAM to spare (IE your commit charge isn't approaching physical RAM total) you simply don't write to the PF much, if ever at all.

Second, your *still* are not reading what I'm saying about paging. The page file is a supplement to RAM as a paging space. When you don't have enough RAM the system will page that data to disk, aka the PF. What you want to do by putting the PF on a RAMDrive is to force all paging into RAM. Yes I see the benefits of that, and so does the OS for the most part* (there are some cases where the system will page a lot, even with RAM free, but only as it's approaching the max), that's why it doesn't page to disk unless it's necessary, IE RAM is getting full.

As for the BlackViper thing, he knows his stuff backwards and forwards. Limiting functionality? Well lemme see here, I don't need Remote Registry, no one needs Messenger which is finally disabled by default, Secondary Logon is pointless to most people, Windows Time - I have a watch, thanks, lots of stuff in services that are pointless and effectively useless little applications consuming CPU time and resources.

Listen, to each his own. I don't know everything, but I do know this much: neither do you.

Paul
Click to expand...
As for services, again, show me how disabling services= increased performance. Don't blow a bunch of smoke up our asses about how someone knows their stuff. All your doing is appealing to the authority. Prove what your saying, that these resources are being taken by the OS. Also read the thread I linked to where Ranma_Sao talks about services and memory/CPU usage.

 
I feel this is an appropriate thread for it's debut...

ewang.jpg
 
rcolbert said:
Phoenix - We can argue all you want about the contents of the services.msc thread which was deleted. There was a whole lot of information in there, and much of it relevant to the discussion at hand. As I recall, it was folks arguing your position (not you) who were making all the attacks and likely the reason the thread is gone.
Click to expand...
Umm, no. Threads don't get deleted, it's still open and I'm awaiting your reply.


First, since a service is a program, and can be pretty much any program, how can you quantify the impact of that program running as being negligible with sweeping statements. All services are as individual as the applications in your start menu. I just don't follow the logic there. Hell, I could run leakyapp.exe as a service if I so chose. Being a service doesn't make a program any more or less resource intensive. A service is defined by its function, not by the fact that it's being run as a service. I have yet to hear any valid contradiction to that assertion.
Click to expand...
Read the replies in the thread I linked to above, esp not Ranma_Sao's comments. I'll not duplicate discussions going on in each thread, so I won't reply to the rest of this post as it's an extension of the other thread.

As to personal attacks, I apologize, The precedent was set by others with a similar viewpoint to yours, and I failed to make the distinction. Although I must note your sarcasm is alive and well.
Click to expand...
Sarcasm is different than a personal attack. I'll be the first to admit to sarcasm. :p Apology accepted.

As to being new here, well, whatever. My main interest here has been as a hobbyist, not an IT professional. In fact, I don't consider this forum to be a source of professional discourse. Truth be told, when I'm here I'm basically just goofing off. It's not as if I decided to start working in IT the same day I signed up for hardforum. Would it help if I whipped out my e-whang?
Click to expand...
As far as being new, I was just noting how you weren't here for past threads, nothing more. There are several members of the forum who agree with what I'm saying, not 1 or 2.

 
djnes: (fixed... yikes!!! I'm going blind!!!) :)

A moment of brevity, just what we needed.

We've all seen this shit before, so many times all of us would just love it if things just disappeared.

I get it, I know all of you do. In my experience, I have done this simple test.

I installed XP Pro SP2 recently on this laptop of mine, here are the specs:

IBM ThinkPad A21m
Pentium 3 800 MHz
256MB PC66 SDRAM
20GB 4200 rpm Hitachi hard drive (ATA33 interface on the Intel chipset, 440BX)

Since this laptop is approximately 4 years old, the hardware it's made with is well established and SP2 has all necessary drivers to ensure functionality across the board, so no newer drivers were necessary for installation.

As far as performance is concerned, those are the only specs that really matter for testing purposes since those 3 components essentially make the computer what it is.

The hard drive is partitioned as ~18GB for the system drive C: and then the remaining ~2GB is assigned as space to store images using True Image D:.

I did the default install of XP Pro SP2 on the laptop then installed BAPCO SysMark 2002 - that's the last version I could actually pay for but I'd love to get the newest one sometime.

I left all XP services at defaults and made no adjustments to the pagefile settings or anything else. I rebooted the computer 5 times in sequence so as to ensure it's working properly and to get some readings on the memory usage when booting to the Desktop.

The first boot was to 112MB according to Task Manager. Mind you this isn't perfect science here, so please don't start hammering away on "you didn't do this" or "you didn't do that" - this is informal, spur of the moment testing methodology to make a point.

The second boot was 110MB when all settled down. The third was 105MB, the fourth was 98MB (I attribute this to the fact that after the 3rd boot, XP begins its self-tuning scheme and the "Windows Tour" bubble/popup doesn't happen again, both things consuming slight resources), and finally the fifth and final pre-testing boot ended up with 92MB of physical RAM usage as defined by Task Manager - also note that running Task Manager consumes a few MB when it's running.

I defragged the drive using the XP default defragger and when that was done, I used True Image to make an image of the system and stored that image on the image partition.

I rebooted the machine and went right into the SysMark suite and let it run until it completed then I made note of the findings.

I rebooted the PC using the True Image CD to restore the image I had created earlier. As soon as this was done, I rebooted again, went directly into the services and started trimming the fat, as I see it. Checking the RAM usage, it leveled at about 92.5MB roughly, almost exactly where it was when I created the image.

The following services were set to disabled and I'll also note their current status before being disabled:

Ati Hotkey Poller (this laptop has a Rage Mobility video chip in it, and I don't use the hotkey feature) (Automatic)
Automatic Updates (I don't use this feature) (Automatic)
Background Intelligent Transfer Service (I don't use this feature) (Manual)
Indexing Service (I don't use this feature) (Automatic)
Infrared Monitor (I don't use this feature) (Automatic)
Print Spooler (I don't use this feature - no printer) (Automatic)
Protected Storage (I don't use this feature) (Automatic)
Remote Registry (I don't use this feature) (Automatic)
Secondary Logon (I don't use this feature) (Automatic)
Security Center (I don't use this feature) (Automatic)
SSDP Discovery Service (I don't use this feature) (Automatic)
TCP/IP NetBIOS Helper (I don't use this feature) (Automatic)
Webclient (I don't use this feature) (Automatic)
Windows Time (I don't use this feature) (Automatic)
Wireless Zero Configuration (I use this, but not during this testing procedure) (Automatic)

Some will wonder why I disable certain features, and that's fine - again, to each his own. On a laptop, some things just aren't needed, and actually, once I get everything up and running to my likings, there are more services that get trimmed/disabled, like IPSEC stuff and the Windows User Mode Driver Framework that Windows Media Player 10 installs and sets on Automatic. But let's move on.

After these services were disabled I noted the RAM usage was down to 73MB at that point, having done nothing else but disable the above mentioned services and then take a peek at Task Manager.

I rebooted once more and upon reaching the Desktop and waiting for hard drive activity to cease, I ran Task Manager and was looking at 67MB of RAM usage.

I re-ran the SysMark suite and took note of the results.

I then rebooted the PC and decided to modify the pagefile settings at that point. I changed it from the default of 384-768MB to 512MB min/max for a static pagefile and then rebooted.

Checking the RAM usage again after the reboot to the Desktop it was 65MB.

I fired up SysMark once more and did another suite. Mind you this took the better part of most of a day to do all this; thank god I have other hobbies.

:D

In the end, the results were clear - at least to me.

The default XP Pro SP2 install gave me a baseline score for comparison: 72.

After the first configuration change (disabling unneeded services) the score rose 6.8% to 77.

After the second configuration change the score rose 2.1% over the second test: 78.

I'm not sure about all of you, but that tells me disabling the services had a noticeable affect on the performance of this old workhorse laptop of mine; setting a static pagefile improved the performance negligibly I admit, but it did improve it.

To be honest, if I had to pinpoint where and why the scores differed, it was more than likely because of the lower RAM usage by the OS to get started and therefore leaving more RAM for the benchmark suite - this would help with the paging activity which obviously would be detrimental to the score also. Trimming the fat off the services left more usable RAM for the applications in SysMark equating to a slight boost in performance.

Tuning the pagefile in the third run through only made that negligible improvement probably because XP didn't have to resize the pagefile during the testing procedure, which I am quite confident the benchmark suite does. I doubt very seriously it stayed stuck at 384MB (the minimum system managed size).

Of course, in today's age of multigigahertz machines, a few percentage points doesn't matter to most people. I never intended to sway the entire world to my way of thinking about disabling services or modifying the Virtual Memory parameters of modern Windows operating systems.

All I'm saying is: it's all relative. It all relates. Disabling services can and does have a measurable effect on system performance, but it's not for every person or user. It can get outright dangerous if you don't know what you're doing, if you don't take the time to consider the results of your experimentation, and you don't research the outcome experienced by others that have already walked the path.

Hobbyists and hackers made the world what it is today, not bookworms and librarians (not to knock bookworms and librarians, mind you). It was forged by the people that kept saying, "What if..." and "Why the hell not..."

We all have something to contribute around here in many ways. As Pheonix86 says:

YMMV.

Paul
 
You must log in or register to reply here.
Back
Top