Performance-oriented Windows tweaking

[Fresh Daemon]

Wait, these things turn off?

I've heard it's so! But any real user knows that night-time is Torrent-time/virus & spyware scan time/update time/defrag time/folding time/XviD conversion time...

 

[SumYungGai]

Originally posted by: Fresh Daemon

Granted, the available memory is increased. What I've seen in all these tests is that there isn't any increase in performance because of this. The reasons for this have been gone over in this thread, but basically, Windows will page out an unused service so although it appears that more RAM is freed up, you aren't actually gaining anything since Windows would have freed the RAM anyway.

It's not like the process just magically goes away. Windows isn't that efficient. While it may be minimal, there's still going to be some memory being taken up. Many PCs, like the many built in 2001, that only have 128MB of RAM would greatly benefit from any increase is available memory.

This is why the benchmarks before and after tweaking, even when all available RAM is used up, don't show any real difference.

I doubt these synthetic tests would be accurate, especially since they aren't geared toward this specific variable.

All that these tweaks do is remove the flexibility from the system, for instance, if you ever needed one of those services, Windows could load it into memory and it'd work. OTOH, if you'd disabled it, then whatever app needed that service just wouldn't run.

Sometimes that's the point; you don't want the app to run. Some of the functionality you're enabling would only be given to some remote attacker.

Many of the services are for specific things, that the average user wouldn't ever use, or would only be needed with some kind of harware change or upgrade.

Also, most of the services you could set to Manual, so then IF you ever need it (or when you do), it would only be started when it is required. If I restarted my PC more often than using the Logical Disk Manager, I would benefit from setting that service to Manual.

That may well be true, but I don't find that to be any useful metric for computer performance anyway. Macs still sell despite the fact that MacOS X takes forever to start compared to XP. So does Linux. Most people here (and most computer users in general) boot and shut down their computers once a day, if that.

Now, I'm not too sure about how this is affected, but what if it was a multi-user environment, where people are logging on and off? Whatever processes are loaded then, would certainly make a difference there, too.

 

[imported_BikeDude]

Originally posted by: SumYungGai
It's not like the process just magically goes away. Windows isn't that efficient. While it may be minimal, there's still going to be some memory being taken up.

For what? Given a process where all threads are blocked waiting for an OS function to return (be it WaitForSingleObject() waiting for a semaphore or a Read() from a socket, etc...) -- just exactly which parts of that process need to be kept in physical memory?

That said, I actually agree. XP fires up too many services that will simply go to rot. That's part of the reason why I prefer 2003 Server on my rigs -- less services to disable. (most are already disabled or set to manual) I certainly do not want indexing service to fire up willy-nilly. But from an increase-performance POV (rather than a neatness POV), I just don't see the benefit.

 

[Fresh Daemon]

Many PCs, like the many built in 2001, that only have 128MB of RAM would greatly benefit from any increase is available memory.

I tried this on a PC from 2000 with 160MB of memory and saw no improvement.

It's not like the process just magically goes away. Windows isn't that efficient. While it may be minimal, there's still going to be some memory being taken up.

Why? Do you know this or are you just guessing that that's how it would work? Computers are very counter-intuitive, probably because the science they rest upon (quantum mechanics) is very counter-intuitive.

I doubt these synthetic tests would be accurate, especially since they aren't geared toward this specific variable.

I don't believe in running artificial benchmarks designed to test only one aspect of your system. If you use your computer to run artificial memory/disk benchmarks then I can't help you. BV's site talks a lot about gaming, so I tested it mostly with games, and saw no improvement.

Sometimes that's the point; you don't want the app to run. Some of the functionality you're enabling would only be given to some remote attacker.

But this isn't called Security-oriented Windows Tweaking, it's called Performance-oriented Windows Tweaking. If you want to discuss how services affect security, go run your own tests and start your own thread.

Now, I'm not too sure about how this is affected, but what if it was a multi-user environment, where people are logging on and off?

Multi-user servers are beyond the scope of this article, and neither BV nor myself claim that this is applicable to networked environments.

Your criticisms seem to be aimed at the fact that this article does not cover things it never claimed to.

 

[SumYungGai]

Originally posted by: BikeDude

For what? Given a process where all threads are blocked waiting for an OS function to return (be it WaitForSingleObject() waiting for a semaphore or a Read() from a socket, etc...) -- just exactly which parts of that process need to be kept in physical memory?

I don't know, I'm not sure of the details. But I've never seen a string of processes only taking up merely 1K each.

Originally posted by: Fresh Daemon

I tried this on a PC from 2000 with 160MB of memory and saw no improvement.

Okay, so you can't see the numerical difference between 28 and 60? I don't care about 160MB. Most don't have that amount (ie, it's non-standard), which is incredibly better for XP than 128.

That's like saying you wouldn't see a difference AT ALL by adding more memory when it's the main bottleneck. If the OS has to page itself into a swapfile, it would surely make much more of a difference to free any amount of RAM. And without a paging file, a freshly booted XP can take nearly 200MB.

Why? Do you know this or are you just guessing that that's how it would work? Computers are very counter-intuitive, probably because the science they rest upon (quantum mechanics) is very counter-intuitive.

Uhh, what? I don't see how any of this related to the topic. It isn't an argument to talk about things that are totally unrelated, and your pseudo-hypothesis is exactly that. Unless you've seen all of those seemingly unused processes take up only a few kilobytes collectivly, you are simply wrong to think it will end up that way.

And it baffles me how you think being counter-intuitive makes you appear correct, or even myself incorrect. That in itself isn't intuitive at all.

I don't believe in running artificial benchmarks designed to test only one aspect of your system. If you use your computer to run artificial memory/disk benchmarks then I can't help you.

But the thing is, those benchmarks are only testing a few variables; not all of them, or even most. It could be completely overlooking the variables we really would like to know, or could be not extensive enough to show anything definitive.

BV's site talks a lot about gaming, so I tested it mostly with games, and saw no improvement.

Since services usually don't take up much of anything when unused, and little paging happens when the program is already in use, it just doesn't make sense to bother with a single application, or one that merely does a limited action.

But this isn't called Security-oriented Windows Tweaking, it's called Performance-oriented Windows Tweaking. If you want to discuss how services affect security, go run your own tests and start your own thread.

Uhh, you mentioned security yourself in the opening post. And part of your argument involves functionality. But it's rare that you would suddenly need to enable a service, and you can always enable it by going to the Services MMC. We aren't using 95 anymore, we don't need to reboot for simple changes. Plus, there's no reason to ignore security, it's an inherent topic for things like services.

Multi-user servers are beyond the scope of this article, and neither BV nor myself claim that this is applicable to networked environments.

It doesn't matter, I was responding to your recent comments. That would mean either you yourself were off-topic, or are just avoiding the issue. And I don't really see how it's "beyond the scope", when it's directly related to performance (yes, booting is part of it, it's a fundamental factor).

Your criticisms seem to be aimed at the fact that this article does not cover things it never claimed to.

No one even mentioned the article. Just what you've been saying in this thread.

 

[xtknight]

Originally posted by: Fresh Daemon

Wait, these things turn off?

I've heard it's so! But any real user knows that night-time is Torrent-time/virus & spyware scan time/update time/defrag time/folding time/XviD conversion time...

:thumbsup::laugh:

Edit: Why is there an empty page 10 after me?

 

[ghendricks]

Hi Fresh Daemon

Very nice work there. I never had time to try these out, but I should someday.

 

[Nothinman]

I don't know, I'm not sure of the details. But I've never seen a string of processes only taking up merely 1K each.

The statistics displayed by taskmgr and the like don't really change because of demand paging, in a lot of cases the OS doesn't know ahead of time if a page is in memory already or needs to be retried from disk, calculating that all of the time would kill performance.

Okay, so you can't see the numerical difference between 28 and 60? I don't care about 160MB. Most don't have that amount (ie, it's non-standard), which is incredibly better for XP than 128.

The point is that the less memory the machine has, the more affect tweaking should have on performance. If you notice virtually nothing with 128M you're going to see even less if you have more memory.

But the thing is, those benchmarks are only testing a few variables; not all of them, or even most. It could be completely overlooking the variables we really would like to know, or could be not extensive enough to show anything definitive.

But the thing is, benchmarks by their very nature only test a few specific variables. There's no real way to gauge how responsive a system 'feels' without just sitting down and using it for a few days or weeks. BV's tweak site specifically targets tweaks for gamers, so it makes sense to test with common games to see if the tweaks have any real affect on gaming.

 

[MysticX23]

WOW!!. All this time, I keep playing with services. Sometimes, I end up crashing my computer or things not working because of it. Now I know, not to even bother. Thanks!!!!

 

[SumYungGai]

Originally posted by: Nothinman

The point is that the less memory the machine has, the more affect tweaking should have on performance.

Also depends on the system setup and what program is being used. If I had more than the average amount of junk installed, and the OS was paging itself to simply open the start menu, I'm not going to be properly testing things like that with some other program that's already in memory.

And if you have a top of the line rig, then you wouldn't even have to think once about doing any tweaking. These things are mainly for just raising the your PC that much higher above the minimum requirements.

If you notice virtually nothing with 128M you're going to see even less if you have more memory.

Who said anything about noticing virtually nothing? And frankly, if don't see the difference in a 128MB system by making some of these changes, then you really aren't fit and are too slow to be testing this stuff.

But the thing is, those benchmarks are only testing a few variables; not all of them, or even most. It could be completely overlooking the variables we really would like to know, or could be not extensive enough to show anything definitive.

But the thing is, benchmarks by their very nature only test a few specific variables.

But the thing is, I just said that.

There's no real way to gauge how responsive a system 'feels' without just sitting down and using it for a few days or weeks.

Well there certainly could be other tests that could get at least slightly more definitive results. Most of these benchmarks are trying to get the most out the system, and it doesn't look like it would deliberately make actions that would hinder performance, even if though those things may be encountered in a real life workload.

BV's tweak site specifically targets tweaks for gamers, so it makes sense to test with common games to see if the tweaks have any real affect on gaming.

I don't care about what BV's site is for. You can't accurately test something in that way because it could be totally overlooking the variable. Hell, even something like load times would be much more realistic of a test for a game. Unless you're not meeting the minimum requirements for the game, something like page swapping probably won't be noticed when it's in use. And if that were the case, then maybe disabling a few service would actually make a difference.

 

[Fresh Daemon]

I don't know, I'm not sure of the details. But I've never seen a string of processes only taking up merely 1K each.

Right, well, when you know something you'll be qualified to talk about it, won't you?

In the meantime, you'll have to forgive me if I take my own hard evidence in preference to the conjecture and speculation of a self-confessed ignoramus.

Okay, so you can't see the numerical difference between 28 and 60? I don't care about 160MB. Most don't have that amount (ie, it's non-standard), which is incredibly better for XP than 128.

160MB is "incredibly better" than 128MB? Why?

Let me ask you this: if I pulled a couple of 16MB SIMMs out of System I, do you think my results would be very different between tweak levels, bearing in mind that the tests maxed out well beyond the physical RAM capacity anyway?

And if you think yes, then let me ask you what the possible use is of a set of tweaks for enthusiasts that lose all value somewhere between 128MB and 160MB of physical RAM, in an age when any enthusiast has at least 512MB?

Unless you've seen all of those seemingly unused processes take up only a few kilobytes collectivly, you are simply wrong to think it will end up that way.

I've given you my evidence and I've explained why it works that way. You've told me that you're taking guesses without any evidence whatsoever. You are being illogical, and it is that, rather than your failure to accept that things can be counter-intuitive, which makes you appear incorrect.

But the thing is, those benchmarks are only testing a few variables; not all of them, or even most. It could be completely overlooking the variables we really would like to know, or could be not extensive enough to show anything definitive.

I think you don't know what computers are used for! We don't run tests of variables on them, we run software which we use to get our work done or play games. It does not matter how greatly something produces an effect at a very low level if it doesn't affect actual programs.

Since services usually don't take up much of anything when unused, and little paging happens when the program is already in use, it just doesn't make sense to bother with a single application, or one that merely does a limited action.

Again, I'm not particularly interested in your guesswork. Test it yourself, you'll find that your results don't confirm your preconceptions.

Uhh, you mentioned security yourself in the opening post.

I mentioned security in passing because one of BV's tweaks represented a serious breach of security. I mentioned functionality in passing because one of his tweaks produced a serious impact on functionality.

This might have been useful information.

Plus, there's no reason to ignore security, it's an inherent topic for things like services.

There are loads of articles and guides on Windows security and services. There are no scientific ones on performance-oriented Windows tweaking. Bearing that in mind, I tested the latter, not the former.

If you want information on security and Windows services, just Google it.

And I don't really see how it's "beyond the scope", when it's directly related to performance (yes, booting is part of it, it's a fundamental factor).

It's beyond the scope in the same way that strapping rocket boosters to a car is beyond the scope of a performance tuning magazine, although rockets are technically a performance-increasing modification like new heads or spray.

No one even mentioned the article. Just what you've been saying in this thread.

The thread is about the article. :roll:

And frankly, if don't see the difference in a 128MB system by making some of these changes, then you really aren't fit and are too slow to be testing this stuff.

This is a really stupid thing to say. Are you telling me you can see a 0.3fps difference, and that people who don't are unfit and too slow? Who are you, Spiderman?

Hell, even something like load times would be much more realistic of a test for a game.

I dislike things that I would have to time with a stopwatch. My own human eyes and hands are too fallible to produce reliable readings. I may do it later anyway, but take my results with a huge pinch of salt.

WOW!!. All this time, I keep playing with services. Sometimes, I end up crashing my computer or things not working because of it. Now I know, not to even bother. Thanks!!!!

Well, you just made all of this work worth it! I managed to save one person from instability in the name of unrealizable performance gains. Kudos to you, mate.

Very nice work there. I never had time to try these out, but I should someday.

You're welcome. If you measure anything, let me know, I'm curious to see other results.

 

[karstenanderson]

what about using a ramdisk mapped as a drive letter as a pagefile? i did that for awhile and it seemed to work alright

 

[Nothinman]

what about using a ramdisk mapped as a drive letter as a pagefile? i did that for awhile and it seemed to work alright

Other than the fact that it's a huge waste of memory and totally defeats the purpose of even having a pagefile or the memory you've wasted on the ramdisk, sure it's a good idea.

 

[gsellis]

Originally posted by: karstenanderson
what about using a ramdisk mapped as a drive letter as a pagefile? i did that for awhile and it seemed to work alright

Not a savings. In the DOS days it was, but now, it removes memory that would more efficiently be used by the OS.

"But wait, some of these boot disc use RAM disk..." You know, I have not figured that one out. Real Windows PE has a "load to memory" option, but it means the OS needs at least 256MB of memory. Loading a RAM disk would take away from loading the OS into memory, which is faster access than making a disk I/O call, even if that is to 'memory'.

MS and others have spent millions of dollars in research, code review, testing, and development to make sure memory management is the best they can do. Some $29 RAMdisk, memory optimization program, or system tuner is not going to beat what they have done and does not in any case I have seen.

So, although it is cool idea, don't expect performance to improve any across the board.

 

[stimpsy]

thanks for the advice

 

[SumYungGai]

Originally posted by: Fresh Daemon

I don't know, I'm not sure of the details. But I've never seen a string of processes only taking up merely 1K each.

Right, well, when you know something you'll be qualified to talk about it, won't you?

In the meantime, you'll have to forgive me if I take my own hard evidence in preference to the conjecture and speculation of a self-confessed ignoramus.

Man, you're just a stupid fu­cking idiot. I don't know what's taking up memory, so that means it's not taking up memory? Hmm, must be more of your "counter-intuitive" logic. Either that or you're just a dumbass.

160MB is "incredibly better" than 128MB? Why?

Are you really that inept? Either you've responded without reading the whole post, or you're just that ignorant. Hmm, maybe because 128 is less than the recommended amount of memory you should have. Like I said before, at that point, the memory is the main bottleneck.

Let me ask you this: if I pulled a couple of 16MB SIMMs out of System I, do you think my results would be very different between tweak levels, bearing in mind that the tests maxed out well beyond the physical RAM capacity anyway?

I don't care about your faulty tests.

And if you think yes, then let me ask you what the possible use is of a set of tweaks for enthusiasts that lose all value somewhere between 128MB and 160MB of physical RAM, in an age when any enthusiast has at least 512MB?

And now you bring up enthusiasts. Well if you're only talking about "enthusiants", why even mention the benchmarks you put up? Seems like it's rather off topic then, or "out of the scope" of the subject at hand.

I've given you my evidence and I've explained why it works that way.

No you haven't. You just said that basically the whole process will be swapped out and never used again, when that clearly isn't true. You saying shi­t isn't evidence.

You've told me that you're taking guesses without any evidence whatsoever.

First of all, I didn't tell YOU anything on that matter. And I never said I made any guesses, nor have I made any.

Jesus christ, all someone says there's one thing they don't know, and you go ahead and apply it to everything they have ever said in they're whole life.

You are being illogical,

Maybe just counter-intuitive. Ever think of that? Well if you did, you would be wrong, because it's counter-intuitive.

and it is that, rather than your failure to accept that things can be counter-intuitive, which makes you appear incorrect.

LOL. Saying that something (an unrelated something at that) is sometimes counter-intuitive, doesn't argue anything. You have no backing for it, and are giving unrelated conjectures.

"Why's the sky blue?"
"Well, it's counter-intuitive."
"Ah, I see now."
"Really?"
"No."

I think you don't know what computers are used for! We don't run tests of variables on them,

Seriously, what the fu­ck? Tests of variables? You've degraded into retardation.

we run software which we use to get our work done or play games. It does not matter how greatly something produces an effect at a very low level if it doesn't affect actual programs.

God damn you are so slow. You totally miss my point. Sure, you run software, but it's not as simple as just running it.

Since services usually don't take up much of anything when unused, and little paging happens when the program is already in use, it just doesn't make sense to bother with a single application, or one that merely does a limited action.

Again, I'm not particularly interested in your guesswork. Test it yourself, you'll find that your results don't confirm your preconceptions.

Test it? Do you even know what I was saying? Why would I test it when I just said it's something that most test don't consider?

I mentioned security in passing because one of BV's tweaks represented a serious breach of security.

Serious breach of security? LOL. Yeah, ok.

And by the way, what you mentioned is also bad for security, so I commented on it. So I can't do what you did? That seems a bit unfair...

I mentioned functionality in passing because one of his tweaks produced a serious impact on functionality.

That wasn't "in passing". It was your way of making a rebuttle. Hardly "passing". And what "serious impact of functionality" was there? If you don't need the service, the only "functionality" you would be losing is, at most, not having to re-enable a service. And normally you don't even have to do anything when it's set to manual.

There are loads of articles and guides on Windows security and services. There are no scientific ones on performance-oriented Windows tweaking. Bearing that in mind, I tested the latter, not the former.

Scientific ones? What in the hell are you talking about?

Oh, and plenty of text-books give some of the advice BV and a few others have said. You're little test isn't enough to "debunk" them.

If you want information on security and Windows services, just Google it.

Huh? LOL, why the hell would you say that? I'm the one saying that you're overlooking it and don't know what you're talking about. And you don't. I'm not the one in need of some learning on that subject.

And I don't really see how it's "beyond the scope", when it's directly related to performance (yes, booting is part of it, it's a fundamental factor).

It's beyond the scope in the same way that strapping rocket boosters to a car is beyond the scope of a performance tuning magazine, although rockets are technically a performance-increasing modification like new heads or spray.

Wow. Way to make an absurd analogy that is completely dissimilar to what was being discussed. While it may not be about gaming, logging on and off, or restarting the computer, is part of the overall performance of the PC, and may matter to people. How is it so absolutely off point to mention that like you're trying to make it out to be?

The thread is about the article. :roll:

This thread is you trying to counter his advice, and give your own.

And frankly, if don't see the difference in a 128MB system by making some of these changes, then you really aren't fit and are too slow to be testing this stuff.

This is a really stupid thing to say. Are you telling me you can see a 0.3fps difference, and that people who don't are unfit and too slow? Who are you, Spiderman?

Wow, talk about stupid things to say. Other than the retarded Superman comment, YOU DIDN'T TEST ON A 128MB SYSTEM. I doubt you even have the wits to even notice the small differences anyway. It's like unless there's some benchmark showing some number, you can't tell how a system is running.

And like I've been saying, you're tests won't show anything conclusive either way. Usually, a service isn't taking up that much, especially when it's idle. Using a game or something won't make any apparent difference, especially if the service doesn't do anything at the time of the test. It also overlooks any slowdowns that may occur when an action causes the service to do something, because it may be checking something just to make sure it isn't needed.

Hell, even something like load times would be much more realistic of a test for a game.

I dislike things that I would have to time with a stopwatch.

So you resort to a test that doesn't even test the change you are making?

My own human eyes and hands are too fallible to produce reliable readings.

That's for sure.

I may do it later anyway, but take my results with a huge pinch of salt.

Everyone should be doing that anyway.

WOW!!. All this time, I keep playing with services. Sometimes, I end up crashing my computer or things not working because of it. Now I know, not to even bother. Thanks!!!!

Well, you just made all of this work worth it! I managed to save one person from instability in the name of unrealizable performance gains. Kudos to you, mate.

People who don't know what they're doing shouldn't be tampering with things like services anyway, unless it's for learning purposes.

 

[Fresh Daemon]

I don't know what's taking up memory, so that means it's not taking up memory

No, if you don't know whether something is taking up memory that means you can't say that it is.

Hmm, maybe because 128 is less than the recommended amount of memory you should have.

Actually, no, 128MB is the recommended amount. 64Mb is the minimum. Microsoft

You just said that basically the whole process will be swapped out and never used again, when that clearly isn't true.

Why isn't it true? What proof do you have?

Tests of variables? You've degraded into retardation

No, that was you who wanted to test variables rather than the performance of the system as a whole.

Bascially, your methodology would be:

If we do X, and it produces no results in our tests, let's modify and alter the test conditions until we do see some results.

That is not scientific method, that is the fallacy of assuming your conclusion.

Serious breach of security? LOL. Yeah, ok.

Actually, it is. Not running any kind of firewall or antivirus will get you the Sasser worm, MyDoom or any of the other horrible things that are out there.

Here's a test for you: disable all firewalls and all antivirus software, and time how long it takes before your computer begins to reset itself without your permission. That's a worm. This was why, before they added the firewall to WinXP, it was recommended to install with the network cable unplugged until you had installed your firewall and antivirus software. Usually your system would be compromised within minutes.

And by the way, what you mentioned is also bad for security, so I commented on it.

How?

That wasn't "in passing". It was your way of making a rebuttle.

That's "rebuttal".

And what "serious impact of functionality" was there?

I noted that Speedfan, which is a very popular program amongst enthusiasts, would no longer install. This is one example.

Scientific ones? What in the hell are you talking about?

Scientific method. Look it up.

Oh, and plenty of text-books give some of the advice BV and a few others have said.

What are their names, who are their authors and who published them?

While it may not be about gaming, logging on and off, or restarting the computer, is part of the overall performance of the PC

No, it isn't. If you think it is, go run DOS. That'll start up, log on and off and restart faster than you could ever believe. Ergo, DOS must have great performance, right? And MacOS X and Linux must suck, because they are slow to boot and restart, right?

I doubt you even have the wits to even notice the small differences anyway.

It needs superhuman wits to notice 0.3fps. For all your blustering, even you, Spiderman (or Superman, since you have trouble reading), couldn't spot that in a double-blind test.

So you resort to a test that doesn't even test the change you are making?

I resort to a test that measures what I actually want to use the computer for.

Here's another car analogy. I want to make a car handle really well, so you say, "Hey, go install a bigger engine, and a turbo, and soften the rear suspension." I try it out, and I say, "This handles worse than before!" To which you reply, "No, it accelerates better. Try accelerating."

Maybe it will - but it's not relevant to me!

Everyone should be doing that anyway.

Everyone is invited to run their own tests. So far no competing figures have been posted and the only people who disagree with me are foul-mouthed people like yourself who can't seem to make three posts without getting banned.

 

[gsellis]

Bump - I thought the alarm clock was a sticky.

 

[imported_BikeDude]

Originally posted by: Fresh Daemon
Actually, it is. Not running any kind of firewall or antivirus will get you the Sasser worm, MyDoom or any of the other horrible things that are out there.

First of all, antivirus products don't really do much. Since they (usually) only protect against known threats, they are little more than a safety net in case you actively try to run infected executables (protection by blacklists). The performance degradation when running certain AV products can easily outweigh the danger of infection -- atleast for experienced users.

Lastly, but not least: Albeit a firewall is nice and all; if you're patched up with SP2 + hotfixes, the known security holes are eliminated, so both Sasser and MyDoom will fail to affect you.

I guess what you meant to say is that a firewall will protect against hitherto undiscovered security holes. (as well as weak passwords in case you enable file sharing -- although it has to be said that being able to access one's own rig using the Internet would be neat...)

 

[Nothinman]

First of all, antivirus products don't really do much. Since they (usually) only protect against known threats, they are little more than a safety net in case you actively try to run infected executables (protection by blacklists). The performance degradation when running certain AV products can easily outweigh the danger of infection -- atleast for experienced users.

Considering that most of the worms in the wild exploit holes that have had patches out for months, it makes sense to run an antivirus unless you're absolutely sure all of the email you're receiving and webpages you're visiting are clean.

 

[gsellis]

Originally posted by: Nothinman

First of all, antivirus products don't really do much. Since they (usually) only protect against known threats, they are little more than a safety net in case you actively try to run infected executables (protection by blacklists). The performance degradation when running certain AV products can easily outweigh the danger of infection -- atleast for experienced users.

Considering that most of the worms in the wild exploit holes that have had patches out for months, it makes sense to run an antivirus unless you're absolutely sure all of the email you're receiving and webpages you're visiting are clean.

Virii can come through port 80 through backdoors in the browser too. Your first indication may be the real time scanner popping. Well known links have been known to be hijacked, but admittedly, not that often.

 

[n0cmonkey]

Originally posted by: gsellis
Bump - I thought the alarm clock was a sticky.

It's a temporary sticky.

 

[n0cmonkey]

Originally posted by: BikeDude

Originally posted by: Fresh Daemon
Actually, it is. Not running any kind of firewall or antivirus will get you the Sasser worm, MyDoom or any of the other horrible things that are out there.

First of all, antivirus products don't really do much. Since they (usually) only protect against known threats, they are little more than a safety net in case you actively try to run infected executables (protection by blacklists). The performance degradation when running certain AV products can easily outweigh the danger of infection -- atleast for experienced users.

Lastly, but not least: Albeit a firewall is nice and all; if you're patched up with SP2 + hotfixes, the known security holes are eliminated, so both Sasser and MyDoom will fail to affect you.

I guess what you meant to say is that a firewall will protect against hitherto undiscovered security holes. (as well as weak passwords in case you enable file sharing -- although it has to be said that being able to access one's own rig using the Internet would be neat...)

Car insurance is there for after you get in an accident. It's a precaution, just like anti-virus and firewalls.

While an anti-virus solution and patching may be reactive, a firewall is proactive. Use one.

 

[QWASZX]

Originally posted by: Fresh Daemon
No, if you don't know whether something is taking up memory that means you can't say that it is.

What have you been reading? God damn you are so effing slow. Please, point out where that was said.

I SAID I DON'T KNOW EXACTLY WHAT CODE IS TAKING UP MEMORY, NOT THAT I DON'T KNOW IF MEMORY IS BEING TAKEN UP. STOP MAKING UP S­HI­T.

Actually, no, 128MB is the recommended amount. 64Mb is the minimum. Microsoft

Ugh. Who mentioned Microsoft? So you would actually recommend 128 as the minimum? Man, you are just out of your mind. 64MB of memory would absolutely CRIPPLE a system.

You just said that basically the whole process will be swapped out and never used again, when that clearly isn't true.

Why isn't it true? What proof do you have?

Nice argument. You're making up crap and then ask me for proof. How can you prove that EVERY SINGLE LAST MEGABYTE is being swapped out? Hmm, task manager doesn't reflect that, and neither do any informational sources.

No, that was you who wanted to test variables rather than the performance of the system as a whole.

!!! NEWS FLASH !!! YOU AREN'T TESTING MORE THAN A FEW SELECT THINGS IN YOUR TESTS.

That's the whole damn point, moron. You're overlooking the variables you're changing. Some generic test isn't going to reflect much of a change, if any.

If we do X, and it produces no results in our tests, let's modify and alter the test conditions until we do see some results.

That is not scientific method, that is the fallacy of assuming your conclusion.

LOL. Scientific menthod? LOLOLOLOL. Scientific method IS SUPPOSED TO TEST THE VARIABLE. That's how the scientific method works, genius. Unless you're specifically seeing if it alters these unrelated things, you aren't going to get good results.

Actually, it is. Not running any kind of firewall or antivirus will get you the Sasser worm, MyDoom or any of the other horrible things that are out there.

Uhh, actually no, it's not a "security breach". And after you're done looking up "breach" in the dictionary, please tell me who said anything about disabling firewalls or anti-virus software.

Here's a test for you: disable all firewalls and all antivirus software, and time how long it takes before your computer begins to reset itself without your permission. That's a worm. This was why, before they added the firewall to WinXP, it was recommended to install with the network cable unplugged until you had installed your firewall and antivirus software. Usually your system would be compromised within minutes.

Uhh, again, no. That's mainly because of the vulnerability in SP1. It's not like it's guaranteed that you'll get some kind of virus "within minutes", especially if you have you're system set up right and know exactly what is running.

And by the way, this is also where extra services is a security issue. It's just leaving more openings for potential vulnerabilities.

And by the way, what you mentioned is also bad for security, so I commented on it.

How?

You're memory is absolutely terrible. And I just mentioned it again above, so I won't repeat that part.

That's "rebuttal".

That's "petty".

I noted that Speedfan, which is a very popular program amongst enthusiasts, would no longer install. This is one example.

You did? Where? I don't see it anywhere in this thread. Try again.

And if you need that particular service, why disable it? Jesus christ, use some common sense. I'm sure BV's advice isn't saying to disable things that are in use.

Scientific ones? What in the hell are you talking about?

Scientific method. Look it up.

LOL. So "scientific articles" means the scientific method? Hmm, I don't quite get that.

Plus it's common sense for some of this stuff (and my books agree). Disabling unneeded junk will free resources, no matter how trivial. And even if it did completely swap out the entire process like you want to believe, I would rather not require any of that to happen in the first place.

What are their names, who are their authors and who published them?

You don't read textbooks? Many have mentioned these things, by many authors. I have mainly Course/Thompson Learning books, and several Microsoft and CompTIA certified ones mention these things. Off the top of my head, the A+ Software/Hardware ones, the various Windows ones, and the Security+ one.

And again I'm repeating myself. Haven't you seen some of the other comments in this thread? Other people are saying some of these things too, but you don't seem to notice.

No, it isn't. If you think it is, go run DOS. That'll start up, log on and off and restart faster than you could ever believe. Ergo, DOS must have great performance, right? And MacOS X and Linux must suck, because they are slow to boot and restart, right?

It's relative, jackass. If DOS didn't have anything I needed, performance wouldn't matter. But if I had a server that took 15 minutes to start up, that could be a problem of availability. So I would want good boot performance, or at least the best I could make it. There's no reason to just completely ignore it as a factor.

It needs superhuman wits to notice 0.3fps. For all your blustering, even you, Spiderman (or Superman, since you have trouble reading), couldn't spot that in a double-blind test.

Holy s­hi­t. I covered this a few times already. I don't care about your faulty tests, and I don't care about gaming performance, because it doesn't really apply. The differences are only noticable under certain circumstances, which could be avoided by some tweaking.

For all your blustering, even you, Spiderman (or Superman, since you have trouble reading),

I'm sorry, I'm paying more attention to the real discussion than your petty offhanded remarks that only make you look foolish.

I resort to a test that measures what I actually want to use the computer for.

I highly doubt you never use anything but games and benchmarking programs, and never have boot your PC.

Here's another car analogy. I want to make a car handle really well, so you say, "Hey, go install a bigger engine, and a turbo, and soften the rear suspension." I try it out, and I say, "This handles worse than before!" To which you reply, "No, it accelerates better. Try accelerating."

Maybe it will - but it's not relevant to me!

Wow. Terrible analogy. The only thing remotely close to this situation is how these tweaks won't make much, if any, difference in something like a game (one that's already open and loaded in memory). The rest of what you said is just garbage. These are simple tweaks for OVERALL performance. If it means a faster boot, that's a plus; if it means hard-paging is slightly faster, that's a plus; if it means means I won't get a slowdown because networking service is waiting for a response it won't get, that's a plus. Freeing a few megabytes of RAM and lowering the amount of items in the process list is also a plus.

Everyone is invited to run their own tests. So far no competing figures have been posted and the only people who disagree with me are foul-mouthed people like yourself who can't seem to make three posts without getting banned.

Maybe if you weren't such a whiny little bitch, that wouldn't happen.

---

Hello, again. Thanks for continuing to be the same prick you always are.

Good bye, again.

AnandTech Moderator

 

[xtknight]

If I want performance, I ditch the antivirus and worthless software firewall as I have for years.

A NAT router will protect from the port-based exploits (unless you're DMZ). Just make sure the ports you do open have secure software running on them.

If you get viruses, the problem exists between the keyboard and chair.

If you use e-mail a lot, get Mozilla Firebird. Besides phishing, you're basically immune to everything. If you visit a lot of shady sites, get something like Opera web browser (now freeware) which has all of its reported exploits patched.