To challenge an old idea

[Perris Calderon]

In our course, and our quest for knowledge, quite naturally most of us are going to hear ideas and reasons from people that know quite a bit more then we do.


because of the respect that we have for these people, and their knowledge, many times we accept what's taught to us to be true...this is natural and correct human behavior...how else to learn, if not to believe our teachers?

Yet still, there are times, where if you think about what's being said, though it's long old wisdom, the ideas and origins fly in the face of current sense.

So, sometimes it is necessary to challenge (introspectively at least), information that has been commonly disseminated;

Now as an example, let me tell you, no matter what you think about the evil empire, I guarantee to you that they at least want to have the best product that they can produce.

Therefore, it doesn't make any sense when a long old and simple fix would still be valid on a later addition of the os....Now I'm not saying that MS will always keep their eyes open, and always be vidulant...what I am saying, is when it looks to you as though the issue should have been simple enough for ms to resolve, at least assume that they've resolved it, and challenge in your mind until satisfied one way or the other.

Three instances that I will sight for you...the IO page lock tweak...ms addressed the issue, yet many sites and experts continue to post this obsolete fix.

The page file tweak...one simply had to say to oneself..."why wouldn't ms simply change this themselves? They must believe they are preparing the computer correctly...this is such an old tweak, and it would be so simple for ms to correct this...I therefore challenge"

Partitions; Ask youself..."why would Microsoft now install it's OS on a single..., they used to install on multiple partitions"...it seems as though Microsoft is convinced their OS is faster on a single partition (microsfts own words; "When performing a clean install, Microsoft recommends...that the system be installed in a single partition on each disk. Under Windows XP, big partitions are better managed than in previous versions of Windows. Forcing installed software into several partitions on the disk necessitates longer seeks when running the system and software").

...yes, it's more convenient for some of you to divide your hard drive, yet unless you do this correctly, you will suffer performance for your convenience...

btw, to parition correctly on a single drive, I'm going to qoute from "jeh"...(Microsoft) is saying that the OS + apps should go into one partition. The system is generally accessing OS files fairly constantly, and if you have your apps in some other partition on the same HD, that will indeed result in longer seeks. And seek time absolutely dominates disk IO time (the actual transfer time is about 10% of the total time for a typical disk IO).

This concern goes away if you put, say, your OS and your apps in separate partitions _on separate drives_. That's BETTER than having them in the same part., again because it means you aren't moving the heads to go back and forth between OS files and app files.


and I'm sure their are other examples..

What I'm saying is this...as far as your gui...go ahead, give yourself the look and functionality that you choose.

BUT...as far as the smoother operation of xp...beleive me, ms will try not to make the same mistake twice, so if the advice you are hearing is old advice...challenge...if the advice you are hearing is new advice...benchmark

perris

 

[Perris Calderon]

comments?

 

[da rock]

couldn't of said it better myself!

 

[xsivforce]

No argument here.

 

[GraLk]

The sound voice of reason.

 

[Dirk Diggler]

If a person tries out a tweak and that person believes the tweak has helped, then that tweak HAS helped.
Whether the tweak actually worked or not is irrelevant. The person feels better for using the tweak and is more likely to be more productive "knowing" that his/her tweak has helped.

P.S.

My P.M.'s are now empty

 

[Perris Calderon]

Dirk, I agree with you, this all goes to the enjoyment of our little box...what I'm saying, is "Challenge, and learn"...

ps...empty your pm, it's full...(they all love you too much)

 

[allan]

I really didn't understand your comment about the OS being installed on multiple partitions. Regardless, I am a strong advocate of partitioning your hard drive. Aside from the fact it makes for better housekeeping, it's just smart to separate your data from the OS. I don't care how careful you are about backing up your data - the one day you forget is the day your system will crash and you won't be able to access the boot drive. And if the data files are on the boot drive, a reformat will wipe them out.

A performance hit because you partition? Please. If your system is so underequipped that you would notice any slow down after partitioning your hard drive, then it probably didn't meet the minimum requirements for XP in the first place. Can you run tests that will show a difference? I don't know - maybe. But forget about tests - in everyday operation there is absolutely no noticeable difference in performance or seek time. With the speed of todays HD's and processors any difference would be so minimal as to be absolutely, positively transparent.

Also, partitions give you an on-disk repository for a backup folder in which you can place copies key files (and folders) as well as an image of your boot drive (using Ghost or Drive Image). One large partition simply makes no sense to me.

 

[jeh]

Originally posted by Dirk Diggler
If a person tries out a tweak and that person believes the tweak has helped, then that tweak HAS helped.
Whether the tweak actually worked or not is irrelevant. The person feels better for using the tweak and is more likely to be more productive "knowing" that his/her tweak has helped.

I couldn't disagree more.

If someone believes that they can levitate, that they actually have levitated, even though videotape shows them just flopping around on their knees and legs... have they "really" levitated? Only if we ascribe useless meanings to the words "really" and "levitated". Meanings which will only lead to further misunderstanding. At best.

Many tweaks show APPARENT benefits but can actually hurt in the long term. Getting rid of your pagefile is one example.

Another good example is the various "memory tuning" utilities. Several I tested, rated rather highly by two public download sites, merely allocate as much memory as they can, thereby forcing the OS to reclaim memory from all other processes and from the file cache. Then the "tuning" program exits. Voila! Look at all that "available" memory! What you don't see, though, is that everything you have running, including the pageable code and data in the OS, and including access to files that were previously in the cache, will now run slower than it would if you'd just left the system alone.

Tweaking without understanding what you are doing, and without having an objective, reproducible set of tests to evaluate ALL results (not just those most immediately apparent), is like throwing sand into machinery without even knowing what the macinery does, let alone how it works. This "disabling paging" thing is a prime example: This OS can't run without paging; "disabling paging" is a term without a referent here.

More generally, to repeat what I've said a few other places -- most "performance tweaks" are non-solutions for problems that most people aren't having in the first place.

 

[Iceman]

is this the same "jeh" that dealer is talking about all the time from microsoft?

 

[Dirk Diggler]

Hi JEH,
if somebody uses the IoPageLockLimit tweak (which is the kind of tweak I was reffering to) and they believe it to work, how can you disagree with it.
Or is your sole purpose of trawling these forums, to look for posts or people to disagree with.
On most if not all your posts that I have seen, all you do is knock. Why not put some of that obvious talent you have to some good. Give us some tweaks that do work, don't just tell us what doesn't work, we'll figure that out for ourselves, eventually.

 

[Perris Calderon]

Originally posted by Iceman38
is this the same "jeh" that dealer is talking about all the time from microsoft?

Jeh is not to my knowledge from microsoft, but he does have a more intimate knowledge of this os then anyone I have personally read...this is his website

 

[jeh]

I'm the same jeh, but I'm not from MS. (And I've asked dealer to stop the quoting; I think any text by me on someone's forum should be posted by me, so it can be challenged directly if someone disagrees.)

(Edit: Oh gawd, now he's posted my URL! Nothing wrong with that, except that it's been an embarassingly long time since I've added anything to the site. Partly because I'm spending too much time posting on web forums. )

Dirk - I am "trawling", to use your word, as my latest in an ongoing series of diversions in an attempt to avoid responsibility. Er, no, that's not it. Seriously, I'm trying to find out what problems people are having and to see what misconceptions there are out there, things that I indeed CAN help with. Since I don't want to be "leeching" I feel it's reasonable to repay the sites in question by posting corrections where incorrect information has been given.

Would you really rather that incorrect information go unchallenged???

don't just tell us what doesn't work, we'll figure that out for ourselves, eventually.

Even in cases where the "tweak" could lead to real problems? You'd rather just experience those problems later, instead of avoiding them?

if somebody uses the IoPageLockLimit tweak (which is the kind of tweak I was reffering to) and they believe it to work, how can you disagree with it.

Because it's easy to show that the OS doesn't read that value from the registry -- it's easy, just use regmon (from www.systinternals.com). (Same for IRQ8Priority, by the way.) So how can the tweak "work"? For any useful definition of the word "work"?

If I create a registry value called "MakeSystemRunFaster" and set it to 1, and if I believe it helps... you're saying that the tweak therefore "works?" What if I set it to 2, will it work even better?

Suppose someone came along and actually posted such a thing, in all seriousness - would you really prefer such a thing not be left unchallenged?

Give us some tweaks that do work

The problem with that is that it's very nearly a null set. I know no one wants to believe this, but in terms of these registry-level tweaks, the system really does run pretty close to optimum out of the box, for most hardware configs and most workloads. So the best I can offer is to explain how and why that can be so, even though it often looks otherwise.

(A high "available" amount of memory, for example, does not mean the OS is somehow at fault for not handing out more memory to apps, just like a high "pf usage" does not mean you're doing a lot of paging.)

Disabling unneeded services is good for security, and to reduce startup times, but that's about it. The only service I know of that is really a constant drain on resources is the Indexing service, and you can turn that off through its own GUI.

(And that's not necessarily a great idea either; the content indexer helps my in my work tremendously -- it can search about 20 gigabytes of source files for the definitions and/or uses of an identifier in literally a second or two. But most people aren't working with code bases that large. As always, it depends on what you're doing.)

A key point here is that "tweaking", or "tuning" as we tend to say when we're focusing on performance areas, can only make tradeoffs. If you are constrained by some hardware limit, tuning does not make more hardware available; it only reallocates the existing hardware. The result of a "tweak" might easily be that an OBVIOUS metric (like the "available" memory counter) gets better, while actual performance gets worse. For that reason you cannot rely on reports like "the system seems to work better". You have to get in and understand what is really happening. And you can't do that with just the information presented by the OS; too much detail is missing and what isn't missing is often labelled in confusing ways (like the "PF usage" graph, not to mention the "system cache" counter).

But, look, I gather you're an admin here... if you really want corrections to NOT be given, fine, I'll withdraw.

 

[Iceman]

Originally posted by dealer


Jeh is not to my knowledge from microsoft, but he does have a more intimate knowledge of this os then anyone I have personally read...this is his website

thanks, for clearing that up, dealer.

 

[xsivforce]

Originally posted by jeh
if you really want corrections to NOT be given, fine, I'll withdraw.

All information is welcome. It is up to the individual user to read through posts and decide what is best for that person.

 

[Perris Calderon]

to quote myself on an earlier thread,

uh oh...

anyway, since no one is going to come to jeh's defense, and it's because of me he came to this site, let me say this, and I'm sure jeh will correct what I'm saying if it's wrong;

Jeh believes that there is not much you can do to make the XP operating system much better...that there is usually a trade off, for performance in another area, for instance, when you turn a service off that you believe you don't need, true, you may not need the service, but somebody does, so this cannot be considered a viable everyday tweak, and in addition, once the service is started, the said service will not affect the performance of most users...

of course none could take issue with setting your GUI in the fashion to suit yourself

I've explained to jeh, and I think he missed the point, that turning off services is intended to speed the boot time, and Personally, I think services should be customized by each user.

Jeh tries to educate, and correct what he believes is misinformation, his interest is not in finding the faults with XP, it's with not letting people do harm with rhetoric.

Now to play the other side, I have some issues with this reasoning...and I have pointed out in jeh's forum, that these are forums, and discussion is an important part of our learning process...when a person says something new, and gives the impression that he's right, period, with no room for personal error, that's an irritating attitude that most people might take offense to...

still, I love the information jeh brings, and everyone knows I love the challenge of "accepted philosophy"...I'm hoping jeh loves the dialog also, and therefore, enjoy the forum process with those of us that are trying to learn as much as possible

(edit)..sorry, jeh posted for himselfe before I finnished these pros


I'm sure everyone agrees, no one would be at all pleased if jeh withdrew

perris

 

[Dirk Diggler]

Hi JEH,
thanks for your reply, I am only a member of the forums here nothing more. I wouldn't want you to stop posting even if I were. I just thought you could be more constructive in your postings.
I have read your reply and I have to agree with the majority of it. The exception being the non-working tweak. There has been for some time a number of tweaks that do not work in XP, people used them in the assumption that they did, and have felt that their system was working better. A placebo effect to be sure.
I think the most obvious tweak to show:-
""If I create a registry value called "MakeSystemRunFaster" and set it to 1, and if I believe it helps... you're saying that the tweak therefore "works?" What if I set it to 2, will it work even better?""
this effect was the "Enable Pefetcher" tweak
"0", "1", "2", and "3" were valid numbers, "5" was an invalid number but people used it and they honestly believed that it worked and were happy with this.

 

[Perris Calderon]

I have entered "MakeXPRunFaster", and I set the value to zero, xp seems to run a fraction af a second slower, so I'm sure this is a valid tweak, and I'll be setting that to one from now on

 

[Dirk Diggler]

You gotta love him haven't ya

P.S.
Did you use a DWORD or a Binary

 

[xsivforce]

I entered "MakeAHotBrunettePopOutOfMyComputer" set the value to "1"...it didn't work.