What do you think about this article / Tweaks debunked?

I posted this on my blog a while back and now come to think of it, I am thinking, is it true what Life
Hacker said in this article?

Debunking Common Windows Performance Tweaking Myths

 

All of these look fake except for 'disabling services'. I think they're tryin gto say to be careful when you do, not that it won't have an effect.

I can tell you you'll gain security and link to CVEs (like the ones used in Stuxnet) but in terms of performance the gains are minimal and depend mostly on other things.

Otherwise every single thing they listed is pretty much a total scam / snake oil.

 

So basically, what that article said that setting your Number of Processors in MSCONFIG / BOOT options to 4 is just a gimmick and really doesn't increase performance but may actually hurt it?

 

I can't count the times I've heard about memory optimizers.

Services are a tricky area. You can gain some security by disabling less than a handful of services, but you can really screw things up in there too. I think I would rather just point new users to more effective security measures and avoid the risk that comes with mucking with services.

 

Everything in the article is true. Except for the disabling services part, which is partially true. Disabling Windows services can increase performance but this applies almost entirely to Windows 2000 and XP and it also matters if the computer is quite old (Athlon XP/Sempron based PC with 512 MB or something). And still the gain would pretty small.
SuperFetch should not be disabled if performance is important. Especially on laptops. It's more important for them than for desktop PCs due to the slower hard drives. It's pretty logical actually. SuperFetch uses idle states to pre-load data that is going to be needed in the near future. When an application is launched some of its data is already loadad, so less data needs to be read from the hard drive. With SuperFetch disabled there would be no pre-loaded data and everything would need to be read from the hard drive. And this takes longer on laptops because their hard drives are slower - 5400 RPM, compared to 7200 RPM for desktop PCs. Which is why pre-loading data on laptops is even more important.

 

ot posts removed

 

Yes always turn off PRINT SPOOLER is my advice.

 

Disabling services definitely makes a noticeable difference on a low spec XP box (i.e. 512 MB - 1 GB of RAM). When combined with other tweaks, notably visual themes, etc... it can even make a substantial difference.

On boxes with 2+ gigs of RAM and multi-core CPU's, probably not though. On Win7, probably not either.

All the other stuff was indeed bogus. Some even laughable.

The problem is these people claiming that disabling services doesn't make a different have probably been running high-end setups/Win7 for so long they forget what life was like prior. And/or they never actually ran barebones back then to actually test the theory themselves either.

 

Clearing Out Windows Prefetch for Faster Startup - it is funny, he actually confirms, that it is not a myth, because when you are gonna run an app, which is not prefetched already, it will load slower, because of other preloaded apps, which have to be dismissed in order to run a desired app. I tested it myself and it looked like that (subjective).

Cleaning the Registry Improves Performance / Clean, Defrag and Boost Your RAM With SnakeOil Memory Optimizer - When the registry size is 100MB and after cleaning 40MB, it is pretty obvious, that it takes less resources, so there is nothing mythical about it. I also use CleanMem, which cleans RAM much better than Windows, so I get no more messages like insufficient RAM. Some people said to buy more RAM, why to waste money, when it can be fixed by a simple utility?! Defrag decreases boot time by a few secs.

Disabling Services to Speed Up the Computer - you can gain a few MB of free RAM (~5 MB per each), my 8 used as little as 180MB of RAM only, and it can actually solve a few problems and increase security, when network services deceptible to common vulnerabilities are disabled. I have only those chosen ones running. Win runs smoothly.

 

Thank you berryracer that was very interesting reading.

 

TweakHound's Bad Tweaks (Windows XP):
http://www.tweakhound.com/xp/xptweaks/supertweaks11.htm

TweakHound's Bad Tweaks Continuation:
http://www.tweakhound.com/windows7/tweaking/14.html

I like this quote from TweakHound:

As for clearing out the Prefetch folder, it will most likely cause the next boot to be slower since Windows will have to cache everything again but this allows you to start with a fresh cache. This isn't something you should do often.

As for registry cleaning tools, they do not 'speed up' your PC but what they do is to allow admins to identify entries that are no longer needed, and if the admin decides so, remove them regardless of what the benefits may be. The element of risk is there.

Registry Junk: A Windows Fact of Life (Mark Russinovich)

Facts about Registry Cleaners (Bitsum)

As for memory optimizers, I am going to be controversial here, although I don't intend to provoke anyone. First and foremost, I genuinely appreciate the reasoning given by memory experts like Mark Russinovich and many others and accept their logic/arguments. However, I make an exception in that if a person needs such a tool (knowing that this would cause a penalty hit), the only ones that I'd link him/her to would be CleanMem and Minimem. I particularly like this article:

Virtual Memory facts and Memory Optimizer scams(Bitsum)

 

Indeed, but if Superfetch is disabled, it is no longer the case, because prefetch files are no longer created. I actually forgot to do it, so I am thankful for this topic.

 

There is no reason to "dismiss" anything. Cleaning the prefetch folder isn't fatal as Windows will rebuild the cache completely after 3 reboots. So not much is lost. However, it is also completely pointless as nothing can be gained either. And no, there is no need to clean "obsolete" files there. Windows does that automatically.

For this to actually work, the registry needs to be not only cleaned but also compacted (rebuilt from scratch), like NTREGOPT does. However, taking a registry down from 100 to 40 is very optimistic. When I've run it in the past I've never gained more than a few megabytes, and this is simply laughable compared to how much memory some common applications use and also compared to how much memory today's systems have.

In Opposite Land maybe. In our world - no.
It's really funny to think that some guy knows more about the inner workings of the Windows kernel than the people who wrote it. Some guy somewhere has a better idea how to optimize Windows than the people who created it. It's just ludicrous.

 

Clearing a cache on reboot makes no sense for an OS. The cache is constantly being replaced through a cache replacement policy.

The CleanMem thing just takes pages that would have been cleared anyways and pages them. Some of them (very dangerously) mark pages that aren't going to be cleared to be cleared.

Your boot process always uses multiple cores. There's like... this insanely tiny window where it uses 1 core (and nothing you do will change that) but that part's over in milliseconds.

 

The point is, what is better, every expert has a different opinion on it, even within Windows circles. Windows makers always do, what they consider to be the best for the majority, like default services running, the metro GUI without Start button, preloading apps to RAM for a faster startup at the cost of lower RAM, that cause problems running apps, which require more RAM. CleanMem author decided, that is better to have more RAM rather than to start some apps a few seconds sooner, I rather go with him on that then with Windows makers. I always go for the facts rather then for a talk, like a washingball, every expert claim, that it does not work, but in reality, it does.

 

It takes an insignificant amount of time to remove those preloaded apps from memory (on our systems when I profile our code, the time taken to clear a cache usually below statistical noise).

Wrong, the registry is accessed as a database - a bit at a time on demand, its not loaded into memory completely. Only resource saved is a bit of disk space.

This one I agree with, since Win XP boot files are defragged and optimised using layout.ini.

And how much more performance do you get from that extra free RAM ?

Cheers, Nick

 

I've already read about these things...

 

So it's better to have unused RAM and slower startup of programs, than to have used RAM and faster startup of programs? What kind of twisted logic is this? The job of the memory is to be used, not for it to sit there empty.

 

Yes, registry defraggers (compactors) have their occasional use. My understanding is when software is uninstalled the registry does not decrease in size, so over time with much installing/uninstalling the registry grows and includes unnecessary space - a registry defragger will remove the space. I don't know about the performance claims some of them make though, such as improving registry access speed. I think that would be barely measurable.

 

If there's no memory contention, you want max. cache. The only question is how effectively the kernel manages memory deallocations from its buffers and caches when needed. And nothing can beat the kernel itself when it comes to this management.
Mrk

 

Once every few years or something like that.

That is basically correct.

There is no improved registry access speed. The registry is loaded into memory and it's a binary tree, meaning thaty any key/value is accessed equally fast. Therefore the size of the registry has no direct impact on registry access or performance whatsoever.
One could argue it has an indirect impact, as compacting the registry means it will take less RAM when it's loaded, which would leave more RAM for other applications. However, the amount of memory gained that way, compared to the amount of memory today's computers have, is simply laughable. Who cares about several megabytes, when a modern system has between 4 and 16 gigabytes of RAM?
Also, there is the theoretical benefit of a faster boot time, because a compacted registry takes less space on the hard drive and should result in the hard drive reading the files faster. However (again), we are talking about ridiculously small differences. Hence the usage of the word "theoretical".

It's simply amazing how people are interested in performing things that have little to (litteraly) no benefit, yet they ignore much more important things. A user should be a lot more concerned what kind of software he/she installs on the system. A badly written piece of software can cause a lot more issues than some unnecessary entries in the registry. It's like being concerned about a splinter when you have a gunshot wound.

I can't believe I actually agree with you on something.

 

Night Raven, if we stick to actual technical stuff, I believe we will agree on many points.
Mrk

 

Probably the most sensible thing posted here. Defragging the registry will save you maybe a few MB of RAM at most (and probably much less, KB are more likely). Focus on other things if you want performance.

 

Not a binary tree, but it is tree based.

The free space in the hive files is part of the data storage structure of the reqistry. It is not anything that is loaded into memory.

Windows XP and newer loads portions of the registry hive ondemand, loads in 16kb chunks via the cache manager utilising 256 slots. Win 2k did map the hive files into the pagable address space (so would free up unused pages in ram allocated to the registry).

 

While I agree with the bulk of your post,if the part I highlighted were always the case there'd never be such a thing as Patch Tuesday.

Microsoft undoubtedly have the greatest knowledge of the Windows system's workings in isolation,it's when all the other 3rd party stuff is added that the gaps appear.