Debunking Common Windows Performance Tweaking Myths

HardOCP News

[H] News
Joined
Dec 31, 1969
Messages
0
Most of you guys know which Windows performance tips work and which ones don’t but this is a good article to forward to friends and family that might not be as savvy.

Besides the tweaks that simply don't work, some of them will actually cause your computer to run even slower—or worse. Let's examine some of the most offensive myths out there regarding PC performance tweaking, and debunk them once and for all.
 
All of his advices are sound except the very last. I've always been an advocate of leaving these three services on (Superfetch, Defender, and Search).
 
The only things I disabled were UAC and Windows Defender (no active internet connection).
 
Great article. I do product support in a segment of the gaming market (high end Microsoft Flight Simulator addons) where there is an unbelievable amount of this type of snake oil accepted as gospel... Gonna post this link on our forums right now actually.
 
Disabling services and all that jazz only gives a placebo effect to the end-user.

As Azhar stated, disabling SuperFetch, and Search is a pretty retarded idea.
 
All of his advices are sound except the very last. I've always been an advocate of leaving these three services on (Superfetch, Defender, and Search).

Disabling services and all that jazz only gives a placebo effect to the end-user.

As Azhar stated, disabling SuperFetch, and Search is a pretty retarded idea.

In his defense he did say to only do so under certain conditions. And I'd to agree with search and defender. If you are using google desktop search or something like that, then it wouldn't make a lot of sense to have windows search enabled. And in the case of windows defender, I would normally leave that on, however some programs (such as symantec endpoint protection) force you to disable it. I think I have seen another anti-malware program do the same.

I prefer defender over stuff like spybot or stuff like that for the average user and it auto updates itself and autoruns. However I can still see somebody wanting/buying other programs. If that is the case there is no need to run a large number of such programs and again I can understand disabling it if you are running a few other programs.

Superfetch is the only one that I don't see a reason to really ever turn off. But for the other two, i could see under the conditions given why you might want to turn them off.
 
Good article. Another common myth is manually setting/disabling the page file helps. There just isn't any real good reason to do this and it could hurt things more than help.

From another forum:

1. Having more pagefile space than you need doesn't hurt anything. In particular it will not encourage more paging than otherwise.

2. You can sometimes get away with running without a pagefile. This does not eliminate paging. It does mean that when paging must occur, it can only be done to virtual memory that is backed by files other than the paging file -- code files, and cached files. Process private virtual memory must stay in RAM for as long as the process exists, even if it's no longer being used by the active code. This is wasteful of RAM.

3. Fears about excess overhead when the OS "resizes" the pagefile are completely unfounded as long as the initial or default pagefile size is large enough. Allowing pagefile expansion (that is, a maximum size larger than the initial or default size) only serves as a safety net. If the extra space isn't needed, there is zero cost. If the extra space turns out to be needed, it's much better to have it than not.

4. A good guideline for the default pagefile size is large enough that actual usage (as measured by performance monitor) is typically 25% or so under your max workload. The allocation algorithms work best when they have a lot of free space to play in. But don't get concerned if it gets up to 50% now and then.

5. "System managed size" is usually more than adequate. And it does not mean the OS is "constantly resizing" the pagefile. It simply sets a default size of 1 or 1.5x RAM, and a maximum of twice that.

6. Fears about disk fragmentation or pagefile fragmentation caused by resizing the pagefile are also unfounded. The original allocation is not affected by the resizing; additional extents are simply allocated as needed. When they are no longer needed (at the next reboot, if not sooner) the additional extents are freed and the pagefile is back in the same state and location as it was to begin with.

(Added) 7. If you don't have a pagefile, any app that tries to create a pagefile-backed section will fail. (As of Vista, at the latest, this no longer appears to be the case). Of course if your current "commit charge" plus the requested size of the new section exceeeds the amount of RAM available for pageable memory, the lack of a pagefile will still cause the CreateFileMapping call to fail.
 
Disabling services and all that jazz only gives a placebo effect to the end-user..
Nonsense. Disabling anything that is running and taking up RAM unnecessarily is useful. There are a ton of unneeded services that run by default, at least in XP. And I have run benchmarks.

Go ask any of the record holders on Xtreme Systems if killing services makes a difference.

The author hurt his credibility in what would have been a good article with that last one.
 
Nonsense. Disabling anything that is running and taking up RAM unnecessarily is useful. There are a ton of unneeded services that run by default, at least in XP. And I have run benchmarks.

Go ask any of the record holders on Xtreme Systems if killing services makes a difference.

The author hurt his credibility in what would have been a good article with that last one.

Benchmarks. Sure it will make a difference in benchmarks. However, everyday computing it will not make a difference depending on system specs. Tweaking your system is good if your trying to squeeze out performance on minimum specs. If your running a quad core rig with eight gigs of ram, SLi video, and a couple of raptors in RAID0 disabling everything that you think you don't use is really just a waste of time.
 
Nonsense. Disabling anything that is running and taking up RAM unnecessarily is useful. There are a ton of unneeded services that run by default, at least in XP. And I have run benchmarks.

Go ask any of the record holders on Xtreme Systems if killing services makes a difference.

The author hurt his credibility in what would have been a good article with that last one.

I see you've completely ignored the author's commentaries. He specifically said that while performance increase might be minimal, the important factor is that turning off services may harm your operating system because when it calls for a service, it won't run.

As for your RAM comment, unused RAM is wasted RAM.
 
Tweaking services is useful for people who are complaining of slow performance. Not everyone is running uber systems with 8GB of RAM. For the single mom with 3 kids who saved her pennies to get a new Vista machine but could only afford one with 1GB of RAM, tuning services to preserve that precious 1GB under Vista will help a lot.
 
I also disagree with the part about not turning off System Restore. For me, it's useless and is just taking up space on my hard drive. I do backups using NTbackup to an external drive in case of a drive crash. But I have never once, ever, used System Restore. So I disabled it.
 
Not everyone is running uber systems with 8GB of RAM. For the single mom with 3 kids who saved her pennies to get a new Vista machine but could only afford one with 1GB of RAM, tuning services to preserve that precious 1GB under Vista will help a lot.

1 GB is perfectly fine for a non-gaming Vista machine with all services on.
 
I see you've completely ignored the author's commentaries.

In my reply, yes, because people tend to read headlines, not the fine print, so his damage was done. Despite backing off his statement a bit, I still think he slights this technique too much. I've speed up many a PC - including the proverbial grandma looking at Picasa photos - by killing services and other startup processes.

As for your RAM comment, unused RAM is wasted RAM.

Utter nonsense. RAM gets fragmented and, at least under Windows OSs, the more that gets loaded into it, the less efficient it becomes at I/O to the CPU. Unused processes sitting in RAM slow down active processes, even if the CPU utilization is zero. At some point, the effect can have a serious hit on performance.

Even the performance of single apps begins to deteriorate because of RAM issues. Ever notice how even Mozilla 3.0, a well-coded app, begins to get slow and jerky if you open too many windows, despite having 4 GB free?

Ever wonder why Windows boots slower and slower over time? Startup exes and processes added by installed applications over time is one big reason. The idea that it doesn't matter how many separate processes are running in the background doesn't maytter, so long as you have not reached your limit of physical RAM, is ludicrous.
 
1 GB is perfectly fine for a non-gaming Vista machine with all services on.
You say this not even knowing the hardware config or installed applications and startup processes? Genius. You must work at Best Buy's help desk.

I recently encountered a relatively recent machine (Intel 820 dual core) with 1GB RAM owned by a grandmother who used Word and Picasa and Quicken that took five minutes to boot up. The problem? Too many startup items and random services loaded at boot. I killed all the unnecessary ones - and did nothing else - and cut several minutes off the boot.
 
your name happen to be BlackViper by any remote chance?
 
I suspect it was more about the needless startup apps and less about the services. I've NEVER seen a speed boost significant enough, from just shutting off the dozen or so unneeded services, to show up within the margin of error of a stopwatch.

The author of the article is correct, most of this crap is placebo effect. Sure there are minimal speed gains, after all, if your machine has fewer tasks to complete, the faster it will get them done. However, anyone who says they can actually notice a difference after disabling services, defragging RAM, or turning off their page file, is just seeing what they want to see.

Most of those services just sit there and use minimal memory while listening and waiting to be used. They aren't actually doing anything and take tiny fractions of a second to load.
 
1 GB is perfectly fine for a non-gaming Vista machine with all services on.

I thought Vista was dog ass slow on 1GB of RAM with every computer I've tried it on. Most recently, I bought the cheapest ($525) full Intel chipset C2D laptop I could get at Fry's and I had to run it for a few days with 1GB until my 4GB upgrade came in from the 'egg. It's much, much better now, and although it does not see the full 4GB, my Ubunutu 64 bit installation on the other partition does.

When people ask me, I always tell them my professional opinion is that they should get at least 2GB of RAM to run Vista, if they can afford it. I think it's happy enough with 2GB, although more is better.
 
You say this not even knowing the hardware config or installed applications and startup processes?

I've run Vista on everything from a 1.6 GHz Pentium M notebook with 1 GB of RAM to my quad core gaming machine. It runs just fine on all of them.

Genius. You must work at Best Buy's help desk.
:rolleyes:
I've been doing solo PC repairs since I was in high school. I'm a computer engineering major currently interning at a large defense contractor. I help design lab hardware for testing aircraft radar systems. What do you do?

I recently encountered a relatively recent machine (Intel 820 dual core) with 1GB RAM owned by a grandmother who used Word and Picasa and Quicken that took five minutes to boot up. The problem? Too many startup items and random services loaded at boot. I killed all the unnecessary ones - and did nothing else - and cut several minutes off the boot.
Any machine will slow down when loaded down with crap. Do you have any other information you want to enlighten us with?
 
I've been doing solo PC repairs since I was in high school. I'm a computer engineering major currently interning at a large defense contractor. I help design lab hardware for testing aircraft radar systems. What do you do?

If I can guess from his username, unless his name is JD, he is or wants to be a lawyer (LOL scumbags.)
 
In my reply, yes, because people tend to read headlines, not the fine print, so his damage was done. Despite backing off his statement a bit, I still think he slights this technique too much. I've speed up many a PC - including the proverbial grandma looking at Picasa photos - by killing services and other startup processes.



Utter nonsense. RAM gets fragmented and, at least under Windows OSs, the more that gets loaded into it, the less efficient it becomes at I/O to the CPU. Unused processes sitting in RAM slow down active processes, even if the CPU utilization is zero. At some point, the effect can have a serious hit on performance.

Even the performance of single apps begins to deteriorate because of RAM issues. Ever notice how even Mozilla 3.0, a well-coded app, begins to get slow and jerky if you open too many windows, despite having 4 GB free?

Ever wonder why Windows boots slower and slower over time? Startup exes and processes added by installed applications over time is one big reason. The idea that it doesn't matter how many separate processes are running in the background doesn't maytter, so long as you have not reached your limit of physical RAM, is ludicrous.

I've never had FireFox get slower when I open more windows. Windows Vista also has never booted slower and slower over time.
 
Back
Top