Windows 7 Needs Liposuction

I'm very sorry you feel that way about windows 7. Hey, good news though! You don't have to use it.

However, I have some... mightily annoying issues with your "tests". That is, you don't actually tell us what you did. I mean, sure, if you're following in the footsteps of Adrian, I guess you'll believe this to be correct practise, but unfortunately it's not. Post your test methodology and how you came to be at these arbitrary numbers, make it repeatable, and we might be more willing to listen to your post.

Unfortunately, your statement about services is EXTREMELY misguided. Services under windows XP were tasked, usually, with performing multiple things. That's why the total number of running services in XP was so low; they were all hosted under a smaller number of processes. Which means that if one failed, they all did (that were under the same process, anyway). Windows 7 tries to remedy that by splitting services up into far more processes, which has the added benefit of increasing how threaded the operating system is.

In addition, Windows 7 has a lot more to do by itself. For example, wireless networking services which simply weren't there when XP was released. Media sharing services that didn't exist 8 years ago. Homegroup, which is a unique new way of networking without the hassle, requires its own services. Many services, when disabled, reduce performance. Superfetch for example, if you have a hard disk and not an SSD.

And finally, if you have not used Windows 7 on a machine with less than 4gb of ram, then you won't know that windows dynamically assigns more memory to the system if more is abailable. Boot the OS with 512mb of installed ram and it automatically scales down memory-hungry services, fitting into a space as small as 250mb.

I'd be glad to hear your testing methodology, feel free to rebuke any of my points.

Posted by Vigilante on Sep 15, 2009 8:35 AM

There are at least two kinds of "evaluations" possible, or perhaps better said, two ends to the spectrum of evaluations; precise, controlled environment laboratory tests, and overall usage and performance evaluations. Both are useful and relevant, as long as you keep in perspective what they are and what they are good for.

I find the information that XWJ has posted to be both relevant and useful, because it is looking from exactly the perspective of the majority of average PC users. Put Windows 7 and Windows XP on the same hardware, run the same programs, and see how it compares. That is immediate, useful and relevant information.

Conversely, I don't see the relevance of the statement that "wireless networking services simply weren't there when XP was released". So what? They are there now, and XP attempts to support them now (although anyone who has ever fought with Wireless Zero might dispute that). The same is probably true of "Media sharing services", although you aren't at all specific about what that means or entails. As for "Homegroup" networking, I think that goes right to the heart of the problem - adding services which clog memory and processing, with dubious value and returns. Although from the name and the functionality one might guess that this is a reincarnation of "Windows for Workgroups"... shudder.

And finally, I would be surprised if XWJ or anyone else had tried Windows 7 with 512 MB of RAM, since Microsoft specifically states that the minimum requirement is 1 GB (or 2 GB if you are going to use XP Mode). In any event, it boggles the mind to think of an operating system that "scales down to a space as small as 250 MB".

jw

Posted by J.A. Watson on Sep 15, 2009 10:42 AM

Thanks for responding and, also, for telling me that you haven't really been keeping up with Windows development post-xp.

Windows Media Player Netowkr Sharing Service is a new service introduced with Windows 7/WMP12 to allow sharing to any attached media devices, xboxes, media extenders, and also allows remote access to your music over the web, if enabled. It's one of the most hyped features of Windows 7, and personally, I do not use it since I personally do not find it necessary. And guess what? Windows realises that and hasn't got the service enabled.

Again, same with Homegroup. Have you never even heard of this? Since it has been known for almost a year that this technology was being introduced, I doubt that you are an IT professional with any real merit. You need to keep up in this industry, or be left behind. Anyway, ignore my digression: Homegroup is an easy way for windows computers at home to share files and folders. No network configuration necessary, no file permissions (manually) need to be set, its a one-click way of getting stuff working. And it works exceptionally well.

But, since you obviously missed the point I was attempting to make, I'll reiterate in simpler terms. I do not use homegroup, and therefore it is not "clogging memory and processing". Just to try and determine the impact of enabling this service manually, in the interests of curiosity, I did. The memory usage of one of my svchost.exe process increased by 1020KB, and nothing else was affected. I find it HIGHLY unlikely that in this day and age, 1MB of memory usage for, what is essentially an intelligent network auto-configuration system, will make much difference to your system performance.

My complaint regarding the information XWJ provided is that he does not state how he got that information. "A similar test I've done running our current application on Windows XP Pro and Win7 Beta on the exact same hardware revealed that XP Pro was faster by 5%! What happened? " Well, which application? Was this startup/launch times? Was it an actual benchmark, or was it some manually timed run through a process it runs? What were the specifications of the machine it was running on? If its struggling to run the operating system by itself, then applications will invariably suffer. Please note: Just because he mentions that it was a, generic "dual core with 4gb of ram" earlier in the article, means in no way he did the tests on that machine.

The report he references from Adrian-Kingsley Hughes is skewed simply by the fact that anyone with common sense knows why upgrades take far longer than clean installs. "Well, hell, I have 600gb of data to migrate before the operating system installs, cataloguing it and making sure it is all in the correct places so my cool and unusual stuff is available when windows 7 is installed", and yet, it's obviously Windows 7's fault that you have so much data that must be migrated. Upgrade installs have absolutely NO bearing on the speed of the operating system, but on the type and quantity of data to be migrated. The fact that AKH actually believes this drivel, and makes other people believe it too, is boggling. The fact that XWJ has jumped on the bandwagon and decided to follow in AKH footsteps with barely relevant conjecture and anecdotal evidence, makes me have doubts over the relevant technique used to come up with such conclusions.

And, lastly, you again misinterpreted me when I spoke about the 250mb reduction. The os functions perfectly when compressed to such small amounts of memory, and is surprisingly fast considering a system with such low amount of memory would nowadays struggle to run most medium-large applications. Word 2007/2010 would use around 100mb by itself for example. The point was, there is nothing "bloated" about the operating system. When memory is there, it uses it. When programs request it, the operating system takes up less memory to allow the application to perform at its best. What is the point of having systems nowadays with 2+gb of ram, and yet only 25% of it is ever used? Hopefully that has un-boggled your mind in this respect.

Regards, Vigilante.

Posted by Vigilante on Sep 15, 2009 11:51 AM

Hey Vigilante. I work with Windows Xp Pro every day. I design Windows XP Embedded systems. I know the operating system quite well. I have been working with Windows XP Pro since its beta staging. Before that I worked with Windows NT 3.1, 3.51 and 4.0 and subsequently Win 2K. I'm also sad to report that I've had to work with all the Win9X variations and WFW 3.11.

I was not following in Adrian's footsteps. His article was a reminder to me of some of the work I had done previously. My focus is on what happens once the OS is up and running. His tests running upgrades confirmed my observations.

This is a blog. I don't particularly trust anything I see in the reviews simply because the relevance of a review is only an small indicator as to how well the OS or other application will work in the "real-world". I'm more interested in relating "how it worked for me". Tables containing piles of stats on Wheystones, floating point ops etc are really boring.

I have been amazed by users all the time when they tell me how they use my programs, processes and systems because they seldom operate it in the manner I think they operate them. I have never once had a user come up to me and tell me the floating point routines were too slow.

The test methodology is EXACTLY as I stated. I installed Windows XP Pro SP3 on a 4GB RAM equipped computer, did some measurements of the application software we sell to our customers. I wiped the drive and installed Win 7 Beta on the EXACT same hardware and tested the operating system with the application again. That eliminates any issues with hardware advantages one system would have over another.

This sort of test is the only truly relevant test to do with application software to determine OS suitability since the only thing that changes from one test to the next is the OS. Yes the drivers and the services figure into the test BUT those are parts of the OPERATING SYSTEM and that was what I was interested in testing. I was expecting Win7 to have some advantages over XP Pro. I was hoping it WOULD be faster, but Windows XP Pro SP3 beat Win7 Beta.

"Straight out of the box" Win7 Beta does not beat XP Pro SP3 "straight out of the box".

Now if I go in and strip out or at least shutdown all the extra crap loaded into the operating system that isn't relevant to our application and our service model I will be able to speed up Windows 7 when its released like I did with XP Pro.

A lot of the extra dross in Win7 is a total waste. Media-Sharing is the absolute LAST thing I would want installed on a computer used in the oilfield.

The only XP "service" that was "tasked with multiple things" is svchost which is specifically designed to be a "service helper" or a wrapper and is used to start networking functions like DHCP client, file sharing, printer services, TCP etc. That is why you'll see it multiple times in the process window in Task Manager.

Wireless services like those that are touted by Microsoft are only useful in those situations where you have a 3G or Wifi card shoved in a PCcard socket. Why install and run it in a desktop CPU that's plugged into a RJ45 connector on the wall? Even XP Pro loads it up when its absolutely not needed.

The biggest complaint I have with Microsoft is that they load up the operating system to cover every conceivable operating mode they can dream up. If they wanted to make their operating system work better, they would build in mechanisms that turn off services that are not needed or relevant and NOT require the user to go in manually and turn the junk services, drivers and applications off. Leaving the service stubs in place set in "Manual" is not the same as not being there at all.

Homegroup in Win7 which I have tested is Win95 Workgroups done a little better than the original 16 bit Windows for Workgroups LanManager based networking found in Windows 3.1 (not 3.1 NT!). It sure took them a long time to discover that networking without a Domain Server was the way a lot of their systems were being used. Workgroup has been a step-child for a long time. I'm not sure that Homegroup is any improvement since our current application ignores any sort of domain or workgroup functionality. It is a pain in the butt though at home since it doesn't work well with XP or Linux.

If you install a XP or Win7 printer driver, print server software gets installed even if you have indicated that you will not be sharing the printer. Why does the OS install components that are not needed?

If you happen to have a computer without a modem in it. XP Pro and Home installer still installs modem drivers, RAS and Windows Dialup services. Why?

This attitude extends even into the applications Microsoft sells. Installing Visual Studio 2008 installs SQL Server even when the programmer may never need SQL Server for any of his application programming projects. (Yes I program in VS dotNet.)

The relevant issue in operating systems used in the real world is the OS bloat and how that affects running the applications you need to do work.

Linux versus Windows is NOT the issue. Linux suffers from its own bloating syndromes.

Blindly being a fanboi-Troll is not a behaviour pattern that makes computers run faster. Just because some functionality included in the operating system was put there by Microsoft or Linus Torvalds is no reason to think its better than what's been previously released.

Posted by Xwindowsjunkie on Sep 15, 2009 1:18 PM

@Vigilante - Your assumption that I "haven't really been keeping up with Windows development post-xp" is incorrect. The truth is that I haven't been keeping up at all with Windows development post-Vista, after having been screwed, dumped on, shafted, misused and mistreated in pretty much every conceivable way by Vista on various computers, both my own and my company's. Before that I had been working with every Windows version since 3.0, and I literally mean every single one.

To jump from that to the conclusion that "I doubt that you are an IT professional with any real merit" is pretty amazing, unless you assume that the only definition if "IT with any real merit" means drinking the Microsoft Kool-Aid. But what I find even more amazing is the fact that Microsoft, its employees and its mindless supporters and promoters (also known as Vigilantes) absolutely refuse to utter the word "Vista". They desperately want the world to forget that they ever produced such a monstrosity, and they are apparently terrified that people will realize that Windows 7 is nothing more than Vista with Lipstick. This is selective amnesia is going to be particularly important when it comes time for consumers who already paid for Vista to turn around and pay again for the patches which have been bundled up and relabeled as "Windows 7".

jw

Posted by J.A. Watson on Sep 15, 2009 2:00 PM

Clearly JW has had a bad experience with Vista as have many others, but jumping online to post bullsh!t on every Windows 7 article is really overkill.
In reality, anyone having even tested the RC of Vista properly would have already known how crap it was before they bought the damn thing. Practically EVERY article written about Vista prior to it's launch was laying into just how bad it was, so what does that say about all of the idiots who still went out any installed it? The majority of people buying new machines with Vista pre-installed wouldn't have a clue whether it was better, worse or the same as XP / 2000 because they only buy a PC so that they can waste their sad little lives on facebook.

Posted by 50943 on Sep 15, 2009 3:38 PM

This comment has been deleted at the users request

Updated by 50943 on Sep 15, 2009 3:40 PM

Oh! Of COURSE! It's those blasted "idiots" who insist on giving Microsoft their money! It's all their fault! But of course, THIS time Microsoft got it right, for sure, absolutely guaranteed, so those who go back and give Microsoft even MORE money are not "idiots" any more, they are... uh... hmmm... "gullible"? "suckers"? Or just ordinary people trying to buy a computer which works, and who are FORCED to buy the computer with Windows pre-installed? Or perhaps those who believe all the hype that Microsoft puts out and most of the press simply repeats verbatim, when anyone who dares say anything to the contrary is shouted down with cries of heresy and "overkill"?

Exactly where is the boundary of "overkill", by the way? If there are 100 articles posted singing the praises of Windows 7, and 5 are posted similar to this one (from Xwindowsjunkie by the way, not from me) saying that perhaps Windows 7 isn't the best operating system in the Universe, is that "overkill"? Or would 10 be "overkill"? Or is the definition of "overkill" the point where the number is large enough that it actually stands some chance of being heard above the din of the Microsoft worshipers?

One thing is still not clear to me, though. Are the "majority of people buying new machines..." with Windows 7 pre-installed also clueless and "wasting their sad little lives on facebook"? Or are they the real geniuses, who can somehow divine that Microsoft really did "get it right" this time by putting Lipstick on Vista and calling it Windows 7?

jw

Posted by J.A. Watson on Sep 15, 2009 6:42 PM

Alrighty, a few things I'd like to clear up then since none of my (completely relevant) criticisms have actually been addressed here.

Firstly: @ XWJ

The problem I had was that you are testing a single application, which was (obviously, due to the nature of its heritage in running on windows XP), designed with windows XP in mind. Which means, in all possibility, the code is based on an 8+ year old operating system in a time when resources were limited and very little emphasis was placed on secure, forward thinking coding techniques. Anyone remember the megahertz wars? Good times for lazy programmers.

That's all fine then, if you keep that kind of test and its results to in house records, perhaps to use for analysis as to WHY it's happening rather than making assumptions. However, when you attempt to say that the entire operating system "fails" because one single application doesn't necessarily use it to it's fullest, I do begin to wonder whether the problem isn't really software or hardware based, but rather, more a case of PEBKAC.

I will completely agree with JW that Vista was far too hyped, far too slow, far too late and far, far too buggy, well into production and post-sp1 states. But, just like previous "blunders", I'm thinking Windows ME here, it was necessary to allow people, programmers and software developers, to know what kind of system to prepare and develop for in future. I'm 100% sure that, had your application that runs just 5% slower on Windows 7 had actually been created having taken notice of new technologies and tools available over the past 5 years, the performance difference would be negligible at worst. Let's face it, Microsoft was never going to revert back to an xp-style codebase just because Vista was particularly terrible, it cost them far too much money and was, in reality, a far superior bit of code. Taking that further, we now have Windows 7, an evolution of windows Vista, which as near-100% compatibility with all vista drivers and software, yet has almost no redemptive issues, which the majority of the IT world has so far, embraced and objectively considered based on multiple software evaluations and other more beneficial metrics for real end users.

The REAL issue here my friend, is that you assume that because your single piece of software performs 5% worse on Windows 7 than on Windows XP, that all software will. And unfortunately for your argument, that is not the case.

As for JW, as 50943 pointed out, had you REALLY been following the information (Even before "post Vista" as you describe), then you would have actually tested Vista before deploying it on your machines, ESPECIALLY after all the information available regarding it. The only person I really see "drinking the Microsoft Kool-Aid" was yourself, by not objectively and logically examining the possible repercussions of placing an untested operating system into your corporate environment. I'd love to say that I feel sorry for you, but in all honesty, we both know that would be a lie. However, you can now be thankful that, thanks to yourself and others like you, we all now have an operating system worth its share of that 80+ percent software ecosystem. Thanks, from me personally, and the rest of the windows world. We owe you one. I'm just sorry you're choosing not to be a part of it.

Updated by Vigilante on Sep 18, 2009 9:41 AM

This comment has been deleted at the users request

Updated by Xwindowsjunkie on Sep 15, 2009 9:35 PM

> The REAL issue here my friend, is that you assume that because your single piece of software performs 5% worse on Windows 7 than on Windows XP, that all software will.

We don't all have the time to run a complete suite of real-world test applications, so very often we have to rely on our own anecdotal experience. However, I fear that many people will find the Windows 7 Bloat Factor affects the applications they use in the same -5% way.

Time was when a new version meant faster, better, more efficient (oh, surprise, it still does: http://bit.ly/kxI0A ;)

Posted by Jake Rayson on Oct 9, 2009 12:15 PM

:s yup they was a time, oh how i miss the glory days.

Posted by CA on Oct 9, 2009 11:17 PM