NoelC

Don't feel bad. Earlier today I was told to "take off my tin foil hat". Then KB3035583 (the "GWX" shill update), which I had previously hidden when it was an optional update magically reappeared today as an important update. So let me get this straight, if I avoid upgrading to Windows 10 in the initial rush, I don't have to deal with an infantile game? Sounds good to me. Waiting - possibly a year or longer - to move to Windows 10 has been my plan all along and this just reinforces the efficacy of that plan. What's next, Microsoft paying customers to upgrade? -Noel

Well, given that I probably won't ever actually run Win 10 on that system except for testing, it doesn't really matter. I put Win 10 TP on the internal backup drive before I received the SSD hardware from which it now boots Win 7, just to have something to run and test the system with. Depending on which drive I put first in the BIOS boot order, I can boot it up on either system now. -Noel

I take it, then, that you have no interest in discussing the actual subject at hand, nor contributing test results. The invitation is open to anyone. -Noel

By the way, the system from which I published the PerformanceTest results above cost a whole $299. Not exactly high end. The GPU is whatever one the Pentium G3220 has in it - pretty much the definition of a low end system. And, while the OS install is not bone stock, both Win 7 and Win 10 have the SAME level of tweaking, and they're not augmented with much of the litany of 3rd party software I normally use on my workstations as its intent is to sit quietly in a corner and serve files. -Noel

Oh, and by the way, this isn't just about web browsing. If www.frickingparasite.com is blocked via hosts, then attempted connections by name from within malware (e.g., to send your skimmed password home) will be blocked as well. -Noel

Oh, hey, just stop with the nitpicky bul*****. If you want to have a meaningful conversation, talk about something other than what who said when and whether your comments on the subject were or were not an opinion. It's tiring, and all the posts are right out there for everyone to read. -Noel

More info: I booted up a Windows XP SP3 (32 bit) VM.Updated the hosts file (with the same one I use on my Win 8.1 system).Installed the latest Windows Update (Windows Malicious Software Removal Tool) and an update to my antivirus software, then rebooted.Brought up IE8 and saw that it displayed my home page data immediately.Cleared the DNS cache.Brought up IE and saw that it displayed my home page data immediately. Using a stopwatch, the time between display of the IE window with white background and display of the web site data was 0.4 secondsTyped in another URL. Same timing.Moved the 500+kbyte hosts file out of the way.Rebooted.Did the same tests and saw no visible improvement in responsiveness. In fact, I measured 2 seconds to display Apple.com, where it had only taken 1.8 seconds with the hosts file in place. This could be human error.Restored the big hosts file.Rebooted again. Times to boot and login (18 seconds and 3 seconds) were the same with or without the hosts file. I don't see a way to start a web tracing tool in the F12 developer tools in IE8 to give millisecond accuracy, but in all observable ways Windows XP is working equivalently with or without a half-megabyte hosts file. If you want to discuss older systems or smaller hardware, reasonably speaking that's outside the scope of this thread. -Noel

It seems to me you may focus too much on what people say. I pay more attention to how things actually work. Tarun has said there'll be a timeout, even with 0.0.0.0. I see no evidence of that. Connections are aborted immediately, per traces from F12 developer tools in IE. Tarun has said that it causes problems with other software. While that can be true at the highest level, understand that it will be the case if that software required a connection to a site that's considered a parasite by the manager of the mvps hosts file and didn't handle an inability to reach the server. That would be, in basic terms, a BUG in the application's web programming. Personally I've seen it block Avast antivirus from gathering tracking information during an update, and when called on it Avast made changes to their product to not fail if they couldn't gather tracking info. I've also seen it block tracking information in some of Adobe Photoshop's web code, and upon reporting it I have received word that they'll be fixing that oversight as well. In practice, it's revealing things you didn't know were happening and may not WANT happening. Tarun has said it's a misuse of the hosts file. I disagree. The intent of that file is to redirect certain names to particular addresses. That's EXACTLY what's being done. I say that redirecting parasite web site names to an address that can't possibly respond is an excellent use of the capability the operating system is providing. Tarun and Jaclaz (indirectly) have said that there will be performance degradation. I'm unable to measure a performance degradation on $10K and $299 computers running Windows 8.1 and 7. Are we talking about some ancient system here?By the way, I misread the advice on the MVPS site. I read it before as saying "there is no need to change the DNS Client service". I see now that it says otherwise. I do agree that it's bad advice. For what it's worth, I never have changed the default status of the DNS Client service. It's always worked just fine. There's a LOT of "rule of thumb" info out there that's been accumulated over the years and which is simply no longer pertinent. Maybe it used to cause problems in Windows ME or something. Or maybe it's just unfounded lore - there's plenty of that out there too. You have a computer. Instead of blabbering about what other people said and when, do some tests. I have done my part, using both a $10K computer and a $299 computer and two different operating systems - and showed that there's no measureable degradation. There's also no instability either. My systems all run without fault between reboots mandated by Windows Updates. So I ask again: Ignoring the lore, where's the measurable downside? -Noel

Or maybe seek a setting to disable the lesser one (with the shared memory) entirely. This, of course, could affect battery life if it's a portable system. -Noel

It kind of doesn't matter how big or small the hardware is when the relative measurements on the same hardware show a slowdown. And I kind of disagree that it's important how a "default" or "untouched" install runs? What seems to matter most to me is how well it can run when well set up. Yes, you could argue that most folks' experience is with un-tweaked systems. For what it's worth, way back when I did find some things faster on a (well-tuned) Win 7 x64 system vs. its predecessor, (a well set up) Vista x64. On the other hand, Vista ran like a pig by comparison to XP x64, and I had to throw out a $1,000+ workstation video card and buy a newer gamer card just to support it. I don't care to compare Win 7 x64 against XP x64, as I really never did head to head testing. I suspect you're right, since XP was trimmer it probably did many things more efficiently. On the other hand, a modern Win 7 or 8.1 system supported by a good GPU is arguably more responsive than XP ever was on the systems of its day. There is no question each new operating system version requires more resources - by absolute measurement. But by cost? By performance? Not so much. A modern $1,000 computer packs a LOT more RAM, disk storage, and processor speed than a similarly priced machine from a decade ago. And while I agree that the code is getting "fluffier" I don't think it's all fluff. There actually ARE some things the new systems do better than the old ones did. Depending on point of view, quite a few things. Consider advancements in display tech, for example. Or the fact that, while XP was pretty stable, it still did exhaust its resources after a time and need at least a log off/on, if not a reboot every so often. Some of those bugs have been fixed since then. We can pine for a system that "might have been", but sadly Digital Equipment Corp. is long gone. What we have today are the two architectures (derived from VMS or derived from Unix) to choose from, and little likelihood of anything to replace them any time soon. It's really a value proposition where "the lesser of the evils" must be evaluated. I figure the best thing to do is try to make the best of what we have. -Noel Edit: Microsoft would like us to believe that the elegance (e.g., glass effect) has been removed from the desktop to enhance performance. Why, then, do we see these kinds of results? This is a test of a mostly-stock Win 7 vs. mostly-stock Win 10 build 10074 on a 2 core 4 GB machine with Intel on-chip graphics I've recently acquired. This shows a pretty significant degradation between Win 7 x64 (red) and Win 10 (blue) that can actually be felt in the UI. In 6 years of OS development only the display of images has gotten better; everything else is noticeably worse. Of course a Marketeer would focus on the Image Rendering result above and say, "See? Windows 10 IS faster!" -Noel

For what it's worth, a lot of us thought that when they eliminated the cascading menus. I take it Win 8.1 is not a contender for you right now. -Noel

Here are my measurements, by the way... Before each test I flushed the DNS Resolver Cache to simulate a worst case scenario. I can repeat these results with small variations in the timing numbers. Note the response times for the first accesses to the site. Big 12 core 48 GB workstation running Win 8.1 x64. Smaller 2 core 4 GB computer running Win 7 x64. Do we need to be talking about even older or lesser systems? I'd have to test using virtual machines. I can do so if it's interesting. -Noel

A note removed in 2013 might have been there for a very long time. Computers have gotten faster (and yes, that statement implicitly excludes those still running hardware from yesteryear). The fact is, that statement has been removed - I suspect because it's no longer viable. And, in case people don't go read, right now the web page says, "...there is no need to turn on, adjust or change any settings with the exception of the DNS Client service..." Jaclaz, do *you* have any suggestions for how to measure "slowdowns" or indeed any downside? You know I'm all about using real, measured data to make decisions on performance. With such information I'd imagine a number of us could do some testing and post our results here. THAT is the power of a forum. Pissing contests are useless (especially against Italians ). FWIW, the mvps hosts file is now over 500 kB. It seems to me that if it were going to make problems I'd be seeing them - which I'm clearly not. But as you noted, I tend to run big systems. Have *you* seen actual problems with it? Finally, I don't buy that a big hosts file causes problems on even a lesser computer. The burden is on those who say it does to show it. I've been using the mvps hosts file for a long time, on systems that were in their day high-end but really were quite tiny and pale by comparison to even a modest computer system today. Granted, the list probably wasn't as big back then, but it was big enough to be effective. FYI, I have it also on another, smaller system with only 4 GB of RAM running Win 7. That system boots up quickly and also navigates to sites it's never been to before in a few 10s of milliseconds. Internet speed tests max out the fiber optic link I have. Where are the problems? -Noel

But it IS faster - all the Microsoft shills say it is, so it MUST be so, right? Argh. Frankly I find it INCREDIBLY frustrating that they keep talking out their butts about Windows 10 being faster. The ONLY friggin' way it's faster than Win 7 is that computers today are generally faster than computers of 2009. Even relabeling hibernation as shutdown and bootup is a scam. My new system boots to the Windows 7 desktop just as the swirling parts of the flag come together. Under 15 seconds, and within 1 second of the speed of booting up Windows 10 with hybrid boot disabled. Actually, when you actually take the time to measure it, Windows 10 is slower or no faster than Windows 7 across the board, and runs the processor measurably hotter to do the same things. I just did more Passmark benchmarking. The desktop doesn't seem any more responsive, and in fact less. It also has an irritating tendency to ignore input occasionally. Fanboys claiming Windows 10 is faster are probably judging a new, fresly installed system against an old loaded-up one. Almost no one knows how to keep Windows from loading up with junkware they don't need running. -Noel

The point is, it's effective, and notwithstanding your inferences simply without practical downside on a modern system. A managed blacklist is an extremely effective means of blocking bad sites. IMO, the mvps list in particular is well-managed. And let's be clear, I don't advocate doing things as you have described to the DNS cache service. Though I started this thread with some general sounding recommendations, I mean to suggest that people just load the particular mvps hosts file. And I'm sorry, even though you bill yourself as an expert, I'm one as well. I simply won't accept a blanket statement like "the Hosts file is loaded into your memory (RAM) at startup, another unnecessary performance degradation" without measurements to back it up. I've already posted my measurements supporting my argument... If you want to debate this further, please make some suggestions for capturing actual measurements of degradation. -Noel

By the way, not to interrupt a jab-Microsoft-fest, but Win 10 can have a desirable look, with a few tweaks... -Noel

No no no... THEY need you to send 15 GB of valuable data thousands of miles. How are they supposed to eat or buy basketball teams otherwise? -Noel

I'm sure the 3rd party software mentioned is good, but don't abandon the possibility of using Microsoft's own Windows Backup to do System Image backups of all critical partitions. It's integrated with Volume Shadow copy Services, and thus with the Previous Versions feature, as well as available to you through the Windows recovery environment (WinRE) at bootup. I used Windows Backup for many years and both recovered from failure and upgraded to new bare metal hardware by restoring such backups. -Noel

+1 for Classic Shell, set to Classic mode. An advantage Classic Shell brings to the party is that it works in Windows 7, 8, 8.1, and 10. Learn it once, and you're done. And it's still actually a better implementation than any of Microsoft's own, IMO. Shown here on Win 10 as an example... -Noel

For me Aero Glass runs just fine in 10074 via a scheduled task that starts AeroHost.exe. It must be run in the SYSTEM account with Highest Privileges, of course. Here's what's in my AeroGlass folder. You can compare dates/sizes with yours... Here's the entry I use to schedule AeroHost: As I recall there may be some issues with using the Task Scheduler in a build 10074 that has been installed as an update, vs. a clean install. Mine's a clean install. -Noel

Probably a minor bug. I noticed that when I was doing my captures I had to expand some of the subheadings. -Noel

This is from memory, but I *think* you can look for information in your Aero Glass debug.log about where the symbols are to be downloaded from. It's possible you could get (or develop) a URL from there that you could use to check to see if the symbols are on the symbol server before reinstalling the KB. Or you could occasionally reinstall it, see if it works, and uninstall it again if the symbols don't automatically download. -Noel

There is a current Windows Update, KB3048043 released this week, that updates DWMCore.dll. I suspect that's what's hindering the Aero Glass product. You could uninstall it and wait for Microsoft to put the debug symbols online. They have done so, often with a delay after the KB comes out, for all versions so far. Personally I don't usually install Windows Updates the day they come out... Better to let other people test them first. I look online for reports from others having installed them. If there are no screams of anguish (or reports that Aero Glass doesn't work) after a while, the updates go in. I go read about what each and every one proposes to do to my computer - which is how I knew that KB3048043 seeks to update DWMCore.dll to 6.3.9600.17795. Making a judgment about whether to install a particular update right away - or wait - based on what problems it purports to fix (in this case a black screen while moving tiles around, which I don't see) is a good idea. It's probably good to install most all updates at some point sooner or later, but if you aren't suffering from a particular problem there's no rush. -Noel

I've not seen it on any other threads, just that one, and it only recently started happening. I've confirmed it with IE, Safari, and Firefox, though with the latter two it only lasts for a second or two, so it may not be noticeable. I suspect IE may be thrashing in a script somewhere, similar to what happens with long Microsoft forum threads. That's what made me mention the possibility it has to do with the length of the text. IE's network monitor shows the last thing requested is /board/favicon.ico. Then there's a 15 second delay, then the page will then respond to scrolling. None of the requests stretches out to the end time; this implies it's something that's running locally in the browser causing the delay. -Noel

I wish you had chosen to expand the Clocks field in the first screen grab. It's possible all of this is because the power management logic has changed a good bit. Were there cases where a powerful machine refused to speed up with intensive operations with the older OS? I don't recall coming across any personally, but it's the kind of thing that's hard to sense. If it takes a few extra seconds to start something that normally starts up instantly, we all just sort of wait it out. -Noel