Slashdot Mirror
Early Speed Tests For Windows 8
adeelarshad82 writes "You often hear in the software industry that performance optimization is one of the last steps in the software development process. That bodes well for Windows 8, considering at the early stage of Developer Preview—even before we've seen an actual beta—the nascent operating system is getting widespread praise for its performance, particularly in startup times. Anecdotal evidence is always encouraging, but PCMag decided to run some very early tests on the OS to see if the reports were wishful thinking or if there was a real, measurable boost in speed. Along with startup and shutdown times, they used several standard industry benchmarks to compare Windows 8 performance with that of Windows 7 running on the same machine."
There are actually two kind of performances, which are both important. The real, actual performance, and how well the OS can make the system feel even under load. It's important to have a snappy feel even if the system underneath is working hard, and this is especially true now that the amount of cores in CPUs and multithreading are increasing. Say what you want, but just the feel of speediness is an important factor. This is why the boot up time is looked at so much too - it's great to quickly get to the desktop, and let the OS load up while you're already started working.
One thing I've noticed with boot up times (and this applies to all operating systems) is that the OS tries to load all programs at once. Usually the limiting factor to this will be hard drive. It's less true with SSD drives, but it's really noticeable with 7200 RPM and slower drives. It usually leads to the whole system crawling for a few minutes after desktop shows up. It would be great if the OS would measure the different loads and UI response times, and actually limit the startup programs. This way you could open your browser and other tools and those would be given priority upon startup process.
I tested the developer preview version briefly and it sure seemed a lot faster and snappier. The startup time is remarkably faster. And according to this PCMag test, seems like the overall speed has been improved a lot too. Good job MS!
This leaves out of the picture that clean installs of any OS are going to be rather fast to start and shutdown, the issues begin when installations get bogged down.
I would have compared the boot-time of Linux to this, but they never state what harddrive they're using, rendering any compare worthless.
...on second thought, let's just compare that against my 1.8Ghz Core2Duo, 2GiB RAM Notebook: Debian Wheezy needs 4 seconds to boot. :O
From the article
Benchmark Windows 7 Windows 8 Percent improvement
Startup time (min:sec) 1:32 0:32 +65
Shutdown Time 10 11 -10
Old hobbits die hard.
Not necessarily. It wouldn't be the first time things have been performance tuned before they're actually working properly.
I heard the opposite.
Does anyone else detect a whiff of shill in the air?
Confucius say, "Find worm in apple - bad. Find half a worm - worse."
The startup time should be faster if you used suspend instead of shutdown, Fedora 15 on my machine with 2GiB RAM, 7200rpm SATA 2 drives and i3 CPU @ 2.93 GHZ, starts back up to a desktop very quickly indeed, But I agree once they load it down with all of the services and other crap that Windows comes with then it will be slower than in these tests. And I did read the article.
Hopefully the Windows classic interface is an option.
liberare massarum ex ignorantia, clausa descendit molestie.
...you boot the bastard on a desktop machine, and then it goes to that horrid Metro screen which makes navigating with a mouse and keyboard painful? It may be fine for touch, but without touch, man....it makes you want to break things.
Then you talk to a Microsoft turfer, as seen on here and other places, they will bald-faced lie to you and say "well, it's not finished yet, who knows what it will be like?"
Then you go to the Microsoft fora and ask Microsoft employees about Metro as being standard for the upcoming desktop, they double-down on it.
Guys, get your friggin' stories straight. All I know for sure is that Metro without touch is a steaming load of bovine excrement backlit by the morning sun so you can see the vapors wafting off it. Fix it.
--
BMO
More than half a minute including BIOS boot, which according to the article is 15 seconds.
History suggest you'll have a lot more chance of finding it on him than on Windows.
surely we should bne thinking about time to be usuable once logged in? We all know Windows doesn't start alot of things till after user login (first used in NT4), so this is what we should be measuring, not how quick it gets to login screen. Reading the article they COULD be using this test but it's not clear.
Also things like with WIndows you NEED some sort of anti-virus installed as well so again not that real world, but looking encouraging and we'll see how many of the extra features not yet implemented impact this.
I've found Windows 7 64 to be an excellent desktop operating system.
I have to agree with this. Windows 7 is a major step up from XP* and very hard to find fault technically. I can choose any OS but I'm using Windows 7.
[*] Mostly due to XP's memory manager being completely crap but that's another story.
No sig today...
I wish the test was done with a weaker machine. I guess the gap is achieved by a better usage of higher HW/FW which will not be noticeable in a non-quad-core machine.
Actually if you want to know what is generally the cause of Windows being slow its all that OEM trialware crapola that gets loaded onto a machine before you ever get it. I consider the new Asus EEE I got pretty light in that it only had nine extra things running at startup, of which only two I found useful (hybrid engine and Asus Hotkey) whereas I've seen as many as nineteen on some dells and HPs. That is why PC Decrapifier is a handy tool to have around.
The second thing that slows Windows down is what I call "granny services" which thankfully MSFT is FINALLY fixing in Windows 8. granny services are the services that MSFT or the OEMs have running to keep granny from calling tech support. That's support for cameras and scanners,media sharing services, etc. It took them a fricking decade but they are finally gonna have services launch by trigger instead of the usual auto/manual crap.
I only hope the new services setting is backported to Windows 7 as many of us have settled into Windows 7 and won't be making the switch for quite awhile. I know in my case i've just finished getting the last of my customers moved off of XP and I doubt seriously any of them will be too keen on jumping onto a new OS next year. With Windows 7 being supported until 2020 it could become the new XP, which while i'm sure that wouldn't make MSFT none too happy without a killer app to make all these multicores obsolete I just don't see folks switching every couple of years like we did in the 90s. Back then thanks to the MHz wars it was worth your while to switch thanks to the huge increases in performance, but now? Frankly any dual core is "good enough" for the majority of the things your average Joe is doing with a PC.
So while I'm happy that MSFT is FINALLY listening to users and making speed a priority I have a feeling windows 8 adoption will be even slower than 7 was. Anybody who has gotten a new PC in the past 4 years frankly has more than enough power to do whatever they want. Why would they switch? Frankly if the best reason they have is Metro and a few speed increases they are gonna be looking at some slow adoption rates IMHO. For most of the new machines I've seen in the end its the HDD not the OS that ends up the bottleneck anyway.
ACs don't waste your time replying, your posts are never seen by me.
You often hear in the software industry that performance optimization is one of the last steps in the software development process.
No you don't, not among sane people. You don't do performance optimization as "one of the last steps" shortly before shipping.
What you hear is that "premature optimization is the root of all evil" (quoting Donald Knuth). What he meant is that you should not bother with complicated performance optimizations when designing the code. Rather, create and implement a good clean design, then test performance and optimize where needed. On the other hand, algorithm choice is one of the biggest performance contributors and initial choices will often be made quite early, so one cannot apply this quote blindly. Read here: http://en.wikipedia.org/wiki/Program_optimization
"When I first heard Daydream Nation it quite frankly scared the living shit out of me." -- Matthew Stearns
Startup time, until login screen, or desktop, or usable system, or fully usable system ....?
The four are different and most people assume you mean the last, when most are measured to the first .... ...and unless you have a laptop why are you turning the machine on and off enough to worry about boot times (in the real world the difference between 20s and 1 minutes is a vast gulf, the difference between 1 minute and 5 minutes is irrelevant)
Puteulanus fenestra mortis
Ideally, a modern desktop OS should be booted once. The rest of the time it should be slipping in and out of sleep.
In practice it seems that a few months between reboots (for OS updates) is easily attainable on some platforms.
When a reboot occurs once every few months, boot time is not terribly important.
I can't help but think that people who marvel at improved boot times are rebooting their machines too much.
Am I the only one that doesn't care about startup times? My debian servers only need to be rebooted for a kernel upgrade. For how infrequently I need to restart, startup times are a minor issue.
I can't say the same for Windows, which requires a restart on half the Microsoft Updates that are installed, many software installations, and of course crashes.
My experience is quite the opposite, when transferring large files from windows to Linux (over a gigabit link) or vice versa it's always the Windows machine that chokes on the IO trashing like a madman (seen with, Debian/Gentoo and Windows XP/Vista/7 with either ReiserFS/ext3/ext4 vs NTFS).
My guess is your distribution isn't set up properly for your workload.
What burns my shorts these days is not the Windows reboot.
It's the automatic updates that only get applied at the end. "PLEASE PLEASE OH PLEASE DON'T TURN ME OFF BECAUSE I WILL FUCK YOUR COMPUTER IF YOU TURN ME OFF IN THE MIDDLE OF UPDATES"
For 20 minutes.
--
BMO
Couldn't have said it better myself.
Zero power consumption with session written to disk, little power consumption with suspend to ram. No need to shut down. Desktop users just need to be educated.
Figures, I just started using Win7 now the talks are of Win8.
XP 64bit has been good to me. I was forced to run Win7
as "Battle Field 3" requires DX11. I still only boot into
Win7 when playing BF3.
I've always assumed the next Windows OS was going to be a
touch screen. Don't know why it comes as a surprise.
Remember the roll out for Surface?
http://www.microsoft.com/surface/en/us/default.aspx
I figure parts-n-pieces will surface [shrug] in Win8.
Kind of a lopsided review; a 3.4GHz quad-core PC with 4GB RAM
Got the power, just no Ram.
About year ago, I tuned my linux box (cheap single-core Intel Atom board, 5400rpm HDD, 1GB RAM) to boot into X in 8s (measured between hitting ENTER key and displaying search results for "asdf" on google in Opera browser). Video: http://www.youtube.com/watch?v=zE_QRZwNGOs
Spoken like a true Windows shill.
My personal experince is that Windows is fragile during updates. YMMV, but I have learned the hard way not to fuck with Windows updating itself or shutting it off while it's doing something "important." Many other people have been conditioned over the years by the exact same things.
I don't care if you say the beach is safe to surf, Kilgore. I don't trust you and nor do I trust Windows.
>Comparing Windows and Linux updates
There is no comparison. Package managers for Linux are quite robust and will pick up exactly where you left off if you so much as hit control-C and then restart. Powering down in the middle would likely fuck /something/ temporarily, but unless you're doing a kernel update, there is likely no reason for a machine to be unbootable. And even then there are the backup kernels you can pick on boot from the handy-dandy boot screen.
>Losing power during write, forcing a fsck
>taking a long time
I haven't had a fsck take longer than a minute since I went to a journaling file system last century, and journaling file systems on Linux are standard issue now. I also noticed when I did an install of Ubuntu 11.10 that btrfs is available. This is spectacular, but I'm sticking with ext4 for now because it's thoroughly debugged and I can trust it.
Sorry to break it to you, shill, but you're full of hot air and you need a tic-tac.
--
BMO
Try: sudo dpkg-reconfigure -a
At least that's always worked for me when I need it to, should power fail, etc. Not to mention ext3/ext4 journaling seems much nicer than using NTFS and having to fallback to CHKDSK when such issues arise, (along with the occasional pre-emptive NTFS defrag).
Personally I find the overall cost of Windows as being too costly to use in my business.
man page for dpkg-reconfigure
You can't be ahead of the curve, if you're stuck in a loop.
Making Windows more optimal might signal a turn of focus toward phones and tablets.
well considering this is version 8... i'd be expecting this to be right near the end of the line and pretty well fully optimized
You should have worded it "I don't run operating systems often, but when I do... I prefer Windows 7."
"Stay cache thrashed, my friends."
Karnal
And in the UK its not cheap so yes, I shut my desktop down when I'm not using it and fully switch off the monitor. Aside from that it has an enviromental benefit downstream at the power station. Sleep mode might not use much power but it still uses some and with all the computers in the world that probably adds up to a lot of CO2 for nothing.
If I remember correctly, the Windows 7 beta performed considerably faster than either the RC or finally the RTM. Windows 8 could actually get slower as development progresses.
"Old hobbits die hard."
Was he taking Viagra?
Vote monkeys into Congress. They are cheaper and more trustworthy.
Well, then jeeze, for fuck's sake, turn auto update off, problem solved.
OF COURSE the test build of Windows 8 runs wickedly quick. Can't you read? It's an early Developer Preview, it's not even a beta yet. They haven't packed it full of the standard train-load of unnecessary services, buggy features, assorted DRM layers, and other miscellaneous bling, crapware, and patented Microsoft Goodness. And by god when it ships, it better have touch-screen enabled by defaultâ¦
I want to mod this insightful, but I haven't played with the preview yet to know if it's actually accurate or not.
My sister opened a computer store in Hawaii. She sells C shells by the seashore.
Well, it's not so straightforward with 7 (2008 and Vista too). Yes, desktop performance & experience is great, but the abomination that WinSxS folder is fucks it up rather ruthlessly for VM, laptop and SSD usage. As it is, there is no way to strip it down to bare minimum and run lightweight.
I have VMs running XP on 2 gigs of disk and 256mb of RAM. Let me see you do that with any of the above mentioned. And don't tell me disk is cheap, because those VMs number in tens for me, and probably hundreds for other people.
Now how about my laptop with Windows partition of 25 gigs consisting of 17 gigs of Windows and tiny program files of 1.7 gigs? I'm fucking scared to run Windows update on it.
"Blah blah blah." - [citation needed]
Provided that Windows can/may download updates during the active session (when the user is actually using his computer),
how come actual updates take so long, while the OS is in mono-task mode, pending for a complete halt, with no user operation in the way?
Slashdot, fix the reply notifications... You won't get away with it...
If it is a desktop, just walk away. If it is a laptop, your battery should last long enough to do the update, so you can just take it with you. What I REALLY hate is going for coffee while waiting for a logon screen, and then you have to wait 2 minutes AGAIN before it becomes stable. Why can't it preload all the important files if it becomes idle while waiting for a logon?
10 ?"Hello World" life was simple then
One word: laptops. Sleep mode still uses power. Hibernate can work, though it usually takes a while anyway. I rarely turn off my Windows machine because it wants to restart.
Thanks for posting from a parallel universe. Could you please upload that magical version of the software you are using somewhere where the guys from Redmond can find it? Thanks.
In return we can send you versions of unix from the 1980s onward that can stop and start just about anything at any time. Our version of Microsoft even had a quite decent version called Xenix that could do that sort of thing.
Give Microsoft time to stick some horribly designed bells and whistles on to destroy any performance gains you're seeing now.
Actually, they shouldn't be including the BIOS time into the boot time because that has nothing to do with the OS.
It should read more like
Benchmark Windows 7 Windows 8 Percent improvement
Startup time (min:sec) 1:17 0:17 +78
78% improvement
Yah, I can't say that I care much about start-up times
Me neither : uptime 07:55:31 up 1532 days, 18:50, 1 user, load average: 0.00, 0.00, 0.00 :)
I must confess that I'm a little behind on kernel updates
Avoid the MS tax, always buy I.B.M. PC's (I Built-it Myself)
So.. you mean it was in suspend and woke up tongue X screen in 8 seconds? My laptop does that faster. Unless you meant it was completely powered down, nothing store in RAM, etc and you pushed the power on button (which may double as an enter key, I have no clue)
It was clean boot, no suspend. By ENTER I meant enter in grub menu.
It spends a lot of its time doing transactions to make sure it can roll-back any changes in the case of failure. Also, being changes to system files, you have to serially modify/patch because you don't want to chance a race condition and fubar the system files.
I suspect that it is all the redundant code used in Windows, where the same fix has to be applied in 20,000 places across the OS. You also have the insane complexity of the registry, and that is where the problems come from when it comes to updates. I suspect that Microsoft does not use a lot of shared library calls, and there is a lack of anyone looking over the code to see where multiple functions could be consolidated down into a single function.
Then again, things like .NET are not small, so the time it takes to copy the files into place and update the registry will take a bit of time with a regular hard drive.
Even with a laptop, I mostly use hibernate. Amount of time to come out of hibernate, even on Windows 7 is only about 20 seconds after BIOS. And that gives you fully usable system, with all the programs you had open previously. Reboot time is absolutely worthless, because I almost never reboot anymore. Once every couple of weeks for updates.
Anthropic principle: We see the universe the way it is because if it were different we would not be here to see it.
Why can't all software do as many of the current web browsers do and remember their state so they can come back up exactly where they were if they are suddenly or even gracefully shut down?
Indeed. There should be some kind of operating system service which could automatically restart any application which were running when the system was shut down. Perhaps it could even interact with them and restore their state such as documents and cursor positions? Oh wait: http://msdn.microsoft.com/en-us/library/aa373524(VS.85).aspx
The restart manager - if used by the application or service - will actually keep track of files which have been scheduled for updating (new version) during restart. If the files are being kept open by an application or service which has registered with Restart Manager, RM can restart the application and avoid a system restart. Note, this replace is transactional - it only replaces a set of files when all files can be closed by restarting processes.
If the processes are closed for some other reason and RM suddenly finds that all the files in a replace set now can be safely replaced, it will proceed to do so. Have you ever noticed that the start menu says that you should restart to install update, only later to find that this notification has mysteriously been removed again? That's Restart Manager jumping in a replacing files.
Reading slashdot one-liner: (irm http://rss.slashdot.org/Slashdot/slashdot).rdf.item | fl title,desc*
...not a great idea for an OS that is a malware magnet.
Plus you just placed another burden on the n00b user. They have to figure out what you just told them to do.
Updates in both MacOS and Windows are far too disruptive for systems released in this day and age.
A Pirate and a Puritan look the same on a balance sheet.
Comment removed based on user account deletion
How about when it decides to force a restart of your computer, and It's Just Too Bad you were in the middle of a fullscreen application that hid the restart notification until the timer expired and your application was force quit on the way to update-land?
Yeah, I'll reboot at the end of the day, if you ask politely like Linux does.
I'm running a board with UEFI and I couldn't agree more. I have about 20 seconds from press the power switch til I see the Windows 7 start screen. Which then shows the desktop and is usable in about 14 or so seconds thanks to the SSD.
One thing I did notice is that my older laptop I have for work has a BIOS that only take around 3-4 seconds. Now I understand that laptops have set hardware so they don't need to do a lot of checks to load other things, but isn't UEFI supposed to be modular? Why not allow a configuration to be set and the machine just tries automatically starting with that unless the user enters the configuration screen to change settings. Or am I just wrong and the different between the desktop and laptop is that the desktop has three different SATA/eSATA controllers in it and their ROMs take forever to initialize while the laptop has only a single controller since they mainly have a max of three devices on them.
Windows 7 64 to be an excellent desktop operating system
Maybe. But my problems with Windows are: * It's expensive; * Restrictive license; * Most apps/add-ons/etc. out there want to sneak in and leave their genes in the OS, which inevitably over time leads to pain-in-the-ass bloat; * Difficult to find simple apps that just does the job with no questions asked, and no strings attached (whatever I need I found it in Ubuntu Software Center, while I had to search for quite a while to simply find a clingy PDF shuffler once on Windows, for which I paid a $10 or something to get rid of the ads and get the extra features, and felt rather uneasy that it might have installed more than what I asked... * Bloat, bloat bloat. After years of using my Ubuntu, it still starts in seconds, and doesn't spent the first 15 minutes IO'ing the main hard drive for whatever reason; I still use Windows only on one computer, for games and nothing else.
My guess is also that it is some file indexing service. It seems to kick in on every Windows 7 machine after maybe 10-15 minutes of inactivity.
Sorry to break it to you, shill, but you're full of hot air and you need a tic-tac.
But to balance it out, everything in your post was wrong.
lol
bite my glorious golden ass.
Mostly due to XP's memory manager being completely crap but that's another story.
Have you compared apples to apples? I run XP x64 as my desktop, and it runs everything just fine, with no memory management issues.
Journalling file systems are good, but ones that support true transactions (with atomic commit or clean rollback) are even better. Which is precisely what NTFS does, and what Windows uses to install updates - so no, it's not "fragile" during updates. Unless you're still on XP.
Just wait to install updates, service packs, .NET frameworks (or their future equivalent), etc, etc... A fresh XP installation books in less than 20 seconds on my machine, more like 15. Install updates, drivers, .NET frameworks...
And it's nearly 2 minutes on my, relatively lean, machine (nothing in startup), with decent modern hardware (Intel 9550 Quado Core CPU, 4 GB RAM, 10K RPM hard disk). That's all the way past the login process until I can USE the machine. This thing should fly. Ubuntu loads in 30 seconds, fully ready to be used.
Well, it's not so straightforward with 7 (2008 and Vista too). Yes, desktop performance & experience is great, but the abomination that WinSxS folder is fucks it up rather ruthlessly for VM, laptop and SSD usage. As it is, there is no way to strip it down to bare minimum and run lightweight.
I have VMs running XP on 2 gigs of disk and 256mb of RAM. Let me see you do that with any of the above mentioned. And don't tell me disk is cheap, because those VMs number in tens for me, and probably hundreds for other people.
Now how about my laptop with Windows partition of 25 gigs consisting of 17 gigs of Windows and tiny program files of 1.7 gigs? I'm fucking scared to run Windows update on it.
The WinSxS folder doesn't actually use all the space you think it does.
It's actually just a link to the file elsewhere on your system.
So despite the fact that it takes up 18 GB and is growing rapidly, that space isn't really gone.
This is MS's standard response. But it's complete horseshit. The OS sees the space as being gone, and low diskspace warnings will be triggered, and file write operations will fail. I don't even believe they're symlinks like they claim. It's so fucking stupid. The folder contains a copy of every version of every driver/dll/whatever that was ever WHQLd or registered and in use at one point by your system. Install Program X? Get every version of its components that MS knows about. This is all done to prevent the rare case of "Install Program X, Install Program Y, Y breaks X because it overwrote a shared library with a different version, Uninstall Y, but X is still broken because Y either left it's incompatible shit behind, or took it with it (leaving none behind)." The traditional fix was to not use programs that were retarded, and to reinstall the broken program if some other program fucked it up.
I don't expect any OS to be able to boot the kernel, start the GUI, and lauch all my services with any great speed. I have DB/2 UDB, Oracle 11g, PostgreSQL 8.4, MySQL 5.5, SQL Server, and Glassfish all firing up when the box is booted. It takes 15-20 minutes, and switching to Linux won't improve that.
Competitions over boot time are only relevant to people who just use their system as a client.
I do not fail; I succeed at finding out what does not work.
When measuring how much faster *Windows* is at booting, you shouldn't include the BIOS times as they do not reflect Windows at all. MS showed a laptop booting in 4-5 seconds.. TOTAL. The graphics wasn't able to switch from low-res BIOS to high-res login in the time it took to boot Windows.
A car analogy. I'm going to measure your 1/4 mile times by including the time it took for you to drive from home to the track.
"Along with startup and shutdown times, they used several standard industry benchmarks to compare Windows 8 performance with that of Windows 7 running on the same machine".
...
But not any other OS and not without Microsoft deciding the criteria
Whatever you say, shill.
--
BMO
You appear to have attached some text to a post that has nothing to do with it. Please find the correct post and try again.
In other word - NO I did not suggest anything remotely of the sort. I would also suggest that you learn a little bit of the simple details of how non MSDOS computer systems work (for example NT) before advancing such opinions that are so incorrect that I jokingly suggested they do not belong in this reality. You don't have to reboot for a minor change anymore. Suggesting otherwise is irresponsible unless it's a single user computer and nobody apart from you is going to care if it's on or off.
Using Win 7 i notice that my desktop is NOT immediately usable. But, the problem is clearly that many of the files and folders I access first are on network drives that are painfully slow to load in windows. The files are stored and served off of oracle mainframes.
Subversion of spatial scale luxury decoration ideas.
oh come on, anybody can configure sendmail by hand