twitter writes "Recent and controversial benchmarks for Windows 7 leave an important question unanswered: 'Is it faster than GNU/Linux?' Here, at last, is a benchmark that pits Ubuntu, Vista and Windows 7 against each other on the same modern hardware. From install time to GUI efficiency, Ubuntu beats Windows and is often twice as fast. Where Windows 7 is competitive, the difference is something the average user would not notice. The average GNU/Linux user is now getting better absolute performance from their computer as well as better value than the average Windows user."
Seriously, and if it takes more that 1 GB of my 500 GB hard drive then there's something wrong.
Why don't they benchmark some more important timings like how long it takes to shutdown, how long it takes to paste text in an email and how long it takes to run a disk defrag.
Boot-up/shut-down are there. I was focused on the Windows 7/x86 & Ubuntu 9.04/x86 'cuz that's what I run. Windows 7 boots about 13 seconds faster and takes about 4 seconds longer to shut down.
Disk I/O is there too. For moving large files around, the numbers were more-or-less comparable. For moving small files (probably comparable to running a disk defrag), Windows 7 got its ass handed to it. Hopefully Microsoft is aware of this and does something about it before subjecting users to it.
Everything took more than 1 GB of hard-drive space installed, but Windows was 3-4 times as big (7.9 GB rather than 2.3 GB).
That's not even remotely true. Every major flavour of Linux comes with more usable applications installed by default than any version of Windows can.
Word, Excel, Powerpoint and Outlook don't come with Windows, real games don't come with Windows, a C compiler, Python, and Java don't come with Windows, there's only one media player installed with Windows and only one browser as well.
What pray tell, besides Microsoft's video editing tool, do you think comes with Windows that isn't on Linux?
The title, at least, is troll-ish. Ubuntu WIPES windows 7 in benchmarks? Even the article concluded differently:
Obviously we're Linux users ourselves, but our tests have shown that there are some places where Windows 7 really is making some improvement and that's good for competition in the long term. However, Linux isn't sitting still: with ext4 now stable we expect it to be adopted into distros fairly quickly.
We understand perfectly your needs about traffic generation and advertisements.
But please, why publish another stupidity like this... when too recently you had a highly criticized "story" about some random guy that found Ubuntu downloads faster than Vista in his home PC's. Please avoid that kind of sh... (how to name it???), that only ends turning people away for your site in the long term.
Eventually, if you can't stop from posting about so called "comparative benchmarks", please do it in the "idle" section.
Our test machine packed an Intel Core i7 920, which in layman's terms has four cores running at 2.67GHz with hyperthreading and 8MB of L3 cache.
(Emphasis added.)
Not sure what kind of laymen the authors hang out with, but all the laymen I know couldn't tell you the difference between a CPU and a hard drive, or the difference between GHz and GB... much less figure out what "L3 cache" is!
Because model numbers tell you nothing about the specifications without a reference sheet handy. People understand "21 Ghz", but not the model number 12675100.
Clearly Microsoft has been listening to us. Vista takes up a whole 8.2GB, while Windows 7 takes up a mere 7.9GB. I can't wait to get a crack at this smaller, slimmer version of Windows!
Why would anyone care about install time? The only interesting part of the install is how much of your hardware works out of the box, and how much of it can be made to work easily.
Of course installation is the easiest feature to review, but this is 2009 - there is nothing interesting about OS installation anymore.
I've been slowly switching from XP to Ubuntu on my work laptop, but I am still stuck with XP at home. I just play too many PC games to give up XP. I really don't care if it boots slower than Ubuntu, or takes longer to shut down. What matters to me is actually using the PC.
Installation time? *Mouse clicks* to install? Seriously? Those have got to be some of the most useless benchmarks I've ever seen.
Startup and shutdown time are marginally more useful benchmarks, but still not really very important unless you're talking about embedded devices, which the desktop version of Windows 7 (obviously) isn't even designed for.
The file copy benchmarks really didn't find a clear winner either, and that was the only arguably significant benchmark. Or are there really desktop users that spend all day copying files between hard drives and USB drives?
I really didn't care all that much about the outcome. I don't have an emotional investment in Windows or Ubuntu, but this was nothing but a pissing contest from someone who wanted to make some poorly constructed graphs showing that their favorite OS beat another OS (and it didn't even do that! Windows won on a few of the tests!)
On my old PC laptop, Ubuntu gets very unresponsive, even with every combination of ATI drivers I use. Both Windows XP and Windows Vista boot as fast, if not faster, on it than Ubuntu did. In fact, Windows Vista was generally more responsive during normal use. There were plenty of times where Vista could easily handle stuff like Firefox with Flash and some other stuff open, but Ubuntu would slow down to a crawl.
Mod me down if you want, but I've found Windows to be faster and more responsive out of the box, especially against modern Linux distributions.
This will probably get me a troll mod, but I have to say that it doesn't matter how much faster Linux is than Windows in raw speed. All that matters is what the user perceives. And I have to say it doesn't look that great for Ubuntu or Fedora or any modern linux distro right now (but that's improving!). Right now I have Fedora 10 on a brand new dual core AMD 4550e (low-wattage, but still) with 4 GB of ram.
Let's start with the GUI since that is most visible. Without compiz, Fedora's Gnome GUI is quite fast, but to the user feels slow. You can see widgets redraw and reorder themselves. When you size a window you can see the contents adjusting. You can see tearing of the edges of window decorations. When moving the windows around you often get tearing. These artifacts actually make the desktop feel slower even though it really isn't at all.
With compiz-fusion on, things get a little bit better. But still resizing a window is very painful, especially one with a lot of widgets in it. Moving a window around is usually fast enough, though. I believe compiz's rendering engine is synced to screen refresh which helps a lot here (OS X did this for years). Still thought the system often just feels slow. Windows take some time to pop up some times. Sometimes I get a window of garbage (instead of a popup menu) and then the menu appears in it. Sometimes the effects (fade in, fade out), are delayed. Fancier effects like beam-in, beam-out (kind of cool and makes windows users take notice!) work well sometimes and then sometimes stutter or are delayed.
Maybe this is related to the recently-talked about I/O kernel bug, but my Fedora 10 box stutters all the time. My cron script that renders my background Earth picture with the proper clouds and day/night lighting will cause video and audio to halt for a complete second *every* time it is run. This never happened on my older, single processor Athlon with Fedora 8. PulseAudio also seems to cause audio to stutter at the slightest hint of any i/o. In this machine, anyway, with Fedora 10 and compiz-fusion, my Gnome desktop is very disappointing from the perception of performance pov. In raw speed I'm sure it beats Windows Vista or 7. But when you're frustrated with the inability to play back video and audio without skips, and the stuttering and delays in rendering GUI elements, none of that matters.
Now use a Vista computer with decent hardware with the effects turned on. Everything is silky smooth. Window resizes, moving windows (even with translucent blurring). Popups are timely and smooth. The system just feels more responsive than my Fedora Gnome desktop. Things like audio and video have a high priority and never stutter.
How can we improve this? Several ways. First GTK with client windows goes a long ways to solving the resize problem. Rather than having asynchronous messages being passed to each and every widget's window by X11, we only deal with events to the main window. Sub windows are all managed by GTK internally, eliminating the sync problem. This should hit mainstream soon when some corner cases are taken care of. From what I've read, KDE users might already enjoy this as Qt is supposed to already do client windows on X11. Then we need to get pulseaudio fixed somehow. And the kernel bug. Development on compiz after the merger with Beryl seems to be stalled as well. Seems like 80% of the work is done, but the last 20% always struggles to get done, especially in open source software. Finally I hope that issues regarding RGBA and ARGB in GTK in particular get addressed (if they still exist). Then hopefully more apps (KDE already can do this) will use ARGB visuals appropriately.
Probably even dafter . Neither is finished, so you don't know what extra logging or debug they're running (well, with Linux you could but you probably can't be bothered).
You also don't know how tuned they are - the dev teams may not have finished all the performance tweaking in the beta, so yes, you get some numbers but unless you want to run the beta in production they are meaningless when it comes to production.
To be fair to TFA though they acknowledge this and are pretty clear that you can't read much into the beta numbers.
But really, it does matter. If I am going to buy a netbook with a 1.6 GHz Atom CPU, 1 GB of RAM and integrated graphics, I'm going to want something that runs fast. On either platform I will have E-mail, basic games, web browsing, videos, music, etc, and whichever one runs the fastest (and the cheapest) is going to be the one someone usually picks. So when Windows 7 comes out and you can either buy the $300 netbook with Linux that runs faster, or the $350 netbook with Windows 7 that runs slower, the choice for any informed customer is obvious.
However, an informed customer (and by informed consumer, I mean someone reasonably intelligent, knows the strengths and weaknesses of Linux, etc) will almost certainly pick Linux
Since when is the majority of the market an informed consumer? Especially the netbook market?
People are going to pick words that they have heard before. Marketing is a huge part of this and what it boils down to is:
Windows XP = good, Vista = bad, Linux = difficult. (Which is sad, mostly because Vista isn't that bad, and Linux isn't that difficult, but marketing is everything to the mass average consumer)
Agreed that Linux is at least as good or better on a netbook for browsing, checking email, and other simple net tasks. However, I think the (vast) majority of people have a more specialized application they wish to run, and in the (vast) majority of cases, it's a Windows app. That is as likely to be true for an informed consumer as an uninformed consumer.
It's not a case of a buyer getting scared of a 'different' operating system. It's a case of a buyer wanting a netbook that will run every program he might ever want over the next few years, without the operating system getting in the way. It doesn't and wouldn't matter that Linux was significantly faster and more capable.
That's true, until they try to plug in the crap they get at Walmart into the USB ports like those damned Lexmark all in one printers and they find they will never work. I tried selling low cost Linux machines in the shop and finally had to wipe them and put back on Win9X/Win2K. Why? Because I ended up having to stick a sign on the front of them that said "Will NOT run Lexmark printers!" Because trying to get one of those bastards to work in Linux will cause you to blow a blood vessel. And needless to say after 6 months they never sold, whereas even the Win9X sold after a few weeks. After all what good is a PC that you can't print from? And sadly with most consumers(at least around here) Lexmark is king, which means good luck getting it to run in Linux.
What I don't get is why someone can't come up with a WINE or Ndiswrapper for those damned printers. I have taken them apart and there really isn't any chips in them, and surely it can't be more complicated that getting the strip of wire and micro firmware that passes for a wireless "card" these days to work in Linux. All the print/scan/fax work is being done by Windows. From what I can tell calling the printer simply hands off everything to the Windows GDI which does all the work. But until I can place machines on the table and know that they will work with those damned Lexmark printers I just can't sell Linux machines. And I wonder how many of those Netbook returns could be traced back to crap like those damned Lexmarks?
There may be relevant performance differences between the operating systems and versions, but this benchmark mostly does not test for them in the general use cases.
* How long does each operating system take to install?
Typical home user installs neither, real IT shops use disk images and other automated deployment tools.
* How much disk space was used in the standard install?
This is only a significant concern for SSD Netbooks. Typical home users will use ad hoc unmanaged storage and fill any available space with porn/music/photos. Power users who need lots of storage understand multiple hard disks. Real IT shops do managed non-local storage.
* How long does boot up and shutdown take?
The benchmarks shows no significant differences for boot up times. Both tests require more iterations and controls to distinguish between clean shutdowns, and ones in which software, first run, and other updates also take place during shutdown.
* How long does it take to copy files from USB to HD, and from HD to HD?
Methodology is flawed because the installation of each operating system has perturbed the free and occupied space layouts on the hard disk. An unbiased test would be USB to/from other installed hard disk instead of USB to/from the boot disk.
Independent of the operating system in use, data located at the outside of a CAV disk can be read and written faster than data located close to the spindle.
* How fast can it execute the Richards benchmark?
Results do not indicate any significant differences in the set. Also, does the Richards benchmark reasonably simulate any particular home or enterprise task set in general, and if so, does the (undisclosed) version of the Richards benchmark employed in this test also reasonably simulate a particular task set?
But IMHO by using the PC they chose they ruined any info that may have been useful. A Core i7 920? Yeah, because so many folks have those quad core monsters lying around their house. A much more useful test would have been on a combination of two "average" machines, which can still be found in most offices, that is something in the P4 2.6GHz-3.6GHz range with 1-2GB of RAM, along with a Quad and whatever machine is the lowest price at Walmart or Best Buy(last I checked these were single core AMD Sempron or Intel Celeron in the 2GHz range with 1GB of RAM).
This would IMHO give us more of a "real world" benchmark and see how the machines that are the vast majority of the market would handle it. Despite being out for a couple of years now Dual Core machines are not what you find in most folks homes and offices, and quad cores certainly have even less of a penetration in these places. With WinXP getting long in the tooth and the economy bone dry when it comes to credit a lot of SMBs are going to be hanging onto those 2.2-3.6 GHz P4s, and the home consumers are going to be hanging on even tighter.
If MSFT "pulls a Win2K" and lets WinXP die out for lack of attention when Win7 comes out these folks are going to need a migration path that hopefully won't involve buying several $$$$ dollars worth of hardware in the case of the SMBs. With real world benchmarks on real world hardware I could hand this data to customers sitting on the fence and let them decide for themselves which upgrade path would be right for them. But putting it on a Quad Core superbeast doesn't really tell me anything. It certainly doesn't tell me how the 3 OSes would operate in the real world in an office full of 3GHz machines. So if the ones who ran these benchmarks read this: Please rerun these tests on machines one would find in your average office or home. Because with the economy as bad as it is an money tight I have a feeling folks are going to be hanging onto machines both at work and at home for a lot longer than AMD or Intel would like.
We run almost exclusively system intensive high resource usage software. (100% CPU & 5GB+ of RAM, RAID arrays pushed to the limit when rendering.) Under these loads we've seen very little performance differences between OSes. Vista x64, Windows XPx64, Windows 7 x64. Across all 3 Windows apps it's effectively a wash.
Similarly I've seen very very marginal improvements while rendering on Linux.
The tests are kind of interesting in a "I suppose that's interesting" sort of way. But on a modern system how fast most OS features act is the split between milliseconds and who really cares?
The summary is highly misleading "Ubuntu as much as twice as fast!" At extremely short unnoticeable tasks which no human would care to measure except in a benchmark.
I've very very very rarely had the OS be a bottleneck. The last time I remember encountering a system slow down on a reasonably up to date system was when I was trying to run Shake on an OSX PPC G4. An older x86 system on Windows and Linux simply smoked it in every possible way. But that was far more to do with being a PowerPC chip than OSX itself. Oh yeah... and Vista's network transfer speeds when it was first released were embarassing. But those have been straightened out as far as I can tell from my experience.
I generally agree, but frankly, OpenOffice.org Writer is still not a drop-in replacement for MS Word if you're doing anything non-trivial. And I've worked on OpenOffice.org Writer -- it has a couple of advantages and a bunch of glaring omissions even today. I haven't tried Word in Wine, but if it's made such strides, then great.
I'm all for Windows and Office and all the rest...
Having used Windows for years, it does not have NEAR the functionality of any given Linux distro without heavy tweaking.
I mean, come on, I have to install office programs, compilers, editors, (non-DRM) media players, (real) CD/DVD burning programs, terminals, secure communication programs, (real) file transfer programs, etc., and that's just the top categories. Let alone all the crap you have to install, just because you're using Windows, like anti-virus and anti-malware programs.
And then there's the lovely day that a program simply... stops working. Why? Who knows! Time to format and reinstall!
Seriously. I have a Windows partition because I like PC video gaming. (Lord, help me, sometimes even I don't know why. I keep all my drivers up to date, but I still get BSOD's a couple times a month.) But I can't stand to try to use it for real work.
I mean, come on, I have to install office programs, compilers, editors, (non-DRM) media players, (real) CD/DVD burning programs, terminals, secure communication programs, (real) file transfer programs, etc., and that's just the top categories. Let alone all the crap you have to install, just because you're using Windows
I always get a kick out of this argument. Has it occurred to you that when Microsoft bundles those applications they get sued to pieces and end up paying billions in fines to the European antitrust extortionists^H^H^H^H^H^H^H^H^H regulators? If you approve of those verdicts and fines, then you cannot simultaneously criticize Microsoft for forcing you to install all those things. Approving of these fines means simply that we accept (or rather demand) the inconveniences they inflict.
I wouldn't mind if they bundled everything they make, and everything they don't make as well! Just put a price tag on it and let the market sort it out. What I can't stand is that they essentially GIVE the software away through bundling deals with the OEM's, but tell them that they can't install any other software that competes with their products, essentially causing the OEM's to eat the difference, and pass the savings on to their customers. THAT'S anti-capitalistic. Unfortunately, government's only answer is to do what they've done, which includes forcing Microsoft to GIVE MORE of their software away to schools, further entrenching their monopoly. Gah!
Microsoft's making all their money from corporate sales, who are basically beholden because of the Office monopoly. All I want is for Microsoft to sell the same piece of software for the same price to everyone. Let them have 42 editions, for all I care, but just box it and price it and let the market sort it out.
How many individuals do you know have paid full retail price for either Windows or Office? If I could go buy either one for what they cost the OEM, the tier-1 Select customer, or the college student -- or if THEY had to pay what -I- pay, then I would consider that competition. I'd even consider it fair to meet in the middle. If Vista Ultimate cost what a new copy of OS X cost, that would seem to be about right. Have you seen what it actually retails for? Scary.
I have personally moved my grandmother and uncle, neither which know ANYTHING about computers. The only problem I have seen is opening horrendously formated word documents and running DX games. Please do not compare a bestbuy installed windows with a downloaded iso linux, they are not nearly the same. When bestbuy installs windows, they find the drivers, install antivirus, add tutorials, etc. When I set up an Ubuntu system, I do the same and they have NO problems!
Linux is just as easy, if not easier to use than windows. Just look at opening programs. In windows you go "start->all programs->adobe->photoshop->start photoshop". In linux you go "Applications->Graphics->Gimp Image Editor". Not to mention installing applications. In windows you have to google-hunt a program, pray it's clean, download, scan, install. In linux you open the package manager, select it and click "apply".
Please stop spreading this FUD that windows is easier simply because some joker being paid $8/hour set it up for you!
Linux has always been rather slow to boot, but as we understand it reducing boot time is one of the goals of the Ubuntu 9.04 release.
What kind of comment is that? Excusing a "slow boot time" with "Linux has always been rather slow to boot." Of course, then we get other benchmarks where it says that Ubuntu betas Windows in booting. IMO, this just goes to show that benchmarks on something that is so hardware dependent can be really silly. That and the user's bias is coming out in defending Linux by saying it's always been slow to boot. If Windows was the one that was so slow, it probably would have been "Windows has always been infamously slow to boot, and Windows 7 is no change." Or whatever.
Also... measuring mouse clicks on an install process? What?
And... comparing the amount if gigabytes and saying that less space used after a fresh install is necessarily better? Becuase, as we all know, a 6 GB installation of an OS is absolutely horrendously huge, given the exorbitant cost of disk storage these days. Man, 1/166th of my 1TB drive gone because Windows! [/sarcasm]
the user's bias is coming out in defending Linux by saying it's always been slow to boot
That's not how I read it. The author didn't seem to be defending Linux with that statement. It was more of a "as we would have expected" statement. He was acknowledging that Linux lost on that metric.
measuring mouse clicks on an install process? What?
The authors seem to acknowledge that this metric was just for fun. The caption for that data says "A bit of a flippant one" and in the intro they say "We also, just for the heck of it, kept track of how many mouse clicks it took to install each OS."
comparing the amount if gigabytes and saying that less space used after a fresh install is necessarily better?
Yes. All other things being equal, a smaller install size is better (more space for other things). Whether or not this particular metric matters to you depends, of course. On a typical desktop machine it might not matter. On some other machines it might. The install size also affects other things people might care about (e.g. how long it takes to do a drive image or backup; how long it takes to scan or seek on the drive;...).
Its ok, the article's first half is a bunch of benchmarks that are utterly meaningless on Windows anyway. Who cares if Window's takes twice as long to install as Linux? I mean seriously, I'm waiting. Are operating installs a frequent event? I can count on my hands and feet the number of times I've performed them. Its all well and good that Ubuntu can install itself faster, but it doesn't matter, because it is by definition an infrequent workload. This is theoretically true for Ubuntu to. After all, wasn't the infinite in place upgradability something that has long been touted as a strength of Debian and co. Thats even more important with Ubuntu, because I sure as hell don't want to reinstall and OS every 6 months.
Same goes for startup and shutdown. Windows Vista was explicitly designed with the idea that in general, the OS is going to be suspended/hibernated, not rebooted. I'd be much more interested in seeing benchmarks of a comparison between the speed with which Windows and Ubuntu are able to hibernate/unhibernate. I've always been curious about this, as subjectively, an older Ubuntu installation hibernation seemed faster than in Windows. Alas, I guess in order to give us that benchmark, the reviewers would have to actually find hardware Linux could suspend on. How does one plot a hard lock on resume anyway, time for the system to reboot and come back up?
The other thing they failed to mention on the I/O benchmarking side is whether or not the drives were set to write cache mode or not in Windows. AFAIK the default for removable media to disable write caching in Windows, but to enable in Linux.
Oh, and why the !@#$ are they benchmarking compute intensive tasks in Python? Is it to exacerbate differences, because the chosen runtime is so absurdly slow? But, in reality, there is no reason for compute intensive tasks to vary on the same hardware. This test is highly dependent on the system services running and the python version. I would consider this more of a benchmark of python instead of Windows/Ubuntu.
+Troll (Score:5, Insightful)
Can I mod this story as troll?
I'm a linux user but this story is anything but serious benchmarking.
Re:+Troll (Score:5, Funny)
Parent
Re:+Troll (Score:5, Funny)
Parent
Re:+Troll (Score:5, Informative)
Seriously, and if it takes more that 1 GB of my 500 GB hard drive then there's something wrong.
Why don't they benchmark some more important timings like how long it takes to shutdown, how long it takes to paste text in an email and how long it takes to run a disk defrag.
Boot-up/shut-down are there. I was focused on the Windows 7/x86 & Ubuntu 9.04/x86 'cuz that's what I run. Windows 7 boots about 13 seconds faster and takes about 4 seconds longer to shut down.
Disk I/O is there too. For moving large files around, the numbers were more-or-less comparable. For moving small files (probably comparable to running a disk defrag), Windows 7 got its ass handed to it. Hopefully Microsoft is aware of this and does something about it before subjecting users to it.
Everything took more than 1 GB of hard-drive space installed, but Windows was 3-4 times as big (7.9 GB rather than 2.3 GB).
Parent
Re:+Troll (Score:5, Insightful)
That's not even remotely true. Every major flavour of Linux comes with more usable applications installed by default than any version of Windows can.
Word, Excel, Powerpoint and Outlook don't come with Windows, real games don't come with Windows, a C compiler, Python, and Java don't come with Windows, there's only one media player installed with Windows and only one browser as well.
What pray tell, besides Microsoft's video editing tool, do you think comes with Windows that isn't on Linux?
Parent
Re:+Troll (Score:5, Funny)
I'm a linux user but this story is anything but serious benchmarking.
Yeah, they left out almost all distros.
Parent
Re:+Troll (Score:5, Funny)
I read the headline and thought installing Ubuntu would wipe a Windows 7 partition.
Parent
Re:+Troll (Score:5, Funny)
Parent
Re:+Troll (Score:4, Informative)
What's wrong? I mean the summary leads you directly to the conclusion you need to be coming to here:
"The average GNU/Linux user is now getting better absolute performance from their computer as well as better value than the average Windows user."
Seriously, that's good enough for me. Don't even need to read the article now...
Parent
Mouse Clicking (Score:5, Funny)
Parent
Re:+Troll (Score:5, Insightful)
The title, at least, is troll-ish. Ubuntu WIPES windows 7 in benchmarks? Even the article concluded differently:
Obviously we're Linux users ourselves, but our tests have shown that there are some places where Windows 7 really is making some improvement and that's good for competition in the long term. However, Linux isn't sitting still: with ext4 now stable we expect it to be adopted into distros fairly quickly.
Parent
Dear /. editors (Score:5, Insightful)
Dear Slashdot editors,
We understand perfectly your needs about traffic generation and advertisements.
But please, why publish another stupidity like this... when too recently you had a highly criticized "story" about some random guy that found Ubuntu downloads faster than Vista in his home PC's. Please avoid that kind of sh... (how to name it???), that only ends turning people away for your site in the long term.
Eventually, if you can't stop from posting about so called "comparative benchmarks", please do it in the "idle" section.
regards,
Parent
Re:+Troll (Score:5, Informative)
I'm guessing the real root of both of your problems is old graphics drivers, unless you really seariously prefer IE over Firefox?
Parent
Re:+Troll (Score:5, Funny)
Parent
I can best them both. (Score:4, Funny)
Re:I can best them both. (Score:5, Funny)
Wow, imagine a beowolf cluster of Essequemodeias!
Parent
Re:I can best them both. (Score:5, Funny)
Parent
Wrong (Score:5, Funny)
My unpatched Windows system can get rooted AT LEAST ten times faster than Ubuntu. Take that, Open Source!
Re:Wrong (Score:5, Funny)
... and this is not a bug, it is a feature! ;-)
Parent
Layman? (Score:5, Insightful)
From TFA:
Our test machine packed an Intel Core i7 920, which in layman's terms has four cores running at 2.67GHz with hyperthreading and 8MB of L3 cache.
(Emphasis added.)
Not sure what kind of laymen the authors hang out with, but all the laymen I know couldn't tell you the difference between a CPU and a hard drive, or the difference between GHz and GB ... much less figure out what "L3 cache" is!
Re:Layman? (Score:5, Insightful)
Because model numbers tell you nothing about the specifications without a reference sheet handy. People understand "21 Ghz", but not the model number 12675100.
Parent
Is that with Virus Software installed? (Score:5, Insightful)
With virus software installed on Windows 7 ubuntu would kill it even more.
Re:Is that with Virus Software installed? (Score:4, Interesting)
Good point. Here's a funny link on the subject http://www.codinghorror.com/blog/archives/000803.html [codinghorror.com]
Parent
Re:Is that with Virus Software installed? (Score:5, Funny)
Ah, a fellow gentooer. Are you done compiling yet?
Parent
Bravo! (Score:5, Funny)
Install time? (Score:5, Insightful)
Why would anyone care about install time? The only interesting part of the install is how much of your hardware works out of the box, and how much of it can be made to work easily.
Of course installation is the easiest feature to review, but this is 2009 - there is nothing interesting about OS installation anymore.
What sold me.. (Score:5, Insightful)
I've been slowly switching from XP to Ubuntu on my work laptop, but I am still stuck with XP at home. I just play too many PC games to give up XP. I really don't care if it boots slower than Ubuntu, or takes longer to shut down. What matters to me is actually using the PC.
Seriously? (Score:4, Insightful)
Installation time? *Mouse clicks* to install? Seriously? Those have got to be some of the most useless benchmarks I've ever seen.
Startup and shutdown time are marginally more useful benchmarks, but still not really very important unless you're talking about embedded devices, which the desktop version of Windows 7 (obviously) isn't even designed for.
The file copy benchmarks really didn't find a clear winner either, and that was the only arguably significant benchmark. Or are there really desktop users that spend all day copying files between hard drives and USB drives?
I really didn't care all that much about the outcome. I don't have an emotional investment in Windows or Ubuntu, but this was nothing but a pissing contest from someone who wanted to make some poorly constructed graphs showing that their favorite OS beat another OS (and it didn't even do that! Windows won on a few of the tests!)
--Jeremy
But can they measure responsiveness? (Score:5, Insightful)
On my old PC laptop, Ubuntu gets very unresponsive, even with every combination of ATI drivers I use. Both Windows XP and Windows Vista boot as fast, if not faster, on it than Ubuntu did. In fact, Windows Vista was generally more responsive during normal use. There were plenty of times where Vista could easily handle stuff like Firefox with Flash and some other stuff open, but Ubuntu would slow down to a crawl.
Mod me down if you want, but I've found Windows to be faster and more responsive out of the box, especially against modern Linux distributions.
The perception of speed is all that counts (Score:5, Interesting)
This will probably get me a troll mod, but I have to say that it doesn't matter how much faster Linux is than Windows in raw speed. All that matters is what the user perceives. And I have to say it doesn't look that great for Ubuntu or Fedora or any modern linux distro right now (but that's improving!). Right now I have Fedora 10 on a brand new dual core AMD 4550e (low-wattage, but still) with 4 GB of ram.
Let's start with the GUI since that is most visible. Without compiz, Fedora's Gnome GUI is quite fast, but to the user feels slow. You can see widgets redraw and reorder themselves. When you size a window you can see the contents adjusting. You can see tearing of the edges of window decorations. When moving the windows around you often get tearing. These artifacts actually make the desktop feel slower even though it really isn't at all.
With compiz-fusion on, things get a little bit better. But still resizing a window is very painful, especially one with a lot of widgets in it. Moving a window around is usually fast enough, though. I believe compiz's rendering engine is synced to screen refresh which helps a lot here (OS X did this for years). Still thought the system often just feels slow. Windows take some time to pop up some times. Sometimes I get a window of garbage (instead of a popup menu) and then the menu appears in it. Sometimes the effects (fade in, fade out), are delayed. Fancier effects like beam-in, beam-out (kind of cool and makes windows users take notice!) work well sometimes and then sometimes stutter or are delayed.
Maybe this is related to the recently-talked about I/O kernel bug, but my Fedora 10 box stutters all the time. My cron script that renders my background Earth picture with the proper clouds and day/night lighting will cause video and audio to halt for a complete second *every* time it is run. This never happened on my older, single processor Athlon with Fedora 8. PulseAudio also seems to cause audio to stutter at the slightest hint of any i/o. In this machine, anyway, with Fedora 10 and compiz-fusion, my Gnome desktop is very disappointing from the perception of performance pov. In raw speed I'm sure it beats Windows Vista or 7. But when you're frustrated with the inability to play back video and audio without skips, and the stuttering and delays in rendering GUI elements, none of that matters.
Now use a Vista computer with decent hardware with the effects turned on. Everything is silky smooth. Window resizes, moving windows (even with translucent blurring). Popups are timely and smooth. The system just feels more responsive than my Fedora Gnome desktop. Things like audio and video have a high priority and never stutter.
How can we improve this? Several ways. First GTK with client windows goes a long ways to solving the resize problem. Rather than having asynchronous messages being passed to each and every widget's window by X11, we only deal with events to the main window. Sub windows are all managed by GTK internally, eliminating the sync problem. This should hit mainstream soon when some corner cases are taken care of. From what I've read, KDE users might already enjoy this as Qt is supposed to already do client windows on X11. Then we need to get pulseaudio fixed somehow. And the kernel bug. Development on compiz after the merger with Beryl seems to be stalled as well. Seems like 80% of the work is done, but the last 20% always struggles to get done, especially in open source software. Finally I hope that issues regarding RGBA and ARGB in GTK in particular get addressed (if they still exist). Then hopefully more apps (KDE already can do this) will use ARGB visuals appropriately.
Re:And... (Score:5, Informative)
"Ubuntu 9.04 we used the daily build from January 22nd."
Parent
Re:And... (Score:5, Insightful)
Probably even dafter . Neither is finished, so you don't know what extra logging or debug they're running (well, with Linux you could but you probably can't be bothered).
You also don't know how tuned they are - the dev teams may not have finished all the performance tweaking in the beta, so yes, you get some numbers but unless you want to run the beta in production they are meaningless when it comes to production.
To be fair to TFA though they acknowledge this and are pretty clear that you can't read much into the beta numbers.
Parent
Re:And... (Score:5, Funny)
What about a story that matters?
Do you make a point of posting in every story that doesn't matter to you? Or was it "cue the douchebag" that you couldn't resist responding to?
You lead a very fatiguing existence, don't you?
Parent
Re:And... (Score:5, Insightful)
Parent
Re:And... (Score:5, Insightful)
However, an informed customer (and by informed consumer, I mean someone reasonably intelligent, knows the strengths and weaknesses of Linux, etc) will almost certainly pick Linux
Since when is the majority of the market an informed consumer? Especially the netbook market?
People are going to pick words that they have heard before. Marketing is a huge part of this and what it boils down to is:
Windows XP = good, Vista = bad, Linux = difficult. (Which is sad, mostly because Vista isn't that bad, and Linux isn't that difficult, but marketing is everything to the mass average consumer)
Parent
Re:And... (Score:5, Insightful)
It's not a case of a buyer getting scared of a 'different' operating system. It's a case of a buyer wanting a netbook that will run every program he might ever want over the next few years, without the operating system getting in the way. It doesn't and wouldn't matter that Linux was significantly faster and more capable.
Parent
Re:And... (Score:5, Interesting)
That's true, until they try to plug in the crap they get at Walmart into the USB ports like those damned Lexmark all in one printers and they find they will never work. I tried selling low cost Linux machines in the shop and finally had to wipe them and put back on Win9X/Win2K. Why? Because I ended up having to stick a sign on the front of them that said "Will NOT run Lexmark printers!" Because trying to get one of those bastards to work in Linux will cause you to blow a blood vessel. And needless to say after 6 months they never sold, whereas even the Win9X sold after a few weeks. After all what good is a PC that you can't print from? And sadly with most consumers(at least around here) Lexmark is king, which means good luck getting it to run in Linux.
What I don't get is why someone can't come up with a WINE or Ndiswrapper for those damned printers. I have taken them apart and there really isn't any chips in them, and surely it can't be more complicated that getting the strip of wire and micro firmware that passes for a wireless "card" these days to work in Linux. All the print/scan/fax work is being done by Windows. From what I can tell calling the printer simply hands off everything to the Windows GDI which does all the work. But until I can place machines on the table and know that they will work with those damned Lexmark printers I just can't sell Linux machines. And I wonder how many of those Netbook returns could be traced back to crap like those damned Lexmarks?
Parent
Re:And... (Score:5, Insightful)
There may be relevant performance differences between the operating systems and versions, but this benchmark mostly does not test for them in the general use cases.
* How long does each operating system take to install?
Typical home user installs neither, real IT shops use disk images and other automated deployment tools.
* How much disk space was used in the standard install?
This is only a significant concern for SSD Netbooks. Typical home users will use ad hoc unmanaged storage and fill any available space with porn/music/photos. Power users who need lots of storage understand multiple hard disks. Real IT shops do managed non-local storage.
* How long does boot up and shutdown take?
The benchmarks shows no significant differences for boot up times. Both tests require more iterations and controls to distinguish between clean shutdowns, and ones in which software, first run, and other updates also take place during shutdown.
* How long does it take to copy files from USB to HD, and from HD to HD?
Methodology is flawed because the installation of each operating system has perturbed the free and occupied space layouts on the hard disk. An unbiased test would be USB to/from other installed hard disk instead of USB to/from the boot disk.
Independent of the operating system in use, data located at the outside of a CAV disk can be read and written faster than data located close to the spindle.
* How fast can it execute the Richards benchmark?
Results do not indicate any significant differences in the set. Also, does the Richards benchmark reasonably simulate any particular home or enterprise task set in general, and if so, does the (undisclosed) version of the Richards benchmark employed in this test also reasonably simulate a particular task set?
Parent
Re:And... (Score:5, Interesting)
But IMHO by using the PC they chose they ruined any info that may have been useful. A Core i7 920? Yeah, because so many folks have those quad core monsters lying around their house. A much more useful test would have been on a combination of two "average" machines, which can still be found in most offices, that is something in the P4 2.6GHz-3.6GHz range with 1-2GB of RAM, along with a Quad and whatever machine is the lowest price at Walmart or Best Buy(last I checked these were single core AMD Sempron or Intel Celeron in the 2GHz range with 1GB of RAM).
This would IMHO give us more of a "real world" benchmark and see how the machines that are the vast majority of the market would handle it. Despite being out for a couple of years now Dual Core machines are not what you find in most folks homes and offices, and quad cores certainly have even less of a penetration in these places. With WinXP getting long in the tooth and the economy bone dry when it comes to credit a lot of SMBs are going to be hanging onto those 2.2-3.6 GHz P4s, and the home consumers are going to be hanging on even tighter.
If MSFT "pulls a Win2K" and lets WinXP die out for lack of attention when Win7 comes out these folks are going to need a migration path that hopefully won't involve buying several $$$$ dollars worth of hardware in the case of the SMBs. With real world benchmarks on real world hardware I could hand this data to customers sitting on the fence and let them decide for themselves which upgrade path would be right for them. But putting it on a Quad Core superbeast doesn't really tell me anything. It certainly doesn't tell me how the 3 OSes would operate in the real world in an office full of 3GHz machines. So if the ones who ran these benchmarks read this: Please rerun these tests on machines one would find in your average office or home. Because with the economy as bad as it is an money tight I have a feeling folks are going to be hanging onto machines both at work and at home for a lot longer than AMD or Intel would like.
Parent
Re:And... (Score:5, Insightful)
We run almost exclusively system intensive high resource usage software. (100% CPU & 5GB+ of RAM, RAID arrays pushed to the limit when rendering.) Under these loads we've seen very little performance differences between OSes. Vista x64, Windows XPx64, Windows 7 x64. Across all 3 Windows apps it's effectively a wash.
Similarly I've seen very very marginal improvements while rendering on Linux.
The tests are kind of interesting in a "I suppose that's interesting" sort of way. But on a modern system how fast most OS features act is the split between milliseconds and who really cares?
The summary is highly misleading "Ubuntu as much as twice as fast!" At extremely short unnoticeable tasks which no human would care to measure except in a benchmark.
I've very very very rarely had the OS be a bottleneck. The last time I remember encountering a system slow down on a reasonably up to date system was when I was trying to run Shake on an OSX PPC G4. An older x86 system on Windows and Linux simply smoked it in every possible way. But that was far more to do with being a PowerPC chip than OSX itself. Oh yeah... and Vista's network transfer speeds when it was first released were embarassing. But those have been straightened out as far as I can tell from my experience.
Parent
Re:And... (Score:5, Funny)
Naw - there's more than one. So you have to line them up single file and deal with them one at a time.
Parent
Re:And... (Score:5, Interesting)
I generally agree, but frankly, OpenOffice.org Writer is still not a drop-in replacement for MS Word if you're doing anything non-trivial. And I've worked on OpenOffice.org Writer -- it has a couple of advantages and a bunch of glaring omissions even today. I haven't tried Word in Wine, but if it's made such strides, then great.
Parent
Re:And... (Score:5, Insightful)
I'm all for Windows and Office and all the rest...
Having used Windows for years, it does not have NEAR the functionality of any given Linux distro without heavy tweaking.
I mean, come on, I have to install office programs, compilers, editors, (non-DRM) media players, (real) CD/DVD burning programs, terminals, secure communication programs, (real) file transfer programs, etc., and that's just the top categories. Let alone all the crap you have to install, just because you're using Windows, like anti-virus and anti-malware programs.
And then there's the lovely day that a program simply... stops working. Why? Who knows! Time to format and reinstall!
Seriously. I have a Windows partition because I like PC video gaming. (Lord, help me, sometimes even I don't know why. I keep all my drivers up to date, but I still get BSOD's a couple times a month.) But I can't stand to try to use it for real work.
Parent
Re:And... (Score:5, Insightful)
I always get a kick out of this argument. Has it occurred to you that when Microsoft bundles those applications they get sued to pieces and end up paying billions in fines to the European antitrust extortionists^H^H^H^H^H^H^H^H^H regulators? If you approve of those verdicts and fines, then you cannot simultaneously criticize Microsoft for forcing you to install all those things. Approving of these fines means simply that we accept (or rather demand) the inconveniences they inflict.
Parent
Re:And... (Score:5, Informative)
I wouldn't mind if they bundled everything they make, and everything they don't make as well! Just put a price tag on it and let the market sort it out. What I can't stand is that they essentially GIVE the software away through bundling deals with the OEM's, but tell them that they can't install any other software that competes with their products, essentially causing the OEM's to eat the difference, and pass the savings on to their customers. THAT'S anti-capitalistic. Unfortunately, government's only answer is to do what they've done, which includes forcing Microsoft to GIVE MORE of their software away to schools, further entrenching their monopoly. Gah!
Microsoft's making all their money from corporate sales, who are basically beholden because of the Office monopoly. All I want is for Microsoft to sell the same piece of software for the same price to everyone. Let them have 42 editions, for all I care, but just box it and price it and let the market sort it out.
How many individuals do you know have paid full retail price for either Windows or Office? If I could go buy either one for what they cost the OEM, the tier-1 Select customer, or the college student -- or if THEY had to pay what -I- pay, then I would consider that competition. I'd even consider it fair to meet in the middle. If Vista Ultimate cost what a new copy of OS X cost, that would seem to be about right. Have you seen what it actually retails for? Scary.
Parent
Re:And... (Score:5, Insightful)
I have personally moved my grandmother and uncle, neither which know ANYTHING about computers. The only problem I have seen is opening horrendously formated word documents and running DX games. Please do not compare a bestbuy installed windows with a downloaded iso linux, they are not nearly the same. When bestbuy installs windows, they find the drivers, install antivirus, add tutorials, etc. When I set up an Ubuntu system, I do the same and they have NO problems!
Linux is just as easy, if not easier to use than windows. Just look at opening programs. In windows you go "start->all programs->adobe->photoshop->start photoshop". In linux you go "Applications->Graphics->Gimp Image Editor". Not to mention installing applications. In windows you have to google-hunt a program, pray it's clean, download, scan, install. In linux you open the package manager, select it and click "apply".
Please stop spreading this FUD that windows is easier simply because some joker being paid $8/hour set it up for you!
Parent
Re:Heh. (Score:5, Insightful)
Another note.
Linux has always been rather slow to boot, but as we understand it reducing boot time is one of the goals of the Ubuntu 9.04 release.
What kind of comment is that? Excusing a "slow boot time" with "Linux has always been rather slow to boot." Of course, then we get other benchmarks where it says that Ubuntu betas Windows in booting. IMO, this just goes to show that benchmarks on something that is so hardware dependent can be really silly. That and the user's bias is coming out in defending Linux by saying it's always been slow to boot. If Windows was the one that was so slow, it probably would have been "Windows has always been infamously slow to boot, and Windows 7 is no change." Or whatever.
Also... measuring mouse clicks on an install process? What?
And ... comparing the amount if gigabytes and saying that less space used after a fresh install is necessarily better? Becuase, as we all know, a 6 GB installation of an OS is absolutely horrendously huge, given the exorbitant cost of disk storage these days. Man, 1/166th of my 1TB drive gone because Windows! [/sarcasm]
Parent
Re:Heh. (Score:5, Insightful)
the user's bias is coming out in defending Linux by saying it's always been slow to boot
That's not how I read it. The author didn't seem to be defending Linux with that statement. It was more of a "as we would have expected" statement. He was acknowledging that Linux lost on that metric.
measuring mouse clicks on an install process? What?
The authors seem to acknowledge that this metric was just for fun. The caption for that data says "A bit of a flippant one" and in the intro they say "We also, just for the heck of it, kept track of how many mouse clicks it took to install each OS."
comparing the amount if gigabytes and saying that less space used after a fresh install is necessarily better?
Yes. All other things being equal, a smaller install size is better (more space for other things). Whether or not this particular metric matters to you depends, of course. On a typical desktop machine it might not matter. On some other machines it might. The install size also affects other things people might care about (e.g. how long it takes to do a drive image or backup; how long it takes to scan or seek on the drive; ...).
Parent
Re:Time for me to upgrade (Score:5, Funny)
There, fixed that for you.
Parent
Re:GUI Efficiency? (Score:5, Informative)
Its all well and good that Ubuntu can install itself faster, but it doesn't matter, because it is by definition an infrequent workload. This is theoretically true for Ubuntu to. After all, wasn't the infinite in place upgradability something that has long been touted as a strength of Debian and co. Thats even more important with Ubuntu, because I sure as hell don't want to reinstall and OS every 6 months.
Same goes for startup and shutdown. Windows Vista was explicitly designed with the idea that in general, the OS is going to be suspended/hibernated, not rebooted. I'd be much more interested in seeing benchmarks of a comparison between the speed with which Windows and Ubuntu are able to hibernate/unhibernate. I've always been curious about this, as subjectively, an older Ubuntu installation hibernation seemed faster than in Windows. Alas, I guess in order to give us that benchmark, the reviewers would have to actually find hardware Linux could suspend on. How does one plot a hard lock on resume anyway, time for the system to reboot and come back up?
The other thing they failed to mention on the I/O benchmarking side is whether or not the drives were set to write cache mode or not in Windows. AFAIK the default for removable media to disable write caching in Windows, but to enable in Linux.
Oh, and why the !@#$ are they benchmarking compute intensive tasks in Python? Is it to exacerbate differences, because the chosen runtime is so absurdly slow? But, in reality, there is no reason for compute intensive tasks to vary on the same hardware. This test is highly dependent on the system services running and the python version. I would consider this more of a benchmark of python instead of Windows/Ubuntu.
Parent