Linked by Thom Holwerda on Wed 4th Feb 2009 19:58 UTC, submitted by M-Saunders
Benchmarks So Windows 7 is supposed to be screaming fast, right? Anecdotal accounts report it booting quicker and feeling snappier than Vista, but the proof is in stats. TuxRadar has benchmarked Windows 7 against Vista and Ubuntu Linux, comparing install time, disk space usage, boot speeds and filesystem performance. The graphs also show how the sparkly ext4 filesystem compares against its older brother.
Order by: Score:
IO
by sharms on Wed 4th Feb 2009 20:11 UTC
sharms
Member since:
2009-02-04

They really need to address that IO performance, it shows up not only in the benchmarks but in day to day use.

Reply Score: 1

RE: IO
by flanque on Wed 4th Feb 2009 20:40 UTC in reply to "IO"
flanque Member since:
2005-12-15

It's a beta. There's no way it's more than 5x slower than Vista for no reason.

It's highly likely there's debug code running. I'd expect this to improve as it gets closer to RTM.

Reply Score: 1

RE[2]: IO
by martinus on Thu 5th Feb 2009 06:28 UTC in reply to "RE: IO"
martinus Member since:
2005-07-06

I highly doubt that IO performance will be better in the release. The problem is that NTFS is just extremely bad. I don't know why but for some reason MS has decided to stop improving their filesystem many years ago.

Reply Score: 1

RE[3]: IO
by REM2000 on Thu 5th Feb 2009 09:00 UTC in reply to "RE[2]: IO"
REM2000 Member since:
2006-07-25

NTFS is a very modern / good File system, which is under constant development. The version in Vista has had extra features added such as being able to fix disk problems on the fly without having to reboot and run chkdsk.

I agree with the other poster, i would expect an increase of speed in the RTM to either match or slightly outperform vista.

Reply Score: 2

RE: IO
by lemur2 on Fri 6th Feb 2009 01:00 UTC in reply to "IO"
lemur2 Member since:
2007-02-17

They really need to address that IO performance, it shows up not only in the benchmarks but in day to day use.


Linux reportedly has an "IO wait bug", apparently yet to be resolved.

http://it.slashdot.org/article.pl?sid=09/01/15/049201

At least they now have a reliable and repeatable means now to make it happen, which often is half the effort towards fixing a problem.

If they fix that issue, presumably the Linux performance would become even better again.

Reply Score: 2

Comment by averycfay
by averycfay on Wed 4th Feb 2009 20:19 UTC
averycfay
Member since:
2005-08-29

Number of clicks to install is a pretty lame benchmark. Who cares?

Reply Score: 1

What a completely useless article
by polaris20 on Wed 4th Feb 2009 21:18 UTC
polaris20
Member since:
2005-07-06

1. How long it takes to install: Who cares? Do you do this daily?
2. How many mouseclicks to install: see above
3. How long it takes to boot: Again, not a huge deal unless you reboot once an hour.

I've ran benchmarks too, using Geekbench 64-bit, Ubuntu 8.10, Windows 7, and OS X.

8.10 loses every time significantly in stuff that matters, such as memory bandwidth, processor performance, etc.

But these are still synthetic benchmarks, not always to be trusted. Still, in real world testing using the apps we use (CFD), Ubuntu is still slower (though not as dramatically).

I hope this isn't an indicator of where Linux is going, performance-wise.

Reply Score: 1

satan666 Member since:
2008-04-18

I've ran benchmarks too, using Geekbench 64-bit, Ubuntu 8.10, Windows 7, and OS X.

8.10 loses every time significantly in stuff that matters, such as memory bandwidth, processor performance, etc.

But these are still synthetic benchmarks, not always to be trusted. Still, in real world testing using the apps we use (CFD), Ubuntu is still slower (though not as dramatically).

I hope this isn't an indicator of where Linux is going, performance-wise.


Ubuntu != Linux

Let me tell you a true story. I work for a company that develops a Java application. To ensure portability, one developer works on Windows XP, one on Mac and I work on Mandriva Linux. The Windows machine and my machine are absolutely identical. After two years of daily usage the Windows machine became so slow (partly because Netbeans is more and more demanding) that my boss had to buy a more powerful computer for the Windows guy. I've been working on the same machine for 3 years and half now and I am still very pleased with my computer.

Reply Score: 8

polaris20 Member since:
2005-07-06

"I've ran benchmarks too, using Geekbench 64-bit, Ubuntu 8.10, Windows 7, and OS X.

8.10 loses every time significantly in stuff that matters, such as memory bandwidth, processor performance, etc.

But these are still synthetic benchmarks, not always to be trusted. Still, in real world testing using the apps we use (CFD), Ubuntu is still slower (though not as dramatically).

I hope this isn't an indicator of where Linux is going, performance-wise.


Ubuntu != Linux

Let me tell you a true story. I work for a company that develops a Java application. To ensure portability, one developer works on Windows XP, one on Mac and I work on Mandriva Linux. The Windows machine and my machine are absolutely identical. After two years of daily usage the Windows machine became so slow (partly because Netbeans is more and more demanding) that my boss had to buy a more powerful computer for the Windows guy. I've been working on the same machine for 3 years and half now and I am still very pleased with my computer.
"

I realize Ubuntu does not equal all Linux distributions. However the tests were consistent with SuSe, CentOS, Debian, and Fedora too.

And your example is but one case of bit rot. Many others (including myself) don't have this issue. However buying a new computer because Windows is slowing down is akin to hitting a fly with a sledgehammer. How about reformatting/reimaging instead? It's free, and it's fast. Well reimaging is, anyway.

Reply Score: 2

rexstuff Member since:
2007-04-06

And your example is but one case of bit rot. Many others (including myself) don't have this issue. However buying a new computer because Windows is slowing down is akin to hitting a fly with a sledgehammer. How about reformatting/reimaging instead? It's free, and it's fast. Well reimaging is, anyway.


Yet his anecdote still stings. The point is that over time, the performance of Windows severely degraded to the point where reimaging/upgrading was necessary. This did not happen to the Linux machine.

You can draw whatever conclusions you like from this. Not too much, though, as anecdotal evidence is at best, anecdotal ;)

Reply Score: 1

abdavidson Member since:
2005-07-06

Ubuntu != Linux

Let me tell you a true story. I work for a company that develops a Java application. To ensure portability, one developer works on Windows XP, <snip>


Let me tell you a true story. Windows XP is now a 7 year old OS.

Using it to attempt to put down the 2 year old version or the not yet out version is... daft.

Reply Score: 2

EvilPixieMan Member since:
2009-01-27

memory?

I jumped at this because the memory usage approaches differ dramatically between the two, and as far as Linux goes, using all your RAM is a Good Thing(TM). But on reading, I can't even see a single reference to memory. The word isn't even mentioned!

Reply Score: 1

Havin_it Member since:
2006-03-10

I'll have you know I do reboot once an hour, you insensitive clod!

/Thanks, current Intel X drivers /:|

Reply Score: 3

jayson.knight Member since:
2005-07-06

1. How long it takes to install: Who cares? Do you do this daily?


Thousands of administrators do this every day, dozens of times a day, so yes I'd say this most definitely matters (assuming they don't have a standard image, which many corporations do not).

Reply Score: 2

Surprised some guys dont 'get it'
by ideasman42 on Thu 5th Feb 2009 02:02 UTC
ideasman42
Member since:
2007-07-20

Im surprised how many people dont get that this is not an in depth benchmark, which is fine.

I like that some guy gets some OS's, does some normal operations and times them, without doing any special tricks to optimize them for either system.

This is really valuable since its what most normal people would run with either OS.

The install time and install clicks are interesting because some linux users havnt used win in a while and vice vercer, for all I know windows is 4 mouse clicks or 40... its been that long, so it puts it in perspective. If you dont care about these then ignore them ;)

A number of recent benchmarks also show linux as slow from 3D tests on intel graphics (rather then on nvidia where it might be competitive), while this is fair because so many people have intel graphics.
Its also nice to see benchmarks that run a set of tasks that linux is fast at.

Reply Score: 5

Vista Install
by zlynx on Thu 5th Feb 2009 02:36 UTC in reply to "Surprised some guys dont 'get it'"
zlynx Member since:
2005-07-20

I installed Vista SP1 on a new system I just built last week. It took quite a while!

First the Vista install. That isn't bad, actually. It asks about four questions and does its thing.

But after that, I installed the motherboard drivers, which took longer than Vista! The installer had to reboot the system after each driver set, it looked like.

Then the video drivers. That was pretty quick but needed another reboot.

The the Bluray drivers and software. Actually, this hasn't happened yet and is really upsetting me. Apparently Windows and everything is just fine running a SATA Bluray drive from the motherboard's second SATA controller, but the Bluray drive diagnostics and the playback software don't believe it exists. Bah. I need to somehow convince the Bluray drive and the RAID-0 to share the primary controller while its BIOS is set to RAID instead of AHCI.

Then after that is the adventure of installing each piece of useful software and updating it. This takes *ages*.

In Linux, most of that stuff is in the distro repository where I can click a series of checkboxes and get it all at once.

But really, I like Windows well enough. I just forget the pain of setting it up.

Reply Score: 2

Nobody needs to tell me
by trenchsol on Thu 5th Feb 2009 10:44 UTC
trenchsol
Member since:
2006-12-07

I don't need benchmarks to see that none of recent Microsoft monster operating systems can't even boot on my notebook. In the same time, the machine is perfectly suitable to run latest Linux, Solaris and BSD systems.

Reply Score: 4

RE: Nobody needs to tell me
by raver31 on Thu 5th Feb 2009 11:11 UTC in reply to "Nobody needs to tell me"
raver31 Member since:
2005-07-06

I don't know...

I find Win 7 compares quite favourable against BSD V7. Well, at least for me, it feels extremely sluggish when running ZFS.

Reply Score: 3

RE: Nobody needs to tell me
by jayson.knight on Thu 5th Feb 2009 23:20 UTC in reply to "Nobody needs to tell me"
jayson.knight Member since:
2005-07-06

I don't need benchmarks to see that none of recent Microsoft monster operating systems can't even boot on my notebook. In the same time, the machine is perfectly suitable to run latest Linux, Solaris and BSD systems.


I don't believe this for a hot minute, Win7 will run on netbooks for crying out loud. Unless your laptop > 5 years old, there is absolutely no way this statement is true.

Reply Score: 1

RE[2]: Nobody needs to tell me
by lemur2 on Thu 5th Feb 2009 23:34 UTC in reply to "RE: Nobody needs to tell me"
lemur2 Member since:
2007-02-17

"I don't need benchmarks to see that none of recent Microsoft monster operating systems can't even boot on my notebook. In the same time, the machine is perfectly suitable to run latest Linux, Solaris and BSD systems.
I don't believe this for a hot minute, Win7 will run on netbooks for crying out loud. Unless your laptop > 5 years old, there is absolutely no way this statement is true. "

The reference I believe was to netbooks, not laptops.

Some netbooks have SSDs for the operating systems storage area, and the SSDs on different models vary from as low as 2GB, many have 4 GB, and some have 8GB.

http://products.liliputing.com/products?id=140
http://products.liliputing.com/products/?id=139

The graph in the article shows the "disk spaced used" after installing each OS. Win 7 uses 7.9 GB for 32-bit, and 11 GB for the 64-bit variant. One would use the 32-bit version for netbooks.

This means that Win 7 might just be able to install on the largest of the SSDs, but this would be just the bare OS with no actual desktop applications to speak of.

OTOH, Ubuntu uses about 2.3 GB, and that includes a full suite of desktop applications.

Edited 2009-02-05 23:41 UTC

Reply Score: 3

RE[2]: Nobody needs to tell me
by trenchsol on Fri 6th Feb 2009 14:49 UTC in reply to "RE: Nobody needs to tell me"
trenchsol Member since:
2006-12-07

My notebook is 2004 model. I could replace it, but why ? To run Vista or Win 7 ? I don't think I need that. The resources needed by an OS are very important for performance. Consider a possibility of running one or more virtual machines. If OS eats up all the resources VM's would crawl.

Reply Score: 2

Why disable the write cache?
by PlatformAgnostic on Thu 5th Feb 2009 11:53 UTC
PlatformAgnostic
Member since:
2006-01-02

It's not clear to me why they'd disable caching in the file copy test. The filesystems of both OSes are likely to be designed to perform better when writes can be coalesced via the cache.

Reply Score: 3

Hey Thanks!
by polaris20 on Thu 5th Feb 2009 14:30 UTC
polaris20
Member since:
2005-07-06

Thanks to all you out there for modding me down for sharing my opinion and experiences! It just goes to show that you can't say anything negative at all about Linux on this side, or face the wrath of the zealots.

Nevermind that I actually use all three major OSes on a daily basis for work, WTF do I know?

Reply Score: 2

RE: Hey Thanks!
by SReilly on Fri 6th Feb 2009 10:05 UTC in reply to "Hey Thanks!"
SReilly Member since:
2006-12-28

Word of advice, chill out man!

I know it's frustrating when zealots and ignoramuses mod you down for your opinions but shouting about it doesn't actually help. In fact, you just end up getting people's backs up and making the situation worse. I found this out the hard way.

Reply Score: 2