For users of Fedora Core releases <fedora-list@redhat.com> wrote:
On Sat, 05 Mar 2005 11:41:32 +0530, Parameshwara Bhat pbhat@ongc.net wrote:
On my home computer and office computer, I have logged the time taken upto logging in, it turns out that Ms windows(XP) is faster than Linux. On both computers, dma is turned on.On my home computer, I have SUSE, Fedora and Knoppix (debian) installed and on office computer, only Fedora. I have measured with all the distros and while knoppix is faster of the distros, still it doesn't measure upto Windows in speed. Both fedora and SUSE take about one and a times longer time, evrything else remaining same.
Any comments ? Or am I missing something ?
Lots of comments. First of all, as others have commented, there is a difference between boot time and speed once its booted up, as well as shutdown time (if you have like a laptop for instance.....).
Unix based Operating systems are designed to boot once and run for a long time. Windows systems are meant to be started and shutdown with each use. Many people don't shutdown Windows every day, but its really designed to do that.
Boot speed is a "visible" but meaningless speed since it only impacts startup. However Gates & Co want your experience to be a good one and that means faster boot speeds. As such, some of the file system checks and other maintenance operations have been pushed to shutdown time, the theory being that once you hit shutdown and walk way, you don't care how long it takes to shut down.
So measuring "boot speed" is a rather meaningless measure of performance.
The next major difference between Windows and Unix OS's is memory management. This is where speed becomes very noticeable. Both systems us a concept called "paging" to move unused programs and data out of memory to make room for programs that are currently running. Windows pages constantly. The system provides a smooth constant level of performance because its always swapping. With Unix boxes, they don't start paging until they run out of physical memory. The effect is that the machine will run like a race horse as long as your don't run out of memory. When you do run out of memory, you start "paging" and suddenly the system slows way down.
With this fundamental difference the more memory you have the faster the Unix box will be, where as with Windows, there is a limit to where memory helps, but a faster hard drive will boost the overall speed. That said, a Linux box (or any Unix OS, including the Mac OS X) is going to be considerably faster at number crunching than the same speed Windows Box since more of the CPU is being used for processing and less for paging.
Now we get into the next reason its hard to measure things, graphics. Xwindows is slow, period. It is no where near as optimized for graphic output as Windows is. Microsoft has worked very hard to get its graphics rendering to be as efficient as possible. This is due in part to Windows being able to directly address the hardware where as under Unix the graphics drivers have to be a lot more portable and layered (Gnome needs GTK+ which needs Xwindows which needs video drivers and then the card. With windows its, Windows->Driver->Card. This is one reason graphic artists like Macs. You get the OS efficiencies of Unix and a solid window system that works with fewer layers. The reason this is the way it is has to do with the nature of the two beasts. Microsoft being a single company can build a very dedicated platform where as the Unix family has to address openess and standards. Lets face it, there are a zillion different Linux, (Free,Net,Open)BSD's, in addition to HPUX, Solaris, AT&T Sys V, etc. that all have to be X compatiable and then you have choices of GNOME< KDE, and a bunch of other Interfaces on top of that. Windows will always win this one.
Finally, the last area of speed that needs addressed is what all is your system's that you are comparing loading? With Windows, you are very unlikely to have a sendmail daemon running, or a telnet/sshd daemon, or a web server or SQL server running. Depending on your install options for Linux, those may very well be running and unless you are looking for them, you may not notice them.
So when Windows starts up, its loads its kernel and graphics system and device drivers. When Linux starts up, not only does it load those, but there is a bunch of other stuff firing off, like time sync (which your system basically stops until it connects to the time server and gets the time and sets your clock). So its really unfair, as I said above to even consider power-on to prompt as a measure of anything meaningful.
So in conclusion, given two identical boxes with sufficient memory to avoid swapping, the Linux box should smoke the Windows box in number crunching, spitting out web pages, etc. Windows is going to win when Graphics are involved or if memory is on the thin side.
Rob
Rob Miracle Photographic Miracles Cary, NC http://www.photo-miracles.com
I have to read the rest but I would like to add something to support one reason Microsoft wants to tout their quick boot times. Although Microsoft tried and tried to fix the blue screen of death they couldn't eliminate the problems, so they switched gears to have a quicker reboot...
On Sat, Mar 05, 2005 18:49:12 PM -0000, James Marcinek (jmarc1@jemconsult.biz) wrote:
[some pages of text cut]
I have to read the rest
While you do it, please remember to not post again pages and pages of already downloaded (=paid, sometimes) stuff. See recent discussions on netiquette.
Thanks, Marco
While all the reasons for linux not 'feeling' as fast as windows booting are very valid and interesting for me, I am not going to explain all that to my girlfriend. She will just say OK and boot into windows if all that she wants to do is check her hotmail. And, yes, when she powers the machine off, she doesn't care how long it takes- shes not waiting for it. Why can't linux shift some of the things it does during boot to doing them during shutdown?
This is another issue of techies wanting things to be 'good' and non-techies wanting things to be fun, or easy, or 'feel' fast. No non-technie is sitting there with a stopwatch. If linux will remain only in the hands of the techies, then there is no need for this discussion. Fact that it is being discussed that it must be addressed. And if we want our girlfriends to boot into linux, then we had better make it fun, easy, and feel fast.
Dotan Cohen http://English-Lyrics.com http://Song-Lyriks.com
Dotan Cohen wrote:
While all the reasons for linux not 'feeling' as fast as windows booting are very valid and interesting for me, I am not going to explain all that to my girlfriend. She will just say OK and boot into windows if all that she wants to do is check her hotmail. And, yes, when she powers the machine off, she doesn't care how long it takes- shes not waiting for it. Why can't linux shift some of the things it does during boot to doing them during shutdown?
If it did, it wouldn't be Linux. I'm sure if you really wanted to hack the startup scripts and take things like syslogd, and a hand full of other daemons or background them you could save a few seconds. However, Linux's roots is Unix and Unix runs on computers that stay up 24/7. If you don't shutdown, just log off and turn your monitor off, it will be as fast, if not faster than Windows.
Rob
If it did, it wouldn't be Linux. I'm sure if you really wanted to hack the startup scripts and take things like syslogd, and a hand full of other daemons or background them you could save a few seconds. However, Linux's roots is Unix and Unix runs on computers that stay up 24/7. If you don't shutdown, just log off and turn your monitor off, it will be as fast, if not faster than Windows.
You really don't want to shutdown your computer that often anyway. Startup and shutdown are stressful on your hardware. The more you do it the higher your chance of something dying. An example being that most computer's fans don't keep going after the computer is shutdown. During the time from when you shutdown, until the computer has naturally cooled, the insides of your computer are exposed to increased high tempertures.
-- Michael mogmios@mlug.missouri.edu http://kavlon.org
On Sat, 2005-05-03 at 13:00 -0800, Michael wrote:
You really don't want to shutdown your computer that often anyway. Startup and shutdown are stressful on your hardware. The more you do it the higher your chance of something dying.
As a colleague once said: "The only time my computer fails to turn on is after I've turned it off". :-)
An example being that most computer's fans don't keep going after the computer is shutdown. During the time from when you shutdown, until the computer has naturally cooled, the insides of your computer are exposed to increased high tempertures.
Not really. As soon as the computer is turned off, the heat sources all disappear, so the temperature has nowhere to go but down.
Sure, I suppose the air temperature in the case might go up a bit as the hot components cool down, but I don't think it would be significant. Not to mention that leaving it on causes everything to stay hot forever, which isn't good for the hardware either.
I think you'll get more stress from power cycling because of voltage spikes, inrush current, etc.
While powering cycling causes stress, so does leaving it on. I suppose you would need to figure out how long it's going to be off and decide which is the lesser of two evils. Leaving it powered up for short stretches would be preferable to frequent power cycling, but turning it off would probably be better than leaving it on but unused for long periods.
With the reliability of modern hardware, I don't think either way poses a significant problem. I leave mine on for convenience.
On Sat, 2005-03-05 at 13:35 -0800, Brian Mury wrote:
On Sat, 2005-05-03 at 13:00 -0800, Michael wrote:
You really don't want to shutdown your computer that often anyway. Startup and shutdown are stressful on your hardware. The more you do it the higher your chance of something dying.
As a colleague once said: "The only time my computer fails to turn on is after I've turned it off". :-)
An example being that most computer's fans don't keep going after the computer is shutdown. During the time from when you shutdown, until the computer has naturally cooled, the insides of your computer are exposed to increased high tempertures.
Not really. As soon as the computer is turned off, the heat sources all disappear, so the temperature has nowhere to go but down.
Not really. This is a dynamic situation with the heat being actively removed and temps non-uniform. You change the heat flow patterns and the temps will spike in some places.
I am an old-timer and remember the big old mainframes, especially the CDC 6600 and CDC 7600. These were cooled with circulating refrigerant. An undisciplined power outage definitely caused problems due to spiking temps. CDC did a study that showed a dramatic increase in failures about 7 days (as I recall) after a power failure.
Sure, I suppose the air temperature in the case might go up a bit as the hot components cool down, but I don't think it would be significant. Not to mention that leaving it on causes everything to stay hot forever, which isn't good for the hardware either.
I think you'll get more stress from power cycling because of voltage spikes, inrush current, etc.
While powering cycling causes stress, so does leaving it on. I suppose you would need to figure out how long it's going to be off and decide which is the lesser of two evils. Leaving it powered up for short stretches would be preferable to frequent power cycling, but turning it off would probably be better than leaving it on but unused for long periods.
With the reliability of modern hardware, I don't think either way poses a significant problem. I leave mine on for convenience.
Graham Campbell wrote:
I am an old-timer and remember the big old mainframes, especially the CDC 6600 and CDC 7600. These were cooled with circulating refrigerant. An undisciplined power outage definitely caused problems due to spiking temps. CDC did a study that showed a dramatic increase in failures about 7 days (as I recall) after a power failure.
I am not that expert on this one, but isn't cooling with a refrigerant similar to a car engine? The temperature of the refrigerant is going up because it stopped circulating?
But most of modern PCs are air-cooled like old Porsches or VW Beatles, and you *were* able to cool down overheated engines by switching off the engine.
If a modern car overheats, you should not switch it off immediatly because of the refrigerant.
On Sun, 2005-03-06 at 02:20 +0100, Markus Huber wrote:
Graham Campbell wrote:
I am an old-timer and remember the big old mainframes, especially the CDC 6600 and CDC 7600. These were cooled with circulating refrigerant. An undisciplined power outage definitely caused problems due to spiking temps. CDC did a study that showed a dramatic increase in failures about 7 days (as I recall) after a power failure.
I am not that expert on this one, but isn't cooling with a refrigerant similar to a car engine? The temperature of the refrigerant is going up because it stopped circulating?
But most of modern PCs are air-cooled like old Porsches or VW Beatles, and you *were* able to cool down overheated engines by switching off the engine.
I think the key to understanding this is the difference between heat and temperature. Liquid coolants differ from gas coolants only in the magnitude of the thermal properties things like heat capacity, etc. (I never learned thermodynamics very well). However the main idea is that if you change heat flow patterns dramatically, then temperatures will also change dramatically. Eventually everything will again reach a steady state with a lower temperature.
If a modern car overheats, you should not switch it off immediatly because of the refrigerant.
-- Regards Markus Huber
On Sat, 05 Mar 2005 15:19:18 -0500, Rob Miracle rwm@photo-miracles.com wrote:
Dotan Cohen wrote:
While all the reasons for linux not 'feeling' as fast as windows booting are very valid and interesting for me, I am not going to explain all that to my girlfriend. She will just say OK and boot into windows if all that she wants to do is check her hotmail. And, yes, when she powers the machine off, she doesn't care how long it takes- shes not waiting for it. Why can't linux shift some of the things it does during boot to doing them during shutdown?
If it did, it wouldn't be Linux. I'm sure if you really wanted to hack the startup scripts and take things like syslogd, and a hand full of other daemons or background them you could save a few seconds. However, Linux's roots is Unix and Unix runs on computers that stay up 24/7. If you don't shutdown, just log off and turn your monitor off, it will be as fast, if not faster than Windows.
Rob
I'm not talking about 'a few seconds'. We have two computers here: and AMD Duron 1ghz with 512 ram and a P4 1.5ghz with 256 ram. On both of them XP loads and connects to the internet (cable modem, no router) in under a minute and 10 seconds. Fedora core 3 takes over two minutes on the AMD computer. I couldn't install Fedora on the P4 computer, but SUSE also took what seemed like forever to load and connect.
The fact is, when she loads XP, she sits by and waits for it to load. On the rare occasion that she is willing to start Fedora, she is not patient enough to sit by and wait. And I agree: a minute is not enough time to do anything. But in two I can make it to the kitchen to pour a drink and open another pack of waffles.
I'm not saying that Fedora in paticular or linux in general HAS to start faster. I personally dont even care. I'm just saying that if we want to make it more accessable then it has to start faster. Or at least appear to. Even if it is doing something in the backgorund that does not interest the end user.
I have to read the rest but I would like to add something to support one reason Microsoft wants to tout their quick boot times. Although Microsoft tried and tried to fix the blue screen of death they couldn't eliminate the problems, so they switched gears to have a quicker reboot...
Also the os is not finished booting when you get a prompt to login, so comparing how long it takes prompt's to show up is not accurate. Showing a prompt x or terminal is the last thing linux distro's do. xp shows the prompt, but still "boots" in the backround. You have to force it to wait for the network to get it to work properly on ad domains some times.
If linux distro's were to do the same thing it would probably show a prompt right after mounting the file systems.
In short xp only pretends to boot faster, ms figured that people would be more understanding if they got a prompt right away, even if the total time of waiting was the same.