Linux Faster?

[requiemnoise]

To simply say something is faster is totally meaningless without context. One can say that Linux boots faster and be correct but then how does that correlate to a real activity like working on photos or playing a game or surfing the web?

If only those things count. The focus of app testing should be based around
library calls and footprint size of the app.

 

[heatlesssun]

I've been multi-booting various windows and linux installations for quite some time now (10+ years) and I've spent many hundreds of hours working in linux at both work and home. I still don't "understand". I see linux as a very viable operating system for servers, databases and for outdated machines... I still don't see how anyone with decent hardware would want to run linux outside of licensing issues much less brag about its performance compared to windows for such activity.

This thread was not intended to bash linux. I am honestly trying to understand the reasoning behind many peoples choices on these forums.

There is no question if you are hosting a server or do heavy computing (engineers and scientists) linux can benefit you considerably. Even when this day arrives for me I will still be dual-booting windows for everything else.

I think there are two primary reasons.

1) Some people love computers and love to play around with something that's as open and easy to manipulate. For them its all about the joy they get from technology.

2) A more colorful crowd is the anti-Microsoft gang. Microsoft is unique in the tech world because it a large and well coordinated lobby of folks who really think Microsoft is an evil company who want to control everything and love to make others think the same. Slashdot, The Inq are sites that come to mind are they are plenty more.

If I had the time I'd love to be a Linux guru, but there's so much to do with Windows and it pays the bills so that's where I focus my time.

 

[heatlesssun]

If only those things count. The focus of app testing should be based around
library calls and footprint size of the app.

You're talking about performance metrics which are important but don't necessarily translate into real world performance. Real world use is the ultimate in app testing.

 

[melteye]

If you really want to understand Linux, you really have to understand why certain distros exist. You have to research each companies or developers history. Yes, it is a burden to understand, but that is part of understanding Linux. Maybe, the creation of too much freedom caused many headaches to new users. There are practically no revenue to be earned as a Linux developer unless you belong to the big supported groups. Most Linux revenue is earned from companies who offer commercial supports. Many IT tests are supported by Microsoft. The large independent studies require funding. There are various fan boy stats you will encounter if you search the internet. Most people will assume these fan boys are bias. I guess you can clearly see, it is very hard to get the words out without the money in the technology business.

If you are a fan of tradezines, you will mostly get the wrong information. Tradezines exist to sell ads. Linux is free, developers will not pour their entire savings to expose their products if they don't even have a revenue model. Every commercial companies exist to support and feed their employees, so why would they constantly support Linux unless they found a niche revenue market for it? When the developers who moonlight as a Linux developer at night has enough money to fund a test by large IT research firms? There are great CEOs like Mark Shuttleworth (Ubuntu) who has poured few millions with constant loss for many years. How long Ubuntu will survive with a constant loss is unknown. They took hardcore Linux like Debian and made it a lot easier.

The information you seek are available everywhere on the net. But, it is good idea to know Linux is the kernel. How they operate is up to developers who package them. If you search around, there will be even a large debate based around Fedora vs Ubuntu's speed test. These speed charts change after each new releases as the kernel, window managers, X, and apps run on top it constantly go through changes. Also, each distros have their own philosophies. Some are cater to newbies, hackers, servers, 100% open, speed, etc. Anyway, you will find your information if you look around. Unless you are asking us to search the links for you.

I actually do understand why certain distributions exist. I also have read around extensively and there is plenty of benchmarks that clearly show linux is better at heavy multi-tasking and heavy computations. The topic of this thread was everyday desktop use and not special tasks.

I'm not actually sure why you're lecturing me about tradezines when I'm talking about actual experience with the distributions.

I've also read plenty of debates between the speeds of different distributions and I've ran both in many cases to see for myself. I am also aware of the different philosophies. This still doesn't change the fact I've experienced sub-par performance for everyday activity such as browsing, opening windows, watching videos, etc. with modern hardware.

My media computer is running linux and I seriously doubt I could watch HD content on that hardware if it was running windows... but this isn't the case for my main desktop.

 

[Joe Average]

Well then, if you truly want outstanding performance in heavy multitasking, Linux wouldn't be the answer.

UNIX would, or even BSD or it's derivatives. I've done similar tasks on Windows that choke it fairly severely, the same tasks drag Linux into the slow lane pretty massively, and the same tasks on a FreeBSD installation had the OS basically saying "What, is that all you want done? Come back when you've got about 10,000 more things to handle simultaneously and we'll talk."

I was extremely impressed and basically blown away at how damned well that OS multitasks, and after using real UNIX for decades in my career (now retired, thankfully), FreeBSD kicks ass.

But I can't stand *NIX or derivatives soooo...

 

[requiemnoise]

You're talking about performance metrics which are important but don't necessarily translate into real world performance. Real world use is the ultimate in app testing.

I'm confused. But that is what you asked for.. You just said, the speed of library calls are the most important feature. You know who is going to lose if it is only based on that idea. The boot time, able to be modular, threads, libraries, and size of the applications all matters when it comes to speed.

 

[synaptical]

So now windows transcodes video, number crunches, or runs web apps faster? No.

People seem to get confused between computer speed and GUI speed. The windows gui seems to be faster and slightly more responsive because of the OS architecture and how the GUI is more integrated into 'doze. In linux, the GUI is just another layer over the system, so it appears subjectively slower. Maybe objectively it even is slower. Once you launch your browser or app and get the probably 0.1 second performance hit you are not going to notice any difference.

But again, that's the GUI, not the raw computing abilities of the computer. Set up two web servers, video renderers, etc. with equivalent hardware, one windoze and one linux, and I would wager that actual performance is probably close to the same, with linux probably getting a slight edge in speed and an even larger edge in stability.

As others have said, boot times can also be significantly faster in linux, with speeds now getting down to the 12-15 second time frame (see bootchart threads on Arch linux forum, for example). Let's see you boot into windows in 15 seconds. Probably not. And that adds up over time, considering all the rebooting you have to do in windows. So take overall performance for windows down another notch.

Overall, therefore, imho linux is faster in most things, with windoze appearing to get a slight edge in the GUI. But that is also a virgin install. Give it a few weeks/months and most windoze systems (meaning the average person running them) start getting bogged down with spyware, registry corruptions, possible viruses, etc. not to mention all the resources that need to be used for scanning, defragging, etc. which you don't have to do in linux. The use of those resources will contribute to a performance hit long-term. And even if speaking of elite users, how many times do they reinstall for drivers, etc.? You don't have to do that in linux, either. For the average user, I think it's safe to say that linux is going to be way faster 6-12 months down the road from the point of fresh install than windoze.

 

[mryerse]

Does linux pre-cache items (like windows prefetch)? That seems to be the difference to me. Also, linux boots way faster than my windows system. But the bottom line to me is that both have things I need that the either doesn't.

 

[synaptical]

Linux can pre-cache, with the preload daemon, for example. Some linuxes, like puppy linux offhand, can also run completely from ram, which make them *extremely* fast. In those cases, subjectively they are faster even than 'doze.

 

[Gorankar]

How easily and fast they can open the app, do what needs to be done in it, close said app, and move on to the next thing they want/need to do, are the only metrics most users gives a damn about. It's horribly unscientific, I know.

 

[requiemnoise]

Does linux pre-cache items (like windows prefetch)?.

Yes, it is called, preload.

That seems to be the difference to me. Also, linux boots way faster than my windows system. But the bottom line to me is that both have things I need that the either doesn't.

That is very understandable. I have multiple different Windows VMs from FreeDOS to Win7 running top of Linux. Not everyone works at Microsoft are evil. Most Linux users just hate their political stands in the technology. In this capitalistic world, money can manipulate everything. That was my point. You will find all your answers if you do your own research. Linux users are heavy bloggers.

 

[synaptical]

I'll bet if someone did an unbiased, long-term study, they would find that windows apps on the whole crash a LOT more often than they do in linux. I can't remember the last time an app crashed on me in linux (known stable apps, not development version apps).

So if that turned out to be true (and from personal experience I believe it is -- would still like to see an unbiased study to verify the impression) and you are being completely honest, you would have to factor that into the long-term equation, too. Something might *seem* faster in the short term, but if you have to open it two or more additional times (after it crashes, causing who knows what loss), that is obviously not "faster" from a productivity standpoint.

 

[melteye]

So now windows transcodes video, number crunches, or runs web apps faster? No.

People seem to get confused between computer speed and GUI speed. The windows gui seems to be faster and slightly more responsive because of the OS architecture and how the GUI is more integrated into 'doze. In linux, the GUI is just another layer over the system, so it appears subjectively slower. Maybe objectively it even is slower. Once you launch your browser or app and get the probably 0.1 second performance hit you are not going to notice any difference.

But again, that's the GUI, not the raw computing abilities of the computer. Set up two web servers, video renderers, etc. with equivalent hardware, one windoze and one linux, and I would wager that actual performance is probably close to the same, with linux probably getting a slight edge in speed and an even larger edge in stability.

As others have said, boot times can also be significantly faster in linux, with speeds now getting down to the 12-15 second time frame (see bootchart threads on Arch linux forum, for example). Let's see you boot into windows in 15 seconds. Probably not. And that adds up over time, considering all the rebooting you have to do in windows. So take overall performance for windows down another notch.

Overall, therefore, imho linux is faster in most things, with windoze appearing to get a slight edge in the GUI. But that is also a virgin install. Give it a few weeks/months and most windoze systems (meaning the average person running them) start getting bogged down with spyware, registry corruptions, possible viruses, etc. not to mention all the resources that need to be used for scanning, defragging, etc. which you don't have to do in linux. The use of those resources will contribute to a performance hit long-term. And even if speaking of elite users, how many times do they reinstall for drivers, etc.? You don't have to do that in linux, either. For the average user, I think it's safe to say that linux is going to be way faster 6-12 months down the road from the point of fresh install than windoze.

Who are you responding to? I don't believe anyone said windows number crunches faster and the GUI is what I'm speaking of... did you read my original post? I said it lags in X-Windows. Go back and actually read the posts. Did you also read that I'm using NOD32 which has less than 1% average overhead.

Yes... defragging while idle really effects performance doesn't it. And as far as boot time I'm currently at about 20 seconds using Windows 7 and I haven't rebooted since the initial install. I use Sleep and it takes a whole of two seconds to turn on.

 

[melteye]

I'll bet if someone did an unbiased, long-term study, they would find that windows apps on the whole crash a LOT more often than they do in linux. I can't remember the last time an app crashed on me in linux (known stable apps, not development version apps).

So if that turned out to be true (and from personal experience I believe it is -- would still like to see an unbiased study to verify the impression) and you are being completely honest, you would have to factor that into the long-term equation, too. Something might *seem* faster in the short term, but if you have to open it two or more additional times (after it crashes, causing who knows what loss), that is obviously not "faster" from a productivity standpoint.

I doubt any application I would be using is going to crash every other time . I've had plenty of software crashes in linux but I'm sure thats not the normal occurence.

 

[cameron122000]

It is "faster" because it uses less resources in general. However, you won't be able to play Windows games on it unless you use Wine, so if that's the only reason why you're thinking about it, don't switch. Wine is still very buggy and is not ready to handle all Windows functions.

Everybody takes Linux in the wrong way. It is not out there to stop you from using Windows, it is an alternative, just like OS X is an alternative. The Windows guys say why would you want to compile your own kernel when I can install my OS in 10 minutes? Windows, Linux, and OS X have different uses and target groups. Windows targets generally business and workforce-oriented users. Linux are the people that like to tinker with their operating system and how their programs work (i.e. developers, etc.), and OS X is targeted towards the creative crowd with Photoshop and tools like that.

You can use some programs cross-operating system, but those are just the overall generalizations of each operating system and their use. There is no "best," there are choices.

 

[synaptical]

Who are you responding to? I don't believe anyone said windows number crunches faster and the GUI is what I'm speaking of... did you read my original post? I said it lags in X-Windows. Go back and actually read the posts. Did you also read that I'm using NOD32 which has less than 1% average overhead.

Yes... defragging while idle really effects performance doesn't it. And as far as boot time I'm currently at about 20 seconds using Windows 7 and I haven't rebooted since the initial install. I use Sleep and it takes a whole of two seconds to turn on.

I'm responding to the people who just spewed crap like "for day to day use linux is slower, it's just not as developed," blah blah blah.

My day to day use includes apache, php, mysql, video encoding, web surfing, spreadsheet use, document writing, etc. Linux multitasking is so much superior to windows to me I can't even imagine having to do all those tasks on windoze, not to mention some of these machines doing duty file and web serving, too. Yeah, just what I want - my server to be "defragging" when "idle," no thanks.

For another thing, whatever might be lost in a millisecond of speed opening an application, so much more is gained in the customization abilities of the linux desktop that it wins hands down on that point alone, in my opinion. Some of the functions can be approximated in 'doze with add-ons (multiple desktops, etc.) but even those don't come close to a personally configured linux desktop. I can open a terminal with a keystroke, transfer any number of files with a simple command, and then go on my way barely without blinking. In windows, you'd have to navigate to the folders, select, cut/copy/paste, etc. Takes a lot longer.

Day to day use in linux is therefore faster overall, I maintain, even when an individual app might take a quarter of a second longer to open. I think one maybe has to grow up a little from the "instant gratification" model to understand that something isn't necessarily better just because it *appears* on the surface to be faster.

 

[requiemnoise]

I actually do understand why certain distributions exist. I also have read around extensively and there is plenty of benchmarks that clearly show linux is better at heavy multi-tasking and heavy computations. The topic of this thread was everyday desktop use and not special tasks.

Then recompile your kernel and remove unnecessary functions that behaves like a server.
Another option is look for a simpler Linux distro.

I'm not actually sure why you're lecturing me about tradezines when I'm talking about actual experience with the distributions.

I can't read your mind. I would love to have that power. You never mentioned your background.

I've also read plenty of debates between the speeds of different distributions and I've ran both in many cases to see for myself. I am also aware of the different philosophies. This still doesn't change the fact I've experienced sub-par performance for everyday activity such as browsing, opening windows, watching videos, etc. with modern hardware.

Are you saying, Firefox is slower on Linux than Windows? Which part? rendering or load time? I think gecko engine is fast. People are working on a different engine for Firefox. Try others.

My media computer is running linux and I seriously doubt I could watch HD content on that hardware if it was running windows... but this isn't the case for my main desktop.

Also, there are many beta drivers you can try that will utilize your GPU for videos for Linux.

 

[synaptical]

Don't forget also that the nested menu model -- which windows is almost totally based on as its main interface metaphor -- is considered extremely poor interface design. The user has to hunt for the app, sometimes through multiple levels of nested menus, first to even find what they want, and then to physically navigate to it with the mouse through narrow ~20-30 pixel corridors, which are subject to popping closed again if the mouse goes even slightly outside the narrow band. That's a major slowdown right there (totally aggravting, too) -- much slower than any actual processing slowness of the GUI, real or imagined.

Then that's "solved" in windows by another poor interface concept: the desktop icon, which loses productivity as you often have to move open applications out of the way, minimize them, etc. to get to the icon. And let's face it, you can't make a desktop icon for every application. The more icons you create to avoid having to go to the menu, in fact, the harder the correct icon is to locate as the desktop gets more and more cluttered. The taskbar concept (which MS largely stole from apple) is a little better - at least that autohides out of the way and is a relatively easy target to hit with the mouse. But even then, realistically it can't contain everything.

By contrast, a single right-click application menu with minimal nested items is superior to windows, as you can click anywhere in the desktop to activate the menu, not have to target a small portion of the lower-left corner to activate the "Start" menu (which itself is stupid since most users are right handed ). For example, if I'm scrolling a web page with the mouse, I only have to move the mouse a fraction of a centimeter to the right to right click the desktop menu, scroll down, and open an app. In windows, I'd have to move the mouse all the way from the right side of the screen to the lower left to activate the start menu, and then hunt for the app (or else usually go to the left also for an icon, probably having to move or minimize the browser or other app to get to it). So in real-world usage, linux is faster.

Even better, most linux desktops encourage keybindings way more than windoze does. Barring brainwaves being able to open apps without any physical intervention, keybindings are obviously the superior way to access applications -- you don't even need to move your hands from the KB. It's much faster even than having to navigate to and click on an icon on a totally clean desktop. And if you compare opening an application with a keybinding to having to navigate and hunt through a nested "Start" menu, there is no contest at all. The keybinding will win every time. So it appears when you take actual *usage* habits into it, linux in most cases is actually demonstrably faster.

 

[requiemnoise]

It just dawn to me. If users are complaining the window manager isn't snappie, so it feels slower, turn on your 3d acceleration if your driver supports it. That will give you options to snap your windows tight like Vista gui. Personally, 3d stuff gives me a headache.

 

[Joe Average]

I've got 46 applications and Windows tools "bound" to keyboard shortcuts... how many have you got, synaptical? As I've stated before, I don't click the Start Menu at all anymore, that's what the Windows key is for, press it and type a letter or two, what I'm looking for shows up on the list, a tap or two and bam, there it is.

And that's for stuff I don't have keyboard shortcuts to... go figure.

 

[requiemnoise]

I've got 46 applications and Windows tools "bound" to keyboard shortcuts... how many have you got, synaptical? As I've stated before, I don't click the Start Menu at all anymore, that's what the Windows key is for, press it and type a letter or two, what I'm looking for shows up on the list, a tap or two and bam, there it is.

And that's for stuff I don't have keyboard shortcuts to... go figure.

I thought you were a Linux GOD. You keep on telling me that other threads. 46? That is very small. Can you satisfy someone with that? Fluxbox... unlimited... That isn't just a launcher. That is even launching commands.

 

[Joe Average]

Uhmmm... you must have me confused with some other GOD, dude. I don't run Linux on my own hardware except in VMs, but I've got more certs for Linux than you probably have years on your ID...

46 keyboard shortcuts that launch 46 different applications with keypresses... too small... you're living up to your sig, I'll give you that much. While I won't place you on ignore using the forum software, I think I've had enough of your non-sensical almost-seemingly-AI-bot generated babbling to just manually not pay attention to you anymore.

Can I get someone to second the motion?

 

[requiemnoise]

Uhmmm... you must have me confused with some other GOD, dude. I don't run Linux on my own hardware except in VMs, but I've got more certs for Linux than you probably have years on your ID...

46 keyboard shortcuts that launch 46 different applications with keypresses... too small... you're living up to your sig, I'll give you that much. While I won't place you on ignore using the forum software, I think I've had enough of your non-sensical almost-seemingly-AI-bot generated babbling to just manually not pay attention to you anymore.

Can I get someone to second the motion?

So you admit, you aren't a Linux expert, but love to bash Linux, because it is fun. Before you say, I am a hypocrite. I have long years of managing Windows projects just like yourself. I have worked on least 8 migration projects in the last 15 years. Usually, a good honesty gets the elder respect for me.

 

[synaptical]

I've got 46 applications and Windows tools "bound" to keyboard shortcuts... how many have you got, synaptical? As I've stated before, I don't click the Start Menu at all anymore, that's what the Windows key is for, press it and type a letter or two, what I'm looking for shows up on the list, a tap or two and bam, there it is.

And that's for stuff I don't have keyboard shortcuts to... go figure.

You are by far the exception. MOST people use point-and-click on icons or navigating the crappy Start menu, I think you would have to agree.

And it sounds like you learned doing it that way in the first place from using linux.

 

[Joe Average]

Actually, I learned keyboard shortcuts from Amiga OS which to me is still superior to every OS made since. Everything since is so lackluster and boring to me... until Windows 7 happened to come along. I don't know what fire was lit under the developer's collective asses, or who spiked the soda machines at Microsoft, but something happened there in 2008 - might have been Bill Gates finally stepping away, who knows. All I know is that something happened and suddenly they created some magic, aka Windows 7.

To be honest, originally it was from my trusty Timex-Sinclair 1000 over 25 years ago, but even that wasn't my first...

 

[Joe Average]

So you admit, you aren't a Linux expert, but love to bash Linux, because it is fun. Before you say, I am a hypocrite. I have long years of managing Windows projects just like yourself. I have worked on least 8 migration projects in the last 15 years. Usually, a good honesty gets the elder respect for me.

I don't bash Linux, I rarely ever bash anything. Bashing means ignorant spouting off at the mouth, whereas what I - and many others around here - do our best to do is have somewhat intelligent discussions about the pros and cons of any given topic at hand.

8 migrations... not sure what that means because it's so out of context and lacking in actual information, but I assure you, you don't wanna see my resume...

I have nothing but respect for intelligence when it's presented; it's the ignorance that'll set me off like a fuse under a magnifying glass on a clear day in Death Valley. And yes, that's some serious simile I pulled outta thin air there... right off the top of my head. Come on, ya gotta be impressed with that one... right... right... right...

 

[requiemnoise]

Linux is fucking slow as far as day-to-day usage is concerned compared to Windows.

I back it up with decades of Windows usage and a shitload of Linux/UNIX stuff to boot. It's just not as well developed and not nearly as fast as the Windows GUI is nor in day-to-day usage, it just isn't.

Sir. What do you mean by this? I am sorry. This was from the same thread few hours ago. Also, haven't you complained and yelled at many people for using VMs to test?

I don't bash Linux, I rarely ever bash anything. Bashing means ignorant spouting off at the mouth, whereas what I - and many others around here - do our best to do is have somewhat intelligent discussions about the pros and cons of any given topic at hand.

I am bit confused.

 

[Joe Average]

If I say Linux is slow and I can prove it, it's not bashing. And I can most certainly prove it...

I run VMs to gain experience with OSes of different kinds, to learn their ins and outs, and how to support them so I can put some cash in my pocket. I don't use VMs for beta testing because they are effectively useless for such purposes. There's a big huge world of difference between just loading up a Linux distro to check it out and see what's what, and loading up an actual beta OS for the purposes of offering feedback to the manufacturer on bugs on a given platform, inconsistencies with the code or drivers, usability issues, etc.

To me, VMs are for "playing around," basically; bare metal hardware is the serious stuff and used for beta testing.

 

[synaptical]

Sir. What do you mean by this? I am sorry. This was from the same thread few hours ago. Also, haven't you complained and yelled at many people for using VMs to test?


I am bit confused.

I don't think it's you.

Linux is fucking slow as far as day-to-day usage is concerned compared to Windows.

It's just not as well developed and not nearly as fast as the Windows GUI is nor in day-to-day usage, it just isn't.

I've already debunked the "well-developed" claim by pointing out only some of the many serious user interface flaws in windows. Joe Average thinks he refuted them -- by agreeing with me that keybindings are superior to the windows interface!

And speaking of well-developed, remember the last major security hole a few weeks ago that infected millions of computers and took MS almost 2 weeks to patch? Supposedly 1 in 3 windows computers are still vulnerable to attack. Wow, that's really well-developed stuff.

http://www.infoworld.com/article/09/01/16/1_in_3_Windows_PCs_vulnerable_to_worm_attack_1.html

Meanwhile, linux servers power at least half the web mostly without major incident.

 

[synaptical]

n/t

 

[requiemnoise]

If I say Linux is slow and I can prove it, it's not bashing. And I can most certainly prove it...

How would you prove it. You said

Uhmmm... you must have me confused with some other GOD, dude. I don't run Linux on my own hardware except in VMs, but I've got more certs for Linux than you probably have years on your ID...

Oh, did I mentioned I... never mind...Yes, you must have tons of Linux certs.
Also, you said you have been beta testing for MS since 1986. What was his name? We might had a mutual friend. Who knows we could all be hanging around on the same BBS.

It's just not as well developed and not nearly as fast as the Windows GUI is nor in day-to-day usage, it just isn't.

So the speed of the GUI makes the difference? I see... ON VM? right.. You like the test the video acceleration to prove Linux sucks in VM. Thank you.

I run VMs to gain experience with OSes of different kinds, to learn their ins and outs, and how to support them so I can put some cash in my pocket. I don't use VMs for beta testing because they are effectively useless for such purposes. There's a big huge world of difference between just loading up a Linux distro to check it out and see what's what

What??? But, you said 20 minutes ago...

I don't run Linux on my own hardware except in VMs

So, which certs you got for Linux? Linux Vms?

 

[kac77]

If I say Linux is slow and I can prove it, it's not bashing. And I can most certainly prove it...

.

Aside from games developed around Direct X and later ported to Open GL. Most linux experts would never say something like that. Especially in the enterprise space. While MSSQL has more features, MYSQL wipes the floor on query returns and inserts of a certain size. Apache as well, which is far thinner, can handle a much larger load than IIS. Even Gnome, if custom compiled can fly like the wind if that was your goal. NFS wipes the floor with anything CIFS/ Samba related. Print Servers are faster as well. Oracle DB's on Linux again are far more responsive.

In what area are you speaking of ? Ease of use is one thing, but the beauty of linux is being able to compile only the services that are needed and thusly be able to obtain a speed far greater than what your average Windows Server would be able to do.

 

[SmokeRngs]

So now windows transcodes video, number crunches, or runs web apps faster? No.

I specifically stated earlier that any type of number crunching between Linux and Windows is going to be basically identical with certain exceptions on the same hardware. The F@H SMP client is an example of this. There is generally a bit of a performance advantage to the Linux client because the client was developed for Linux and ported to Windows. Due to the differences between the underlying OSes, the Linux client is actually more efficient in real world usage. In the case of video encoding, a lot depends on the encoder you are using and the settings you are using. If you are using the same encoder on both OSes with the same settings, you are more than likely going to get the same results unless the encoders are not optimized to the same extent. This is again a case of efficiency and doesn't actually have anything to do with the OS but the program itself.

 

[kac77]

I specifically stated earlier that any type of number crunching between Linux and Windows is going to be basically identical with certain exceptions on the same hardware. The F@H SMP client is an example of this. There is generally a bit of a performance advantage to the Linux client because the client was developed for Linux and ported to Windows. Due to the differences between the underlying OSes, the Linux client is actually more efficient in real world usage. In the case of video encoding, a lot depends on the encoder you are using and the settings you are using. If you are using the same encoder on both OSes with the same settings, you are more than likely going to get the same results unless the encoders are not optimized to the same extent. This is again a case of efficiency and doesn't actually have anything to do with the OS but the program itself.

I think you are speaking of best case scenarios / systems that are tuned for a particular experiment. However, real world cases would show the opposite.

A perfect example would be a system with 1GB of RAM one running Vista the other Linux. In this case the hardware is the same. However, due to the nature of Vista I can guarantee you that if you were to encode video or number crunch the lInux system would beat the Windows environment every single time using the same codec. Probably by large margins as well. Why? Because the OS doesn't get out of the way of the primary function you are trying to perform. Vista in this example would be busy using cycles dealing with prefetch and probably a good deal of other services with this amount of memory.

You would have to give each system 2GB maybe a little more in order for Vista to match the performance of the linux distro. Now if you were comparing just base kernel to kernel, yeah without all of the other crap running you would be more likely to see similar performance. The problem is I know you can convert video using a particular codec with no GUI in Linux, however, in Windows exactly how would you perform that feat? How would you even install the codec? Would the codec even allow you to install it since all of the libraries needed for a msi executable would be missing?

 

[SmokeRngs]

I think you are speaking of best case scenarios / systems that are tuned for a particular experiment. However, real world cases would show the opposite.

A perfect example would be a system with 1GB of RAM one running Vista the other Linux. In this case the hardware is the same. However, due to the nature of Vista I can guarantee you that if you were to encode video or number crunch the lInux system would beat the Windows environment every single time using the same codec. Probably by large margins as well. Why? Because the OS doesn't get out of the way of the primary function you are trying to perform. Vista in this example would be busy using cycles dealing with prefetch and probably a good deal of other services with this amount of memory.

You would have to give each system 2GB maybe a little more in order for Vista to match the performance of the linux distro. Now if you were comparing just base kernel to kernel, yeah without all of the other crap running you would be more likely to see similar performance. The problem is I know you can convert video using a particular codec with no GUI in Linux, however, in Windows exactly how would you perform that feat? How would you even install the codec? Would the codec even allow you to install it since all of the libraries needed for a msi executable would be missing?

I have yet to see much of a difference of UI overheard with regards to available processing power with either Linux or Windows on a modern system. Yes, you can have much lighter UIs for Linux than you can with Windows but the UI even in Windows doesn't eat up a lot of CPU power. Looking at my current uptime with Win7, which is over 5 days, dwm.exe has used a total of 2 hours of processing time. That's not very much. There are plenty of background tasks in Windows but there can also be a lot of background tasks with Linux depending on what distro you use. Don't get me wrong, I'm a Linux fan and generally prefer it over Windows, but I'm not going to lie about Windows' system resource usage.

As for your encoding situation, you're mixing up a few things. First of all, a codec and an encoder are two different things. I do not need an h.264 codec installed to use an x264 encoder. Also, you can do encodes in Windows without a GUI. There are plenty of command line encoders for Windows. I'd say most of them are CLI encoders but you don't see that due to the GUI front end. I know the x264 encoder I use in Windows with Handbrake is CLI only and the GUI does nothing but start up the CLI encoder with the options selected. I haven't tried it, but I bet you could boot to command prompt only in Windows and run the x264 CLI encoder. I also doubt you would see much difference in processing times.

 

[kac77]

I have yet to see much of a difference of UI overheard with regards to available processing power with either Linux or Windows on a modern system. Yes, you can have much lighter UIs for Linux than you can with Windows but the UI even in Windows doesn't eat up a lot of CPU power.

Please check the example I gave with Vista. You know that it's overhead is huge and most definitely affects performance, which is why you brought up Windows 7. You wouldn't have done that unless overhead was an issue.

As for your encoding situation, you're mixing up a few things. First of all, a codec and an encoder are two different things. I do not need an h.264 codec installed to use an x264 encoder. Also, you can do encodes in Windows without a GUI. There are plenty of command line encoders for Windows. I'd say most of them are CLI encoders but you don't see that due to the GUI front end. I know the x264 encoder I use in Windows with Handbrake is CLI only and the GUI does nothing but start up the CLI encoder with the options selected. I haven't tried it, but I bet you could boot to command prompt only in Windows and run the x264 CLI encoder. I also doubt you would see much difference in processing times.

Actually I'm not. A codec, in the definition of this dialogue, is a collection of libraries based on a specification for video or audio. Within the codec you have the ability to encode or decode or both depends on the package. However, in terms of linux you most definitely aren't going to encode or decode without the baseline codec installed, or use a package that contains the libraries from multiple codecs.

Yes you can use Handbrake in CLI. But you didn't answer my original question. How are you going to install it without a GUI? Meaning if by some reason someone was to strip out the overhead, the GUI in Windows (Vista), how are they going to download since it's not hosted by FTP, and then extract it in order to run it? No one in windows will do that mainly because it's a PITA to do and it's far far harder to do there than in Linux. Therefore the example of testing Vista with 1GB of RAM and no GUI isn't likely to happen if ever. So we are stuck with the GUI, so to make it fair we run Linux with it's GUI. Now if we run both with 1GB of RAM, which is right at the mark where Vista fits within it's memory footprint. At the end of encoding when the file is created (possibly before) Vista will undergo a tango with prefetch. This eats up cycles which one of the OS's endures while the other doesn't. So in the end while both are using the same libraries one of them will most definitely finish before the other.

 

[SmokeRngs]

Please check the example I gave with Vista. You know that it's overhead is huge and most definitely affects performance, which is why you brought up Windows 7. You wouldn't have done that unless overhead was an issue.

I mentioned Win7 because that's what I have installed at the moment. Before this I was running Vista64 and Handbrake encoded at the exact same rate with the exact same version of Handbrake using the exact same settings. The time to encode is based primarily on raw CPU speed.

Actually I'm not. A codec, in the definition of this dialogue, is a collection of libraries based on a specification for video or audio. Within the codec you have the ability to encode or decode or both depends on the package. However, in terms of linux you most definitely aren't going to encode or decode without the baseline codec installed, or use a package that contains the libraries from multiple codecs.

You are mistaken because a codec is not required for encoding when the encoder has everything built in for encoding. The x264 encoder included with Handbrake will not playback x264 encoded files or h264 encoded files. You do not need an installed and loaded codec to encode and that is all that matters.

Yes you can use Handbrake in CLI. But you didn't answer my original question. How are you going to install it without a GUI? Meaning if by some reason someone was to strip out the overhead, the GUI in Windows (Vista), how are they going to download since it's not hosted by FTP, and then extract it in order to run it? No one in windows will do that mainly because it's a PITA to do and it's far far harder to do there than in Linux. Therefore the example of testing Vista with 1GB of RAM and no GUI isn't likely to happen if ever. So we are stuck with the GUI, so to make it fair we run Linux with it's GUI. Now if we run both with 1GB of RAM, which is right at the mark where Vista fits within it's memory footprint. At the end of encoding when the file is created (possibly before) Vista will undergo a tango with prefetch. This eats up cycles which one of the OS's endures while the other doesn't. So in the end while both are using the same libraries one of them will most definitely finish before the other.

You said nothing about downloading or installing anything while at the command line. You only said that you couldn't encode at a command line with Windows and that is false. Quit trying to change your argument because it doesn't make a damn bit of difference.

I'm also not going to take my system down to 1 gig of RAM and it's actually impossible for me to do so as I'm running 2 2gig sticks and I wouldn't even bother messing with my system to test what you say with a 1 gig system. Also, who's now talking about trying to setup a system that might favor one system over another? Face it, the vast majority of people doing video encoding are likely to have 2 gig of RAM at minimum no matter what OS they are using. Also, RAM caching isn't going to play much of a part in encoding, at least with x264 while doing DVD encodes. Either OS is going to free up whatever RAM is needed to make room for the encoding and will not be trying to smash other things in RAM so the caching is not going to be a factor. Also, when at the command line, I doubt Windows is going to be trying to pre-cache stuff which wouldn't even run when at a command line. Besides, with the encoding I have done I don't think I have gone above 150MB RAM usage for the CLI x264 encoder. RAM usage would more than likely increase with HD material but I have not done any HD material as of yet. Hell, I do a lot of other stuff including web browsing, watching movies, downloading stuff and moving files around while I'm doing the encoding and it doesn't affect my encoding speeds that I have been able to see and I've actually measured it. If that stuff doesn't noticeably affect my encoding speeds, the background OS tasks definitely won't affect the encodes.

You keep jumping around in your arguments and it's not helping you. I said before that I actually prefer Linux over Windows for the majority of stuff but I'm not going to say I get better performance in CPU intensive apps on one OS over another when it's not true.

 

[Klob]

I've generally found that bootup, login, and the big one....getting to a functional desktop..is generally quicker versus Windows.

Really? I've found the opposite to be true and the FUD spread about Linux booting faster than Windows by Linux zealots is just that, pure FUD. I do run Linux on one PC so I'm not saying this because I am a Linux hater.

 

[synaptical]

Really? I've found the opposite to be true and the FUD spread about Linux booting faster than Windows by Linux zealots is just that, pure FUD. I do run Linux on one PC so I'm not saying this because I am a Linux hater.

Let's see you boot to windows in 13 seconds. That's not FUD, that's fact. Most of the FUD is coming from the windows crowd, as usual.

 

[Joe Average]

*snip*


Let's see you boot to windows in 13 seconds. That's not FUD, that's fact. Most of the FUD is coming from the windows crowd, as usual.

That 13 second boot didn't just happen by itself - it required a concerted effort to MAKE Linux boot that fast, just as using BootVis on a laptop I have using XP 32 bit and a 5400 rpm hard drive with DDR 333 RAM can boot in 13.68 seconds after I hit it with a boot optimization pass.

Would we then argue the .68 second discrepancy, because I'd expect Linux to be far more accurate than just saying "13 seconds" and not carry it out to the 5th decimal place or something.

Oh yeah... What's your point?

Out of the box, a Win XP install will boot faster than a given Linux distro, such as Ubuntu - and I use Ubuntu as the example here because it's the most popular Linux distro on the planet, so it's comparable in general. After the install you can tweak the living hell out of both to optimize the boot speed, obviously.