Linux vrs Windows - FPS
by Jason Nabors · in Torque Game Engine · 02/29/2004 (12:30 pm) · 51 replies
Figured alot of you Linux users would find this enjoyable.
I recently tested the downloadable precompiled TorqueDemo's
for Linux and for Windows. I ran the multiplayer fps mod.
I found that under Linux, the game has a much higher Framerate then it does under windows even with lesser hardware.
For Instance: at 1024x768
Windows XP box: Athlon 64 3200+, 1 gig ddr400 ram, Nvidia
5950 Ultra video card
FPS: 111
Linux(Mepis) box: P4 2.2ghz, 1 gig ddr333 ram, Nvidia Geforce 4
TI 4200 video card
FPS: 157
As you can see, even with lesser hardware Linux out performed the Windows Box.
I recently tested the downloadable precompiled TorqueDemo's
for Linux and for Windows. I ran the multiplayer fps mod.
I found that under Linux, the game has a much higher Framerate then it does under windows even with lesser hardware.
For Instance: at 1024x768
Windows XP box: Athlon 64 3200+, 1 gig ddr400 ram, Nvidia
5950 Ultra video card
FPS: 111
Linux(Mepis) box: P4 2.2ghz, 1 gig ddr333 ram, Nvidia Geforce 4
TI 4200 video card
FPS: 157
As you can see, even with lesser hardware Linux out performed the Windows Box.
About the author
#22
Are you running that in 1024x768? Or what resolution? bit level?
I was using the latest Forceware drivers from nvidias website.
Version: 53.03
Release Date: December 9, 2003
Where do you get version 55 or greater? Special developers section?
Were you running the demo from GG website, or your own compiled version? I used the demos from GG website for both the linux and windows machines.
Thanks Robert, that was interesting, I'll have to see if I can get those drivers you speak of.
03/04/2004 (10:58 am)
Ah nice to know, Are you using windows 64bit beta OS? Windows XP?Are you running that in 1024x768? Or what resolution? bit level?
I was using the latest Forceware drivers from nvidias website.
Version: 53.03
Release Date: December 9, 2003
Where do you get version 55 or greater? Special developers section?
Were you running the demo from GG website, or your own compiled version? I used the demos from GG website for both the linux and windows machines.
Thanks Robert, that was interesting, I'll have to see if I can get those drivers you speak of.
#23
Jasons results are antedotal at best and can't generalize this concolusion that Linux runs games faster on lower spec hardware.
But you could only make semi-useful antedotal comparisons of video drivers. And even then, the brand of motherboard, ram and bios version and other things would come into play.
Adhoc benchmarking is even more useless as a general performance indicator anymore, especially when the performance relies on so many variables that are out of the control of what is being benchmarked. nVida is tweaking ( cheating in some cases ) with its driver for specific games.
As a highly specific performance indicator that might be another issue all together.
Certain things in Torque can limit the framerate regardless of the graphics card and the processor. For example when I was playing around with the fxGrass stuff I got the same FPS results on my 7500 Mobility as my Radeon 9500 ( softmodded 9700 ) once the view distance got past a certain distance the frames per second did not scale linearly anymore.
This was NOT an indicator of anything regarding the graphics card, driver or OS, but an indicator of where a bottle neck in Torque is. Both on WinXP
Now lets say that on Linux or MacOSX this was not the case and the fxGrass frames per second scales linearly, that means that the bottle neck is in the Windows specific portion of the code or possibly the driver implementation but it does not automatically mean the OS is to blame.
One of the things I do on a daily basis is track down performance anomolies and do peformance analysis ( benchmarking ). We are re-writing a massively multiplayer game server system right now that is moving from handling a few thousand conccurent users which right now is about 23 mega bits a second to tens of thousands of concurrent users and trying to lower the bandwidth at the same time!
03/04/2004 (10:58 am)
It would be useful to see what the stats for particular graphics cards with what version drivers playing back a specific journal file are.Jasons results are antedotal at best and can't generalize this concolusion that Linux runs games faster on lower spec hardware.
But you could only make semi-useful antedotal comparisons of video drivers. And even then, the brand of motherboard, ram and bios version and other things would come into play.
Adhoc benchmarking is even more useless as a general performance indicator anymore, especially when the performance relies on so many variables that are out of the control of what is being benchmarked. nVida is tweaking ( cheating in some cases ) with its driver for specific games.
As a highly specific performance indicator that might be another issue all together.
Certain things in Torque can limit the framerate regardless of the graphics card and the processor. For example when I was playing around with the fxGrass stuff I got the same FPS results on my 7500 Mobility as my Radeon 9500 ( softmodded 9700 ) once the view distance got past a certain distance the frames per second did not scale linearly anymore.
This was NOT an indicator of anything regarding the graphics card, driver or OS, but an indicator of where a bottle neck in Torque is. Both on WinXP
Now lets say that on Linux or MacOSX this was not the case and the fxGrass frames per second scales linearly, that means that the bottle neck is in the Windows specific portion of the code or possibly the driver implementation but it does not automatically mean the OS is to blame.
One of the things I do on a daily basis is track down performance anomolies and do peformance analysis ( benchmarking ). We are re-writing a massively multiplayer game server system right now that is moving from handling a few thousand conccurent users which right now is about 23 mega bits a second to tens of thousands of concurrent users and trying to lower the bandwidth at the same time!
#24
Then I remembered the DetonatorXP line of nVidia drivers, so I went out and got myself DetonatorXP (40's build, I can't remember exactly) from June 2002. That's over 20 months ago. I installed it, fired up torque, and whaddya know, absolutely brilliant. My rendering went back to what it was before; clear, crisp and absolutely no jitters. In fact, before I installed the ForceWare upgrade, I was indeed running a DetonatorXP driver!
From my experience, drivers can make all the difference in the world when it comes to performance. As of now, the 55.3 AND 55.2 ForceWare build have been nothing but disaster for me, contrary to what Robert said.
03/04/2004 (11:38 am)
I tend to agree with Jarrod's point of view, simply because of the shear discrepancy in drivers. A few days back, I tried installing the new nVidia ForceWare driver (55.2), dated December 2003; I have 733MHz and a GeForce2 MX 440 running on WinXP Pro. I fired up Torque, and lo-and-behold my frame rate was extremely poor. I didn't have any FPS tracking, but I can tell you it was as if I was running a cheap integrated graphics card, or more correctly, like a software renderer. It was virtually unplayable.Then I remembered the DetonatorXP line of nVidia drivers, so I went out and got myself DetonatorXP (40's build, I can't remember exactly) from June 2002. That's over 20 months ago. I installed it, fired up torque, and whaddya know, absolutely brilliant. My rendering went back to what it was before; clear, crisp and absolutely no jitters. In fact, before I installed the ForceWare upgrade, I was indeed running a DetonatorXP driver!
From my experience, drivers can make all the difference in the world when it comes to performance. As of now, the 55.3 AND 55.2 ForceWare build have been nothing but disaster for me, contrary to what Robert said.
#25
So we are stuck with whats put out to the public. Thats what I did my test on.
03/04/2004 (12:10 pm)
@Kamal, I agree with you that the drivers seem to make a huge difference. But are you going to ask your consumers to load older versions of a driver so they can play your game? Doubtful.So we are stuck with whats put out to the public. Thats what I did my test on.
#26
03/04/2004 (3:59 pm)
@Jason: Absolutely. It's assummed that the latest drivers by default SHOULD provide you with the best performance, as is the case most of the times. But, what I've gone through was just another example in how old drivers (they are still available on nVidia's website) or even different flavours of drivers can make a dramatic impact in performance. Your benchmark test may be, or may not be, influenced by this premise. That's all :).
#27
There sure was a difference in performance on the windows machine used in this test. FPS jumped from 111 to 250. This is more like what I expected from the hardware on the windows box. These were beta drivers though, and not officially released. Never the less they are staying on my system. Well until I put linux on that box and check fps.
Here is the link to those drivers incase anyone out there wants to try them.
downloads.guru3d.com/downloadget.php?id=741&file=5&evp=5add535c606f3d4c36c4f3c55...
03/04/2004 (9:42 pm)
I got ahold of some Nvidia drivers tonight and the version was 56.57 There sure was a difference in performance on the windows machine used in this test. FPS jumped from 111 to 250. This is more like what I expected from the hardware on the windows box. These were beta drivers though, and not officially released. Never the less they are staying on my system. Well until I put linux on that box and check fps.
Here is the link to those drivers incase anyone out there wants to try them.
downloads.guru3d.com/downloadget.php?id=741&file=5&evp=5add535c606f3d4c36c4f3c55...
#28
I really like that idea. Would answer those question quite nicely as well as give an indi an idea if there game should run "as good" on a potentially targeted PC that they don't have access to everyday in dev. (Like me and the mac)
As I put my indi hat on I really don't care if windows is faster or slower than a mac or Linux. I just care that I can get my game to reach as many buyers as possible. This info would help me makes some of those design idea vs machine limitation. (The if I implement that I will have to ensure it still runs well on X)
It would also be nice to see a break down of what games sell on what OS's of games that can run on all three. Would answer the other question of "if I create a game that is like this, where would I expect my customer base mostly be?" Might drive you to move to that OS and forget support for all others or might drive you to make that OS the developement one so you KNOW it will work and work well on it.
But all that is leading to a different topic than this thread.
03/05/2004 (4:19 am)
Jeff Tunnell,I really like that idea. Would answer those question quite nicely as well as give an indi an idea if there game should run "as good" on a potentially targeted PC that they don't have access to everyday in dev. (Like me and the mac)
As I put my indi hat on I really don't care if windows is faster or slower than a mac or Linux. I just care that I can get my game to reach as many buyers as possible. This info would help me makes some of those design idea vs machine limitation. (The if I implement that I will have to ensure it still runs well on X)
It would also be nice to see a break down of what games sell on what OS's of games that can run on all three. Would answer the other question of "if I create a game that is like this, where would I expect my customer base mostly be?" Might drive you to move to that OS and forget support for all others or might drive you to make that OS the developement one so you KNOW it will work and work well on it.
But all that is leading to a different topic than this thread.
#29
Conclusion: Either is a great platform for Torque, performance will vary based on drivers.
The following system was used in this test.
P4 2.2ghz with 1 gigabyte of DDR333 ram, agp x4, Nvidia GeForce 4 Ti 4200, 120gb WD 7200 rpm HD, built-in C-Media sound.
=================FPS:MAX/Min
Linux(Mepis)==========160/80
Linux(Mandrake 9.2)====92/34
WinXP Pro(53.03)======70/20
WinXP Pro(56.57)======130/63
For the Max-FPS I went down by the tree in the village and looked straight up into the sky. The min I just looked around and recorded the lowest fps I could achieve.
All tests were ran in 1024x768 with 32 bit depth
The current Nvidia drivers for windows seem terrible. There was a huge increase in performance when I switched from 53.02 to 56.57. I had no access to the beta drivers for linux So I could not compare for a difference in driver performance for that platform.
A few things for those wanting to conduct your own tests. I ran into some problems after installing linux. The MBR(master boot record) was changed by linux and windows would not install onto the drive. I had to remove the drive from the system, attach it to another windows system and partition it, then reinstall back into test system. I was told you could Fdisk /mbr the drive, but from the WinXP cd that wasn't an option. It does offer a Fixmbr and Fixboot option, neither worked.
For those who don't know how to display FPS within torque, bring down the console window by pushing ', then type metrics(fps);
Close the console window by pushing ' again.
These tests weren't done to say Linux or Windows is better then the other, but to help you as a developer on Torque see what kind of performance you can expect between platforms. I invite others to post results of their systems as well. Or you can argue how its meaningless and not scientific, coJARRODugh. Either way, I found it interesting. :)
-Jason
edit: spacing on table
03/07/2004 (10:56 pm)
Ok, I've had some time(and troubles I might add with MBR) to do some testing between Linux and Windows. In short with the "Latest" drivers for both systems, Linux faired somewhat better. With beta drivers for windows, Windows came close to the same performance.Conclusion: Either is a great platform for Torque, performance will vary based on drivers.
The following system was used in this test.
P4 2.2ghz with 1 gigabyte of DDR333 ram, agp x4, Nvidia GeForce 4 Ti 4200, 120gb WD 7200 rpm HD, built-in C-Media sound.
=================FPS:MAX/Min
Linux(Mepis)==========160/80
Linux(Mandrake 9.2)====92/34
WinXP Pro(53.03)======70/20
WinXP Pro(56.57)======130/63
For the Max-FPS I went down by the tree in the village and looked straight up into the sky. The min I just looked around and recorded the lowest fps I could achieve.
All tests were ran in 1024x768 with 32 bit depth
The current Nvidia drivers for windows seem terrible. There was a huge increase in performance when I switched from 53.02 to 56.57. I had no access to the beta drivers for linux So I could not compare for a difference in driver performance for that platform.
A few things for those wanting to conduct your own tests. I ran into some problems after installing linux. The MBR(master boot record) was changed by linux and windows would not install onto the drive. I had to remove the drive from the system, attach it to another windows system and partition it, then reinstall back into test system. I was told you could Fdisk /mbr the drive, but from the WinXP cd that wasn't an option. It does offer a Fixmbr and Fixboot option, neither worked.
For those who don't know how to display FPS within torque, bring down the console window by pushing ', then type metrics(fps);
Close the console window by pushing ' again.
These tests weren't done to say Linux or Windows is better then the other, but to help you as a developer on Torque see what kind of performance you can expect between platforms. I invite others to post results of their systems as well. Or you can argue how its meaningless and not scientific, coJARRODugh. Either way, I found it interesting. :)
-Jason
edit: spacing on table
#30
That is what you critized for, is the trying to claim one OS was superior over the other, now you seem to understand that apples to monkeys distinction.
Here is a hint, wrap the table with code tags, it will then render in a fixed width font.
as such
03/08/2004 (9:09 am)
Here you are compairing nVidia drivers for a SPECIFIC card, on a specific machine and Operating System(s), and not claiming that one OS is better than another because of your limited test, which is NOT the same thing you initially claimed, and got called out on, not just by me, but other people as well.That is what you critized for, is the trying to claim one OS was superior over the other, now you seem to understand that apples to monkeys distinction.
Here is a hint, wrap the table with code tags, it will then render in a fixed width font.
as such
S:MAX/Min Linux(Mepis) 160/80 Linux(Mandrake 9.2) 92/34 WinXP Pro(53.03) 70/20 WinXP Pro(56.57) 130/63
#31
I am not going to tell Linux is better. I will tell you that it is better for ME.
Now, it has been tested and found to be true by many, many, many network administrators that you can get MUCH greater performance with a Linux based server over a Windows based server. The networking performance is extremely good in a Linux machine. Another item to point out is that *nix in general is a more stable development platform.
I have been using Linux for at least 3+ years now. I have not regretted it at all. I have learned so much about *nix that it has made me extremely valuable at work. I also use Windows machines at work every day for general tasks and deploying systems. I just had XP installed on my work machine a couple months ago. In general it is okay. I would not buy it for myself.
This is a personal decision to use Linux. Torque compiles fine and there are some issues with the editor, but other than that it runs great.
If you want to understand wby some Linux users are of the opinion that Windows is inferior then consider this. In software development there is a thing called the citadel. This is where code is developed in a very controlled environment. The reasoning behind this is to make sure there are just the right amount of cooks stirring the pot. In opposition to this there is a thing called the bazaar. Get as many cooks stirring the pot as you can. One side of the pot may taste different (or horrible) than the other. However, eventually the best tasting stuff rises to the top. The idea behind the bazaar is that a "bug" will be obvious to someone. When you consider that Microsoft has a limited number of developers, while Linux is literally gaining a limitless supply of developers, then it is just a matter of time before the "quality" of Linux software dwarfs that of any commercial OS.
Sorry for the long post.
Thanks,
Frank
03/12/2004 (5:56 pm)
Ahem,I am not going to tell Linux is better. I will tell you that it is better for ME.
Now, it has been tested and found to be true by many, many, many network administrators that you can get MUCH greater performance with a Linux based server over a Windows based server. The networking performance is extremely good in a Linux machine. Another item to point out is that *nix in general is a more stable development platform.
I have been using Linux for at least 3+ years now. I have not regretted it at all. I have learned so much about *nix that it has made me extremely valuable at work. I also use Windows machines at work every day for general tasks and deploying systems. I just had XP installed on my work machine a couple months ago. In general it is okay. I would not buy it for myself.
This is a personal decision to use Linux. Torque compiles fine and there are some issues with the editor, but other than that it runs great.
If you want to understand wby some Linux users are of the opinion that Windows is inferior then consider this. In software development there is a thing called the citadel. This is where code is developed in a very controlled environment. The reasoning behind this is to make sure there are just the right amount of cooks stirring the pot. In opposition to this there is a thing called the bazaar. Get as many cooks stirring the pot as you can. One side of the pot may taste different (or horrible) than the other. However, eventually the best tasting stuff rises to the top. The idea behind the bazaar is that a "bug" will be obvious to someone. When you consider that Microsoft has a limited number of developers, while Linux is literally gaining a limitless supply of developers, then it is just a matter of time before the "quality" of Linux software dwarfs that of any commercial OS.
Sorry for the long post.
Thanks,
Frank
#32
But for some reason I feel Jarrod will argue, its what he does.
03/12/2004 (7:13 pm)
Right on Frank!But for some reason I feel Jarrod will argue, its what he does.
#33
03/14/2004 (11:46 am)
You guys are the ones arguing, I stated a FACT. Nowhere did I argue anything abuot windows vs linux, I stated a FACT that your "comparison" was invalid, others agreed, you argued Jason. You re-canted with your last post that did not try to conclude that windows was faster but just compared video driver versions on each OS, and Frank is talking about Linux on servers something is not even close to the topic of playing games. Bringing unrelated arguments as rebuttles is just argumentative and trolling.
#34
03/14/2004 (12:38 pm)
Jarrod always has something to say, take over the world already with your endless knowledge! 
#35
While the original paper by Eric S. Raymond has some nice intellectual musings on the topic, they have been misrepresented, as you just did : the "time" you mention could be infinite, and no one has shown with any degree of scientific or even empirical evidence that having more eyeballs is better. To the contrary : the worst way to salvage a project in troubles is to add yet more eyeballs. This is not only true of software but of most production processes, be them industrial, artistical, etc.
The complexity of management grows with numbers, exponentially.
This is why the Linux kernel is developed by a very small group of engineers, etc. And the same for every succesful Open Source project.
Fact : one of the most secure BSDs out there, OpenBSD does NOT trust the auditing of its source code just to anyone who can get it, but to a very limited core of engineers.
And I could go on with just about every succesful Open Source initiative...
Again, Open Source development does not mean total chaos.
You've obviously never talked with people who work in big kitchens (I have, and have worked in a few of them) : the more cooks, the lower the quality of the food.
And, as this post is totally off-topic for this thread, Frank's also was.
Have a nice day
03/14/2004 (12:50 pm)
Yeah, the comment on the cathedral vs bazaar (not citadel, but cathedral is the original image Frank ;))While the original paper by Eric S. Raymond has some nice intellectual musings on the topic, they have been misrepresented, as you just did : the "time" you mention could be infinite, and no one has shown with any degree of scientific or even empirical evidence that having more eyeballs is better. To the contrary : the worst way to salvage a project in troubles is to add yet more eyeballs. This is not only true of software but of most production processes, be them industrial, artistical, etc.
The complexity of management grows with numbers, exponentially.
This is why the Linux kernel is developed by a very small group of engineers, etc. And the same for every succesful Open Source project.
Fact : one of the most secure BSDs out there, OpenBSD does NOT trust the auditing of its source code just to anyone who can get it, but to a very limited core of engineers.
And I could go on with just about every succesful Open Source initiative...
Again, Open Source development does not mean total chaos.
You've obviously never talked with people who work in big kitchens (I have, and have worked in a few of them) : the more cooks, the lower the quality of the food.
And, as this post is totally off-topic for this thread, Frank's also was.
Have a nice day
#36
03/14/2004 (2:55 pm)
And for some reason I think Jason will just continute to make personal attacks it is whate he does.
#37
>While the original paper by Eric S. Raymond has some nice >intellectual musings on the topic, they have been >misrepresented, as you just did : the "time" you mention >could be infinite, and no one has shown with any degree of >scientific or even empirical evidence that having more >eyeballs is better. To the contrary : the worst way to
Where is the research for defending the cathedral (yes, I did forget the correct term) model of development?
>salvage a project in troubles is to add yet more eyeballs.
This would most likely be the case because of schedule. However, it does depend on what you mean by salvage.
>This is not only true of software but of most production >processes, be them industrial, artistical, etc.
>The complexity of management grows with numbers,
>exponentially.
You are mixing apples and oranges. The bazaar by its very nature does not require the traditional forms of management to get a project done.
>This is why the Linux kernel is developed by a very small >group of engineers, etc. And the same for every succesful >Open Source project.
Linux does have a hierarchy of software engineers / hackers. However, they all started out as Linus did, at the bottom. If you have a good modification for the Linux kernel you submit it. If it is good it is included. Sometimes it takes a while to prove a change will be good. I myself have submitted a kernel driver to the Linux project. It was not included. One reason is because it is for a device that is not sold and is very scarce. The other is because I never really pursued it. Had the device been in production I would have pushed more. The point is if someone has a better idea for doing something they can get it included. The kernel development of Linux is not elitest.
>Fact : one of the most secure BSDs out there, OpenBSD does >NOT trust the auditing of its source code just to anyone >who can get it, but to a very limited core of engineers.
I would imagine their development model is similar, but perhaps more restrictive. The source is available. Anyone can audit it. They would be fools not to allow submissions. Yes, you still need control of distributions. For a very large piece of code, like the kernel, you must have reviews for stability. All open source projects have this kind of hierarchy. It is just scaled to size.
>And I could go on with just about every succesful Open >Source initiative...
Really? Do go on...
>Again, Open Source development does not mean total chaos.
Neither does the bazaar.
>You've obviously never talked with people who work in big >kitchens (I have, and have worked in a few of them) : the >more cooks, the lower the quality of the food.
The food analogy is to provide a visual picture on what the idea behind the concept was. You being a professional kitchen man I will defer to your experience on this issue. I would not dare to imply that this is the proper way to make a dish. So I will just have to come up with some other analogy.
>And, as this post is totally off-topic for this thread, >Frank's also was.
Since when is flame-bait off topic in an off topic flame-bait thread anyway?
>Have a nice day
You to.
Take this piece of code:
char *h = "Hello";
char *w = "World";
printf("%s %s\n", *h, *w);
Can you find the bug(s)?
Lets see how many chaotic/bazaar developers it takes to find the problem. If you don't know please respond with "Bug? I don't know." in your post.
Thanks,
Frank
BTW - The "cathedral vs the bazaar" is not some "idea" of how to create code. It was an attempt to describe "how" the code is being developed. The industry is undergoing a paradigm shift. The question large code developers are asking is, "How do we deal with code when it gets so huge?". The "bazaar" may have some of the answers.
03/14/2004 (6:01 pm)
>Yeah, the comment on the cathedral vs bazaar (not citadel, >but cathedral is the original image Frank ;))>While the original paper by Eric S. Raymond has some nice >intellectual musings on the topic, they have been >misrepresented, as you just did : the "time" you mention >could be infinite, and no one has shown with any degree of >scientific or even empirical evidence that having more >eyeballs is better. To the contrary : the worst way to
Where is the research for defending the cathedral (yes, I did forget the correct term) model of development?
>salvage a project in troubles is to add yet more eyeballs.
This would most likely be the case because of schedule. However, it does depend on what you mean by salvage.
>This is not only true of software but of most production >processes, be them industrial, artistical, etc.
>The complexity of management grows with numbers,
>exponentially.
You are mixing apples and oranges. The bazaar by its very nature does not require the traditional forms of management to get a project done.
>This is why the Linux kernel is developed by a very small >group of engineers, etc. And the same for every succesful >Open Source project.
Linux does have a hierarchy of software engineers / hackers. However, they all started out as Linus did, at the bottom. If you have a good modification for the Linux kernel you submit it. If it is good it is included. Sometimes it takes a while to prove a change will be good. I myself have submitted a kernel driver to the Linux project. It was not included. One reason is because it is for a device that is not sold and is very scarce. The other is because I never really pursued it. Had the device been in production I would have pushed more. The point is if someone has a better idea for doing something they can get it included. The kernel development of Linux is not elitest.
>Fact : one of the most secure BSDs out there, OpenBSD does >NOT trust the auditing of its source code just to anyone >who can get it, but to a very limited core of engineers.
I would imagine their development model is similar, but perhaps more restrictive. The source is available. Anyone can audit it. They would be fools not to allow submissions. Yes, you still need control of distributions. For a very large piece of code, like the kernel, you must have reviews for stability. All open source projects have this kind of hierarchy. It is just scaled to size.
>And I could go on with just about every succesful Open >Source initiative...
Really? Do go on...
>Again, Open Source development does not mean total chaos.
Neither does the bazaar.
>You've obviously never talked with people who work in big >kitchens (I have, and have worked in a few of them) : the >more cooks, the lower the quality of the food.
The food analogy is to provide a visual picture on what the idea behind the concept was. You being a professional kitchen man I will defer to your experience on this issue. I would not dare to imply that this is the proper way to make a dish. So I will just have to come up with some other analogy.
>And, as this post is totally off-topic for this thread, >Frank's also was.
Since when is flame-bait off topic in an off topic flame-bait thread anyway?
>Have a nice day
You to.
Take this piece of code:
char *h = "Hello";
char *w = "World";
printf("%s %s\n", *h, *w);
Can you find the bug(s)?
Lets see how many chaotic/bazaar developers it takes to find the problem. If you don't know please respond with "Bug? I don't know." in your post.
Thanks,
Frank
BTW - The "cathedral vs the bazaar" is not some "idea" of how to create code. It was an attempt to describe "how" the code is being developed. The industry is undergoing a paradigm shift. The question large code developers are asking is, "How do we deal with code when it gets so huge?". The "bazaar" may have some of the answers.
#38
This thread has really gone downhill in the last ten posts. This was an interesting discussion about the range of performances one could see on different bits of hardware. Now it's just a pissing contest. :-/
C'mon guys.
03/14/2004 (9:16 pm)
Yeah, wow.This thread has really gone downhill in the last ten posts. This was an interesting discussion about the range of performances one could see on different bits of hardware. Now it's just a pissing contest. :-/
C'mon guys.
#39
Sorry for the stains... ;)
03/15/2004 (7:34 am)
Btw, I'm a big proponent of Open Source, I was just trying to debunk a myth, albeit in an off topic way. (the myth being more eyeballs to work on project == teh g00d ;))Sorry for the stains... ;)
#40
This is a good discussion in that respect, as it brings to light the many issues including passion of a user for an OS. It probably would not be good juju to tell a customer it would work better on a Mac when they own a PC for instance. It would also not be good to ignore issues on one platform because it is assumed to be the fault of the platform. Even if it is, then it should be worked around if possible. Despite the degradation of the conversation, a lot can be learned from this exchange.
Thanks and I apologize if I insulted anyone,
Frank
03/15/2004 (8:34 pm)
I do find it interesting that the there can be such a variance in performance. It really shows that we have our work cut out for us when trying to solve performance issues on multiple platforms and OSes. Is it the hardware, software, driver, or a bug in our game? This is a good discussion in that respect, as it brings to light the many issues including passion of a user for an OS. It probably would not be good juju to tell a customer it would work better on a Mac when they own a PC for instance. It would also not be good to ignore issues on one platform because it is assumed to be the fault of the platform. Even if it is, then it should be worked around if possible. Despite the degradation of the conversation, a lot can be learned from this exchange.
Thanks and I apologize if I insulted anyone,
Frank
Torque 3D Owner Robert Blanchet Jr.
Also, 111 fps is extremley low for a 64bit processor. I have an AMD 64 3000+ with 1gig PC3200 Ram and a Geforce 4 Ti4600 and I get no less then 175 fps in Torque, and when I mean I get no less then 170 I mean it. Highest I get in Torque is 233 fps just sitting there looking at the sky.
For nvidia hardware on windows make sure you are using the 55 or greater drivers, there are problems with any below that with performance.