This is the fourth time www.atheos.cx (running AtheOS) hit the front page of/. and it still have not crashed a single time due
to the load it generates. This time (with a new link and much work spendt on the TCP/IP
stack lately) it even seem to be quite easily accessible. It choked quite badly in the
beginning but after moving images and static downloads to a spearate HTTP server
(on the same machine) it now seems to be quite responsive.
I have 1GB RAM in the machine I type this on (running AtheOS 24/7) and it does not boot any (noticable) slower now than when I had 128MB. I don't have BeOS installed so I have no first-hand experience on how fast it boots but I can't figure out why it should slow down. Maybe they do some very expensive initializing of page-list's or something in that direction who slows it down when you add "to much" RAM. AtheOS might use a few mS longer to initialize the "memory manager" but it doesn't take long enough to notice.
The memory manager in BeOS is quite broken though. It have a fixed-size disk cache (configurable before boot but with a very limited maximum size AFAIK) that takes away much of the benefits you normally get's by adding more RAM. AtheOS have a dynamic cache and will utilize all unused memory as disk cache so adding more RAM will greatly speed up things touching a lot of disk-data (like when compiling large projects). On BeOS you will just end up with more free RAM.
It was a driver patch (to the PS2 mouse driver) not libatheos. I have always said that I will accept new drivers and driver patches as long as they are of a reasonable quality.
Can you please show me one single statement anywhere where I threaten to close-source AtheOS? You couldn't find it? Ok, then how did you conclude that I have often threatened to make AtheOS closed source? I have never threatened to do so and I will not threaten to do so in the future.
I have said that I'm might going to change the lisense to a more liberal one in case the GPL would cause problems for non-GPL drivers, apps, or other "third-party" components somehow. How you managed to twist that into "often threathened to close-source" is beyond me.
Hehe, this is the third time the server has been slashdotted and each time the
slashdot hord have triumfated happily when the server have appeared dead
after a few minutes:)
Well the fact is that the server has never crashed due to the slashdot effect. It is
always bogged down by the heavy load though and being hooked to the net through a 2Mb
link there is no way it can handle the incredible load created by such a slashdot
article.
The reason it lost it's uptime today was that I took it down to add another 128MB of
memory to be able to increase the maximum number of
simulatnous HTTP connections allowed (it is mentioned on the news-page but as you probably
have noticed having anything on the atheos.cx right now doesn't mean that it is accessible
to everybody:)
PS: This is my first/. posting with the AtheOS web-browser so please be kind with
me if it end's up all screwed:)
Re:dont want simulated, I want the real thing!
on
AtheOS Interview
·
· Score: 2
Well, then you are not looking for a desktop OS. A Ferarry doesn't pull trees all that well so it must be inferiour to the tracktor. Right?
Re:My take on AtheOS... I want my shell, dammit!
on
AtheOS Interview
·
· Score: 1
>1) I wouldnt use it because I doesnt have console. Not only does AtheOS not have a console, the developers seem to actively _SCORN_ the notion of a console (text) shell. Foo on them!
It have an XTerm like terminal emulator that allow you to run BASH or any other shell you might desire.
>2) Read the information about the way the software is structured. It's in stark contrast to Linux and Linus' vision. Read it...it has this in the kernel, that in the kernel....kernel kernel kernel. They rave on about all the nifty kernel features. (most notably, video drivers and user interface elements, and most of what seems to be the user-space part of pthread libraries, all jammed in the kernel)
I don't know what website you have read but it ceratinly ain't mine (www.atheos.cx). Only the essentials are in the kernel. Everything else is in user-space including the GUI and it's drivers.
From www.atheos.cx: "The GUI is server/client like X11 but communicate through the native messaging system"
Graphics drivers are loaded by the user-space GUI server, not by the kernel.
How did you managed to mix pthreds into all this?
>As I say, if you read Linus' mails in the kernel development list, that'd make steam come out of his ears. Linus practices a great deal of restraint regarding kernel patches. Linus clearly (and IMHO, rightly) believes that video drivers dont belong there -- that and OS must not be held hostage to a video card. Linus has rejected all sorts of "nifty features" from entering the kernel tree, so that stability remains paramount.
Like the frame-buffer device (and all it's graphics kernel-driver)? Someone should tell Linus that a bunch of graphics drivers has been smuggeled into the Linux kernel.
>As I say, it's a stark contrast. I'd be willing to play around with it...but I doubt I'd use it or make it my primary OS. I'm just not a "graphical computing uber alles" type of guy. I want my shell, dammit! I wanna be able to ditch all the chrome!
Sure, no problem. But if you hate the GUI so much, why did you continue to read past the "Desktop OS" part?
Yes. I have been developing AtheOS solely under AtheOS for about 3 years now.
>Is anyone working on a GUI RAD tool for AtheOS (something in the spirit of KDevelop or Glade)?
No.
>Are you trying to gather the momentum among application developers now or is it too early?
It's a bit early. I have never ever anounced AtheOS anywhere myself since many important "desktop features" and other things that I whould like to have firmly defined from the beginning are still missing.
>Do you have a Component Model yet? If not are you planning to add one soon? A CM would definitely help keep the look and feel consistent and help avoid code duplication. Two things that marred Linux's acceptance as a desktop OS.
Nothing like COM or CORBA but AtheOS have an object oriented design and it is component based in the sence that you can build abstract components that communicate over the flexible "builtin" event system.
It seems like there is a lot of confusion about the AtheOS connection to the Intel CPU. AtheOS is *not* designed around the Intel architecture. What I said on the front page is: "AtheOS is a free operating system for the Intel architecture". This does not mean that it was designed around the Intel CPU! 99% of AtheOS is highly portable C/C++. It use a flat paged memory-model, no Intel specific segmented memory model. I only use assambler when there is no way to solve the problem in C (which is VERY rare). There is very little that tie AtheOS to the Intel platform. The main reason it only runs on Intel CPUs today is that I don't own any other hardware that is worth writing an OS for. I have had portability in my mind when implementing AtheOS. I have not been very fixated on it, but I can't think of one design decision that should be a major bottleneck when it comes to porting it. I just dont feel like fork out the required $$$ to get a multi-CPU Alpha/SGI/MAC/Whatever machine, and then learn a new architecture from scratch and port AtheOS. For now my dual-celeron do all I want. When a new CPU comes around that I whould like to use, and can afford, I am pritty sure I will port AtheOS to it.
Hope this clearify some things. BTW I have no problem seeing why the uni-architecture discussion started, after all the story, both at kuro5hin and here at slashdot implied that AtheOS was designed tightly around the Intel architecture. Believe me it's not, or when the server stop bouncing around in my office, take a look at the sources:)
>Anybody else noticed that all the titles in the shots say "AltOS"?:)
Ups..:) Some of the screenshots is very old. From when AtheOS indead was named AltOS. The reason for the name-change is that AltOS is tradmarked. I spent a lot of time renaming stuff, but I totally forgot the screenshots:(
Re:I Suspect they Ripped off a BSD
on
AtheOS
·
· Score: 1
> Writting a complete, fully-featured, modern OS > with a modern look & feel without anyone finding > out about it. This would take quite a few coders > a fairly signifigant amount of time.
Or one guy about 4 years.:) The reason not anyone found it is simple: Until about 6 months ago there was no web-server, and no references to it what-so-ever, other than on my own PC. Even after I put up the server (mostly for the fun of running a server on AtheOS)I did not anounce it anywhere. The reason for not anouncing it is that I just don't think it is finnished enough yet. I do this for fun, and I whould like to finnish more of the fun tasks before giving them away:) I believe that to many cook's at an early state will indead end with a lot of mess. I know I will propably get flamed for that statement, but after all I wrote the OS, I want to get it the way I want it. If you don't like it, there is (as mentioned many times already) a lot of OS projects out there. Find one you like.
And about the BSD ripoff: There is no BSD code in AtheOS. The OS is written from scratch. The NE2000 driver is ported from Linux, and kmalloc() is ripped from the linux kernel. I took the string functions (memcpy(),strcmp(),etc, etc) from Linux, and maybe a few other small general utility functions and stuffed into the kernel. The rest of the kernel, and the highlevel (user-space) API's, and the appserver is written entirely from scratch. Most of the application is from the various GNU packages though, I just ported them.
Remember, you get what you pay for, and you got it for free:)
>This also means that you get all BIOS-related problems like 1024-cylinder-stuff. 8-(
No, not unless you have an ancient BIOS that don't support the "new" 64-bit IO functions. All the disks I am running AtheOS from is >4GB, no problem.
Re:ERmmm... make that 'kuro5hined' before ...
on
AtheOS
·
· Score: 1
>Er. Sorry about that.:-)
No hard feelings:)
>Curiosity: How many hits, or what percentage, roughly, seemed to come from kuro5hin?
I'm not sure at this stage. I have stopped the cron-jobs updating the stats to reduce the load as much as possible.
>I have no idea if we're even capable of lashdotting someone yet-- the readership is not all *that* big.
It is not *that* bad, but still it adds to the load. Last time I checked the server-stats before slashdot made that impossible (about 9 hour after the kuro5hin article was posted) I had received 8000-9000 hits. This is 6000-7000 more than normal. Last time I checked it (a few hours ago) I received 25000+ hits per hour.
>I'd be interested to hear numbers from your side, if you've got 'em.
I can't get to the status pages since, well... someone is using my server:) I can give an update when thing's are calming down a bit.
Re:ERmmm... make that 'kuro5hined' before ...
on
AtheOS
·
· Score: 1
>Better this way than people just not being interested though:)
Sure, much more enjoyable.
>What is the server hardware out of curiosity and what's the outgoing bandwidth of the site?
The hardware is a K6-200 with 256MB RAM, and the bandwith is 2Mb/S. I just recompiled the kernel and rebooted the server to get rid of a lot of debug output that was sendt through the serial link and bugged down the server. It is now running 3-4 times faster than with the debug output enabled.
>I'm looking forward to looking at the site.. maybe by Sunday...
:)
Re:ERmmm... make that 'kuro5hined' before ...
on
AtheOS
·
· Score: 2
What have I done to deserve this?:)
Typical to my luck to be "slashdotted" from both the real thing, and another large news-site simultanously:(
A little update on the server: It is very overloaded by now, but it is alive:) It seems to be quite stable, since the only problem so far is that it is overloaded. It has not crashed or misbehaved in any other way. (Ohh, and yes, the server do run AtheOS)
Hehe. Yes it is definitively fake! Pasted together in PhotoShop. Even the source code in the CVS repository have traces of the airbrush on most of the glyphs:)
Seriously: I know the GUI is a bit inconsistent, but it is shaping up (Some of the screenshots are rather old).
Sorry to keep this thread off-topic. Just couldn't resist:)
This is the fourth time www.atheos.cx (running AtheOS) hit the front page of /. and it still have not crashed a single time due
to the load it generates. This time (with a new link and much work spendt on the TCP/IP
stack lately) it even seem to be quite easily accessible. It choked quite badly in the
beginning but after moving images and static downloads to a spearate HTTP server
(on the same machine) it now seems to be quite responsive.
I have 1GB RAM in the machine I type this on (running AtheOS 24/7) and it does not boot any (noticable) slower now than when I had 128MB. I don't have BeOS installed so I have no first-hand experience on how fast it boots but I can't figure out why it should slow down. Maybe they do some very expensive initializing of page-list's or something in that direction who slows it down when you add "to much" RAM. AtheOS might use a few mS longer to initialize the "memory manager" but it doesn't take long enough to notice.
The memory manager in BeOS is quite broken though. It have a fixed-size disk cache (configurable before boot but with a very limited maximum size AFAIK) that takes away much of the benefits you normally get's by adding more RAM. AtheOS have a dynamic cache and will utilize all unused memory as disk cache so adding more RAM will greatly speed up things touching a lot of disk-data (like when compiling large projects). On BeOS you will just end up with more free RAM.
It was a driver patch (to the PS2 mouse driver) not libatheos. I have always said that I will accept new drivers and driver patches as long as they are of a reasonable quality.
I have said that I'm might going to change the lisense to a more liberal one in case the GPL would cause problems for non-GPL drivers, apps, or other "third-party" components somehow. How you managed to twist that into "often threathened to close-source" is beyond me.
Kurt Skauen
Well the fact is that the server has never crashed due to the slashdot effect. It is always bogged down by the heavy load though and being hooked to the net through a 2Mb link there is no way it can handle the incredible load created by such a slashdot article.
The reason it lost it's uptime today was that I took it down to add another 128MB of memory to be able to increase the maximum number of simulatnous HTTP connections allowed (it is mentioned on the news-page but as you probably have noticed having anything on the atheos.cx right now doesn't mean that it is accessible to everybody :)
PS: This is my first /. posting with the AtheOS web-browser so please be kind with
me if it end's up all screwed :)
Well, then you are not looking for a desktop OS. A Ferarry doesn't pull trees all that well so it must be inferiour to the tracktor. Right?
It have an XTerm like terminal emulator that allow you to run BASH or any other shell you might desire.
>2) Read the information about the way the software is structured. It's in stark contrast to Linux and Linus' vision. Read it...it has this in the kernel, that in the kernel....kernel kernel kernel. They rave on about all the nifty kernel features. (most notably, video drivers and user interface elements, and most of what seems to be the user-space part of pthread libraries, all jammed in the kernel)
I don't know what website you have read but it ceratinly ain't mine (www.atheos.cx). Only the essentials are in the kernel. Everything else is in user-space including the GUI and it's drivers.
From www.atheos.cx: "The GUI is server/client like X11 but communicate through the native messaging system"
Graphics drivers are loaded by the user-space GUI server, not by the kernel.
How did you managed to mix pthreds into all this?
>As I say, if you read Linus' mails in the kernel development list, that'd make steam come out of his ears. Linus practices a great deal of restraint regarding kernel patches. Linus clearly (and IMHO, rightly) believes that video drivers dont belong there -- that and OS must not be held hostage to a video card. Linus has rejected all sorts of "nifty features" from entering the kernel tree, so that stability remains paramount.
Like the frame-buffer device (and all it's graphics kernel-driver)? Someone should tell Linus that a bunch of graphics drivers has been smuggeled into the Linux kernel.
>As I say, it's a stark contrast. I'd be willing to play around with it...but I doubt I'd use it or make it my primary OS. I'm just not a "graphical computing uber alles" type of guy. I want my shell, dammit! I wanna be able to ditch all the chrome!
Sure, no problem. But if you hate the GUI so much, why did you continue to read past the "Desktop OS" part?
Yes. I have been developing AtheOS solely under AtheOS for about 3 years now.
>Is anyone working on a GUI RAD tool for AtheOS (something in the spirit of KDevelop or Glade)?
No.
>Are you trying to gather the momentum among application developers now or is it too early?
It's a bit early. I have never ever anounced AtheOS anywhere myself since many important "desktop features" and other things that I whould like to have firmly defined from the beginning are still missing.
>Do you have a Component Model yet? If not are you planning to add one soon? A CM would definitely help keep the look and feel consistent and help avoid code duplication. Two things that marred Linux's acceptance as a desktop OS.
Nothing like COM or CORBA but AtheOS have an object oriented design and it is component based in the sence that you can build abstract components that communicate over the flexible "builtin" event system.
It's not down. It is a bit loaded though :)
The server link is only 2Mb/s so it tend to be a bit hard to access when slashdotted.
Hope this clearify some things. BTW I have no problem seeing why the uni-architecture discussion started, after all the story, both at kuro5hin and here at slashdot implied that AtheOS was designed tightly around the Intel architecture. Believe me it's not, or when the server stop bouncing around in my office, take a look at the sources :)
Ups.. :) :(
Some of the screenshots is very old. From when AtheOS indead was named AltOS. The reason for the name-change is that AltOS is tradmarked. I spent a lot of time renaming stuff, but I totally forgot the screenshots
> with a modern look & feel without anyone finding
> out about it. This would take quite a few coders
> a fairly signifigant amount of time.
Or one guy about 4 years. :) :) I believe that to many cook's at an early state will indead end with a lot of mess. I know I will propably get flamed for that statement, but after all I wrote the OS, I want to get it the way I want it. If you don't like it, there is (as mentioned many times already) a lot of OS projects out there. Find one you like.
The reason not anyone found it is simple: Until about 6 months ago there was no web-server, and no references to it what-so-ever, other than on my own PC. Even after I put up the server (mostly for the fun of running a server on AtheOS)I did not anounce it anywhere. The reason for not anouncing it is that I just don't think it is finnished enough yet. I do this for fun, and I whould like to finnish more of the fun tasks before giving them away
And about the BSD ripoff: There is no BSD code in AtheOS. The OS is written from scratch. The NE2000 driver is ported from Linux, and kmalloc() is ripped from the linux kernel. I took the string functions (memcpy(),strcmp(),etc, etc) from Linux, and maybe a few other small general utility functions and stuffed into the kernel. The rest of the kernel, and the highlevel (user-space) API's, and the appserver is written entirely from scratch. Most of the application is from the various GNU packages though, I just ported them.
Remember, you get what you pay for, and you got it for free :)
No, not unless you have an ancient BIOS that don't support the "new" 64-bit IO functions. All the disks I am running AtheOS from is >4GB, no problem.
No hard feelings :)
>Curiosity: How many hits, or what percentage, roughly, seemed to come from kuro5hin?
I'm not sure at this stage. I have stopped the cron-jobs updating the stats to reduce the load as much as possible.
>I have no idea if we're even capable of lashdotting someone yet-- the readership is not all *that* big.
It is not *that* bad, but still it adds to the load. Last time I checked the server-stats before slashdot made that impossible (about 9 hour after the kuro5hin article was posted) I had received 8000-9000 hits. This is 6000-7000 more than normal. Last time I checked it (a few hours ago) I received 25000+ hits per hour.
>I'd be interested to hear numbers from your side, if you've got 'em.
I can't get to the status pages since, well... someone is using my server :)
I can give an update when thing's are calming down a bit.
Sure, much more enjoyable.
>What is the server hardware out of curiosity and what's the outgoing bandwidth of the site?
The hardware is a K6-200 with 256MB RAM, and the bandwith is 2Mb/S. I just recompiled the kernel and rebooted the server to get rid of a lot of debug output that was sendt through the serial link and bugged down the server. It is now running 3-4 times faster than with the debug output enabled.
>I'm looking forward to looking at the site.. maybe by Sunday...
Typical to my luck to be "slashdotted" from both the real thing, and another large news-site simultanously :(
A little update on the server: :)
It is very overloaded by now, but it is alive
It seems to be quite stable, since the only problem so far is that it is overloaded. It has not crashed or misbehaved in any other way.
(Ohh, and yes, the server do run AtheOS)
Hehe. Yes it is definitively fake! Pasted together in PhotoShop. Even the source code in the CVS repository have traces of the airbrush on most of the glyphs :)
:)
Seriously: I know the GUI is a bit inconsistent, but it is shaping up (Some of the screenshots are rather old).
Sorry to keep this thread off-topic. Just couldn't resist