Disk is cheap?
Quoted 3 lines
Everyone who says, disk is cheap, ought to donate me one. Everyone who says, memory is cheap, has to send me some.
:-)
Perhaps a more convincing argument may be that in embedded devices, disk as well as memory and CPU power are _not_ cheap.
The more resources Linux requires, the less are it's chances of being accepted as a viable alternative in embedded systems.
Quoted 3 lines
I'm still stuck with a P-133, 56 MB RAM (60-70 ns, some EDO, some FPM) and not only Linux but also W2K on a 2.1 and a 0.8 GB HDD.
That would be _a_ _lot_ for an embedded system!
----------------------------------------------------------------
Robert Kaiser email: [email protected]
SYSGO RTS GmbH
Am Pfaffenstein 14 phone: (49) 6136 9948-762
D-55270 Klein-Winternheim / Germany fax: (49) 6136 9948-10
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/
I built a embedded dvd/cdda/mp3 player based on linux, using a p200mmx with 24mb with a bus of 75mhx, but it still takes about 20 seconds to boot, I think that an embedded device (for home use) should boot in less than 5 seconds, how could be possible with a slow p133? (I've also tried a p133 on 66mhz of bus and it takes almost 35 seconds to boot)
However, first or last old hardware will finish, and who wants to build an embedded device should use high cost embedded hardware (high cost for me).
P.
Quoted 18 lines · Robert Kaiser
Quoted 3 lines
Everyone who says, disk is cheap, ought to donate me one. Everyone who says, memory is cheap, has to send me some.
:-)
Perhaps a more convincing argument may be that in embedded devices, disk as well as memory and CPU power are _not_ cheap.
The more resources Linux requires, the less are it's chances of being accepted as a viable alternative in embedded systems.
Quoted 3 lines
I'm still stuck with a P-133, 56 MB RAM (60-70 ns, some EDO, some FPM) and not only Linux but also W2K on a 2.1 and a 0.8 GB HDD.
That would be _a_ _lot_ for an embedded system!
Quoted 10 lines
----------------------------------------------------------------
Robert Kaiser email: [email protected]
SYSGO RTS GmbH
Am Pfaffenstein 14 phone: (49) 6136 9948-762
D-55270 Klein-Winternheim / Germany fax: (49) 6136 9948-10
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/
---------------------------------------------------------
Patrizio Bruno
DADA spa / Ed-IT Development Staff
Borgo degli Albizi 37/r
50122 Firenze
Italy
tel +39 05520351
fax +39 0552478143
PGP PublicKey available at: http://www.keyserver.net/en/
---------------------------------------------------------
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/
In article <[email protected]>,
Quoted 5 lines · [email protected] (Patrizio Bruno) writes:
I built a embedded dvd/cdda/mp3 player based on linux, using a p200mmx with 24mb with a bus of 75mhx, but it still takes about 20 seconds to boot, I think that an embedded device (for home use) should boot in less than 5 seconds, how could be possible with a slow p133? (I've also tried a p133 on 66mhz of bus and it takes almost 35 seconds to boot)
Usually most of the startup time is spent by the BIOS doing extensive self-test stuff and for firing up services (http, inetd, sendmail, ...) that many embedded systems have little use for.
I have a 25MHz 386EX (~2.2 Bogomips) here that boots Linux out of ROM in roughly 30 seconds. Most of _that_ time however is spent decompressing the kernel.
Quoted 2 lines
However, first or last old hardware will finish, and who wants to build an embedded device should use high cost embedded hardware (high cost for me).
Hmm, since embedded devices are usually built in large scale series, cost is quite an important factor. Neverthless, even if you leave that aside, there is the even more important question of size, power consumption and heat dissipation: Any processor that requires a fan is simply unacceptable in most embedded designs. (Plus, you don't need the speed of an 1GHz Athlon to control a washing machine...) That is the reason why relatively low-performing chips such as i386EX, AMD Elan SC[45][012]0 are very popular in embedded devices. These chips are by no means "old hardware" !
Did you know that about 95% of the worldwide microprocessor production end up in some sort of appliance/embedded system ?
I'm pretty sure these "low-end" processors will not go away as quickly as you might think. Whether they will be running Linux in the future will depend on Linux's resource requirements.
Cheers
Rob
----------------------------------------------------------------
Robert Kaiser email: rkaiser AT sysgo DOT de SYSGO RTS GmbH http://www.elinos.com Klein-Winternheim / Germany http://www.sysgo.de
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/
Quoted 13 lines · Robert Kaiser
Date: Wed, 31 Jan 2001 17:12:47 +0100
From: Robert Kaiser <[email protected]>
To: Patrizio Bruno <[email protected]>, [email protected]
Subject: Re: Disk is cheap?In article <[email protected]>,
Quoted 5 lines · [email protected] (Patrizio Bruno) writes:
I built a embedded dvd/cdda/mp3 player based on linux, using a p200mmx with 24mb with a bus of 75mhx, but it still takes about 20 seconds to boot, I think that an embedded device (for home use) should boot in less than 5 seconds, how could be possible with a slow p133? (I've also tried a p133 on 66mhz of bus and it takes almost 35 seconds to boot)
Quoted 8 lines
Usually most of the startup time is spent by the BIOS doing extensive self-test stuff and for firing up services (http, inetd, sendmail, ...) that many embedded systems have little use for.
I have a 25MHz 386EX (~2.2 Bogomips) here that boots Linux out of ROM in roughly 30 seconds. Most of _that_ time however is spent decompressing the kernel.
[...]
If someone would modify kernel to allow LZO compression then the decompression would be 3 times faster at the expense of compressed part of the image being 9..10% larger (assuming LZO1X-999/9 method used by lzop -9)
(I have done decompression speed tests on Pentium MMX 166 MHz)
Best regards,
Wojtek
--------------------
Wojtek Pilorz
[email protected]
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/
Hi!
Quoted 10 lines
Quoted 5 lines
I built a embedded dvd/cdda/mp3 player based on linux, using a p200mmx with 24mb with a bus of 75mhx, but it still takes about 20 seconds to boot, I think that an embedded device (for home use) should boot in less than 5 seconds, how could be possible with a slow p133? (I've also tried a p133 on 66mhz of bus and it takes almost 35 seconds to boot)
Usually most of the startup time is spent by the BIOS doing extensive self-test stuff and for firing up services (http, inetd, sendmail, ...) that many embedded systems have little use for.
Actually, most of that time is spent running bash/sleep 1. Startup scripts tend to be poorly designed.
Quoted 3 lines
I have a 25MHz 386EX (~2.2 Bogomips) here that boots Linux out of ROM in roughly 30 seconds. Most of _that_ time however is spent decompressing the kernel.
You might want to set up XIP and run kernel directly off the ROM...
Pavel
Signature · 6 lines
I'm [email protected]. "In my country we have almost anarchy and I don't care." Panos Katsaloulis describing me w.r.t. patents at [email protected] - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected] Please read the FAQ at http://www.tux.org/lkml/
Quoted 7 lines · On Sam, 03 Feb 2001 you wrote:
Quoted 4 lines
Usually most of the startup time is spent by the BIOS doing extensive self-test stuff and for firing up services (http, inetd, sendmail, ...) that many embedded systems have little use for.
Actually, most of that time is spent running bash/sleep 1. Startup scripts tend to be poorly designed.
Yes!
Quoted 6 lines
Quoted 3 lines
I have a 25MHz 386EX (~2.2 Bogomips) here that boots Linux out of ROM in roughly 30 seconds. Most of _that_ time however is spent decompressing the kernel.
You might want to set up XIP and run kernel directly off the ROM...
Hmm, that board has only 512KB ROM. I can fit a minimal Linux kernel and root-FS in that, but only if it's compressed. ROM, in my experience, is more expensive than RAM, so it often makes sense to save ROM space even at the expense of using a little more RAM.
But I'm curious: is there a simple procedure to set up a linux Kernel to execute from ROM ?
----------------------------------------------------------------
Robert Kaiser email: [email protected]
SYSGO RTS GmbH
Am Pfaffenstein 14 phone: (49) 6136 9948-762
D-55270 Klein-Winternheim / Germany fax: (49) 6136 9948-10
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/
Quoted 5 lines · On Sam, 03 Feb 2001 you wrote:
Quoted 2 lines
Actually, most of that time is spent running bash/sleep 1. Startup scripts tend to be poorly designed.
Yes!
I'm not so sure. I'm using RedHat 6.2, and it seems the only time a startup script calls sleep is when it gives you a chance to do interactive startup, and when you are looking for an NIS server. You could certainly remove those.
All the other calls to sleep are in the stop sections, where you want to make sure the thing died before proceeding.
But paring down the startup scripts is a good idea. For something like an embedded device, you might even want to go with a custom init, that just runs your main program.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/
Quoted 2 lines · On Die, 06 Feb 2001 you wrote:
But paring down the startup scripts is a good idea. For something like an embedded device, you might even want to go with a custom init,
Yes, I'm using busybox (see busybox.lineo.com). It's a multi-call binary that contains a simplified init, a shell and a host of other nice things.
Quoted 1 line
that just runs your main program.
Well the _very_ minimal approach is to have your main program _be_ init :-)
----------------------------------------------------------------
Robert Kaiser email: [email protected]
SYSGO RTS GmbH
Am Pfaffenstein 14 phone: (49) 6136 9948-762
D-55270 Klein-Winternheim / Germany fax: (49) 6136 9948-10
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/
Quoted 3 lines · Robert Kaiser
Quoted 2 lines · On Die, 06 Feb 2001 you wrote:
But paring down the startup scripts is a good idea. For something like an embedded device, you might even want to go with a custom init,
Plug: How about jinit (my init) ;-)
http://www.penguinpowered.com/~vii/programs/linux/jinit
http://john.snoop.dk/programs/linux/jinit
Boot script time with the supplied example scripts is 12-13 seconds to login prompt under 2.4 on my old K6-2. Jinit has integrated service stop/start functionality. Also on the page are links to other source available inits.
[...]
Signature · 5 lines
http://www.penguinpowered.com/~vii - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected] Please read the FAQ at http://www.tux.org/lkml/
Quoted 11 lines · Robert Kaiser
Perhaps a more convincing argument may be that in embedded devices, disk as well as memory and CPU power are _not_ cheap.
The more resources Linux requires, the less are it's chances of being accepted as a viable alternative in embedded systems.
Quoted 3 lines
I'm still stuck with a P-133, 56 MB RAM (60-70 ns, some EDO, some FPM) and not only Linux but also W2K on a 2.1 and a 0.8 GB HDD.
That would be _a_ _lot_ for an embedded system!
Oh this common missconception that embedded system equals small systems. There are embedded systems that outrun supercomputers without sweating, have gigs of RAM and sometimes if you look at them closly even have the names of well known big iron companies on their boards. The whole embedded term is just so weakly defined and everybody seem to have his personal definition.
Ralf
Quoted 8 lines · Ralf Baechle
Quoted 6 lines · Robert Kaiser
Perhaps a more convincing argument may be that in embedded devices, disk as well as memory and CPU power are _not_ cheap.
The more resources Linux requires, the less are it's chances of being accepted as a viable alternative in embedded systems.
Quoted 7 lines
Quoted 5 lines
Quoted 3 lines
I'm still stuck with a P-133, 56 MB RAM (60-70 ns, some EDO, some FPM) and not only Linux but also W2K on a 2.1 and a 0.8 GB HDD.
That would be _a_ _lot_ for an embedded system!
Oh this common missconception that embedded system equals small systems.
I didn't say that. I'm talking about what's common in embedded systems and it is _very_ common for an embedded system to be small both in mechanical dimension and computational horsepower. This is not a misconception but simply practical experience.
Quoted 3 lines
There are embedded systems that outrun supercomputers without sweating, have gigs of RAM and sometimes if you look at them closly even have the names of well known big iron companies on their boards.
Sure these systems exist, but trust me, they are a very small fraction of a very big market.
Quoted 3 lines
The whole
embedded term is just so weakly defined and everybody seem to have his personal definition
How about this: An embedded system is a computer system designed to fulfill a particular purpose.
Since embedded systems are usually made in large quantities, there is a strong pressure to make them as cheap as possible. Thus, the amount of resources assigned to a system is usually just barely enough for it to fulfill it's purpose. If it has gigs of RAM, you can be sure it needs every bit of it for doing whatever it is supposed to do.
Running on low resources is symptomatic to all embedded systems.
Helau
Rob
----------------------------------------------------------------
Robert Kaiser email: [email protected] SYSGO RTS GmbH http://www.elinos.com Am Pfaffenstein 14 http://www.sysgo.de
D-55270 Klein-Winternheim / Germany