Whisky is a great base for computer naming (even though if you have 4000 machines and a 8-character limit it will not work very well).
I use Great Single malts for my favourite machines (Ardbeg for my G4, Port-Ellen for my Firewall and Auchroisk for my laptop). Machines that I dont like that much, particularly those running windows, can be named using nasty American blends like Jim Bean (Huh!).
Looking at www.maltmadness.com most people will find more Whiskys than they have computers (and they are rated as well).
If you are using American, Irish, Canadian and Scotch Wiskies and still cant come up with more names, just add bottling etc:
Pipelining as a concept was introduced with RISC. The idea of RISC is that all instructions should take exactly ONE clyck cycle to execute (otherwise that fuctionality should be achieved by a series of other instructions).
When (and only when) each instruction takes one clock cycle to execute you can divide each instruction inte several sub-tasks, for example Fetch - Decode - Execute - Store. Now a pipeline can be introduced:
instruction 1 F D E S
instruction 2 . F D E S
instruction 3 . . F D E S
instruction 4 . . . F D E S
In this way work is parallellised. Four instructions are executed in parallell, during 4 clock cycles each, giving a throughput of 1 instruction per second.
CISC basically means that instructions take different number of clockcycles to execute. With such an instruction set you can not use Pipelining.
The i386 cpus today transform the CISC instructions into RISC instructions, and apply pipelining. For the price of a translator the major drawbacks with CISC are avoided.
It is probably true that Intel has to put more research and silicon into a CPU than Motorola to achieve the same performance. But this is not because Intel are stuck with building a CISC cpu - it is because Intel must emulate the i386 CISC instruction set on a RISC cpu core.
The code tested computed some kind of geometric mean... It is not surprising att all that performance can be improved significantly by optimising against parallell instructions (such as SSE/SSE2). There is no guarantee that any major improvements will be seen on an "ordinary application".
However, I will find use for this information and I will try Intels compiler and compare it to GCC.
Very minor changes in the code of this kind of high-performance applications can result in very big speed-ups, with any compiler. It would be interesting to see some real world problem (some PDE-model or something) based on for example BLAS (Basic Linear Algebra Solver, or something, www.netlib.org), being computed with gcc/icc and see the "real" difference.
What about setting up a linux machine and connect X-terminals to it, thus providing multiple users with internet access, but they are on the same machine. Or a windows terminal server. Or ssh in and run applications that are forwarded over X. Or port forwarding.
And, windows 98/ME does this automatically if you have a windows LAN with one computer connected to the internet, doesnt it?
First, I never said Aunt Tillie would ever compile a kernel the old way;)
A good thing when configuring a kernel (using xconfig) is to change as few things as possible. Do not add drivers to devices you might buy in 6 months. If you get a new device you can just recompile the kernel then. Most devices can be compiled as modules - that is a good idea for devices that are not needed for getting the system up.
Is you saw, I forgot "make bzImage". That command should be given after "make clean" and before "make modules".
A good start (if you usually fail when compiling your kernel) is to start "make xconfig" and exit and save immediately (that is the same as "make config" give default answer to all questions). Try to compile this standard kernel. It might not run your system, but it will tell you whether the compiler is working well. If this test fails you probably have to upgrade the compiler (be careful with gcc 3 though). I prefer Slackware and Debian, and the compilers in these distros usually work well. Yesterday Redhat 7.2 turned out to have a good compiler too (for the purposes of compiling the kernel).
If this test compilation worked then "make clean" and "make xconfig" again. Now change:
-> CPU-model (do not pick a newer one than the one you have)
-> SMP?
-> SCSI adapter (if you have none, turn off scsi)
-> Network adapter (as module)
-> Sound (you can take a bunch of them, as modules of course)
-> You might want ramdisk and loopback device as well (In block devices section, I think).
-> You might want vfat (but probably not umsdos)
When it comes to ISDN and USB I have no experience - I never use it.
Now exit and save and "make bzImage"
The new kernel is arch/i386/boot/bzImage
now:
#cp bzImage/boot/new.kernel
add to/etc/lilo.conf
image=/boot/new.kernel
root=/dev/hda1 (where your root is)
label=new
read-only
now:
#/sbin/lilo (i386 only!)
if lilo gave you no errors the kernel is properly
installed. If you are running sparc just edit/etc/silo.conf, do NOT run silo. If you are running some other arch I dont know...
...it is a good thing to not be root when compiling the kernel, untar it in your users home directory and compile it there.
If you still fail:
-> What distribution are you using?
-> i386?
-> What is failing?
I mostly run debian (when it comes to linux) nowadays. Why? Becasue of the dselect convenience. I have thousands of pre-configured free packages to choose from.
Did I compile a lot of stuff when I ran slackware? Yes.
Do I know how to build debian packages? No.
Would I be able to build a debian package if I found out I needed to? Probably.
I dont feel I would have more freedom if there was a very simple program that created.deb-files (maybe there is). In fact, freedom for me is not worrying about how the.deb-files were created in the first place.
Why would an automatic kernel-compiling-wizard give more freedom to users than the opportunity to choose from a set of precompiled kernels?
But of course, I can not argue that more choice is less freedom. Hopefully the tool gets so good everyone uses it.
tar -xzf linux-2.4.17.tar.gz
cd linux
make xconfig
make dep
make clean
make modules
make modules install
...and make it boot...
I mean, If you think that is hard you probably wont be able to give any useful instructions to a kernel configuration program at all... Maybe not even know you need a new kernel...
What is nice with linux (compared to Windows) is that very few things happen "behind your back". The system does not change itself. I find this very comforting.
And modern distributions tend to make it quite easy anyway... I installed Redhat 7.2 from isos today for the first time in over a year. All hardware was autodetected and worked without any tweaking at all (then I felt like compiling an own kernel to play DVDs well, but that another history)
I have used 2.4.* since the very beginning on my firewall (to get IP-tables). arch/i386/kernel/bluesmoke.c is broken for my IBM PC Server, Dual P90 since 2.4.5 or something but just replacing with the old version works fine.
I am running the entire system in RAM (loading a 96 Mb almost empty ramdisk from floppy, populating the filesystem via rcS from/dev/hda and then shutting down the hard drive). I had problems with this in 2.2 and early 2.4 (perhaps 2.4.10) because for some reason double amount of RAM was required. Allocating 20 Mb of RAMdisk cost me 20 Mb of ram, but when I filled the filesystem another 20 Mb was used. sync did not help. Unmounting and remounting helped... but it was my root filesystem and no good solution.
Conclusion:
If you run old hardware, puts little load on it, and patches the kernel before you compile it you will be quite satisfied with 2.4.*;)
Oh, btw. I run 2.4.* on a laptop (with more than enough RAM). 2.4.* has been stable and satisfactory since the first release, for me.
I can't help it, but as a non-gamer I find all computers on the market today more than sufficient for my needs. Any CPU that I can buy brand new is enough. Memory is more or less free, and 512 Mb is more than I've ever needed. A 20Gb hard drive makes it. Most graphics adapters can do 1600x1200x85.
I just want something that is "completely" silent. My Mac G4 is not silent. My IBM Thinkpad is not silent. The Sun Blade 100 at my work is not silent.
There was an article a few days ago on Slashdot telling how to build a quiet performance PC. I believe dropping "performance" could make it even more silent.
How do I build a machine (to run my favorite free os) that is completely silent?
- Hey, I want to buy this computer.
- Ok, do you have license for Hard Drive class B, CPU class A and Full featured NICs?
- What does CPU class A mean?
- That it can boot uncertified operative systems, and run uncertified software.
- Sure, I am a scientist, scientific use you know.
When do you think we will hear this kind of dialog? The scary thing is that most people wont care, and the government will say that "you are not allowed to use any frequency you like with your cellular phone either, so whats the problem".
Computers are cool because they can be programmed to do anything. The devices are supposed to provide basic services.
If shit like this is put in computers, will the computer still feel like a universal tool? There is of course a limit when a computer is to restricted to feel like a computer. Most geeks here appreciate computers because they are so flexible.
Hopefully, there will always be a quite large demand for computers like they are today. And even though most people might buy "computers" that are more like a Nintendo with WWW/email/DVD/office, for scientific uses and in the industry, truly flexible and programmable computers will be needed.
Fully carbonized! That means (?) you can develop a program in a mixed Linux/OS X environment, and get Aqua look and feel in OS X. I think it is time to learn Qt;)
Since W2k i have found Windows to be not all that bad anymore
That is exactly what I am saying. But even though that is and continues to be my opinion I have bad surprising experiences weekly. And still I say; "W2k is quite good, despite the fact that i hate MS".
What currently is annoying at our site is that the built in backup program in W2k refuses to do what we tell it, it keeps backing up the same old share all the time, regardless of what we tell it to. (This is so crazy I could not possibly have made it up)
Another annoying feature is this HAL (Hardware abstraction layer). It prevents us, you or anyone from moving a W2k installation between two computers with different HALs, even though you tell Sysprep to check for a better HAL. It turns out that we CAN move an installation from a Compaq Desktop to a Dell if setting the HAL to "Standard", but under no circumstances in the other way.
Not to talk about Outlook Express that:
- Stores EMAIL on the local computer by default
- Stores THAT INFORMATION in the users profile
That is: all user settings, temporary files, and other files are stored in the profile, EXCEPT the users EMAIL which is stored locally and LOST when the user logs out.
One time out of 100 the users roaming profile is not copied back to the server when the user logs out, and the users loses all changes made in the profile, including My Documents.
The idiotic default settings in "folder options", a good practice is to change ALL the values from their default to their opposite.
What is so strange is that every time one hear such things about MS, it is at least a slight surprise. I never learn
I work daily in a MS environment, and to motivate myself I think thoughts like "W2k is quite stable anyway" or "It is quite simple install a Bla Bla in Windows". The result is that I am always suprised how bad Windows actually is.
So, even if MS has exactly the plans you describe, and even if I completely believe you, it is hard to actually understand it, to grasp the very idea of it.
You say "its just not gonna happen. I say so to! We dont want it to happen. But lets assume it is about to happen. And lets assume we can stop it - IF we see it coming.
Our good beliefs in MS, our bad imagination, or our opinion, makes us claim it wont happen, and that is exactly what will make it possible!
I just heard on Swedish television that this could possibly be used in microchips in the future... at 20 nK I doubt it... Did the journalists find that out themselves, or has anyone else heard any more details?
I just wonder... exactly what are you going to do with that piece of information? Is it actually useful for any purpose at all? I can hardly come up with an example of less relevant data. Voltage of the motor in a DVD-player?
For people who know a thing or two about cpus, MHz is and has been a useful measure of performance, but primarily in the sense that 333MHz >> 60MHz. If we were often interested in FLOPS those values would have been seen more. Today 1 Hz != 1 Instruction per second...
I guess the kernel developers can use the MHz value for something, but they can probably figure it out too. And of course, I guess that the internal clock frequency of a cpu will be mentioned in a technical specification, available for anyone who might need it along with other information like the function of all pins. And for you to.
Get over it, its like getting used to the metric system;)
Apart from the containment vessel, no radioactive waste will be produced at all.
The reactor itself will be radioactive in a harmful way for about 100 years (this ultimately depends on what material it is built of). This is much better than the 10000+ years for a fission reactor, and totally acceptable, I think.
An old coal power plant is also quite radioactive, at about the same magnitude as a fusion power plant.
Slashdot Mirror
Whisky is a great base for computer naming (even though if you have 4000 machines and a 8-character limit it will not work very well).
I use Great Single malts for my favourite machines (Ardbeg for my G4, Port-Ellen for my Firewall and Auchroisk for my laptop). Machines that I dont like that much, particularly those running windows, can be named using nasty American blends like Jim Bean (Huh!).
Looking at www.maltmadness.com most people will find more Whiskys than they have computers (and they are rated as well).
If you are using American, Irish, Canadian and Scotch Wiskies and still cant come up with more names, just add bottling etc:
ardbeg1975
ardbeg17years
ardbeg_caskstrength
etc.
Dude, your 110GHz Dell consumes 450kW, and requires its own diesel generator...
Why bother porting Linux to the PowerPC when
BSD already runs fine on it in the form of
NetBSD?
Why bother porting NetBSD to the PowerPC when Mac OS already runs fine on it?
When (and only when) each instruction takes one clock cycle to execute you can divide each instruction inte several sub-tasks, for example Fetch - Decode - Execute - Store. Now a pipeline can be introduced:
instruction 1 F D E S
instruction 2 . F D E S
instruction 3 . . F D E S
instruction 4 . . . F D E S
In this way work is parallellised. Four instructions are executed in parallell, during 4 clock cycles each, giving a throughput of 1 instruction per second.
CISC basically means that instructions take different number of clockcycles to execute. With such an instruction set you can not use Pipelining.
The i386 cpus today transform the CISC instructions into RISC instructions, and apply pipelining. For the price of a translator the major drawbacks with CISC are avoided.
It is probably true that Intel has to put more research and silicon into a CPU than Motorola to achieve the same performance. But this is not because Intel are stuck with building a CISC cpu - it is because Intel must emulate the i386 CISC instruction set on a RISC cpu core.
The code tested computed some kind of geometric mean... It is not surprising att all that performance can be improved significantly by optimising against parallell instructions (such as SSE/SSE2). There is no guarantee that any major improvements will be seen on an "ordinary application".
However, I will find use for this information and I will try Intels compiler and compare it to GCC.
Very minor changes in the code of this kind of high-performance applications can result in very big speed-ups, with any compiler. It would be interesting to see some real world problem (some PDE-model or something) based on for example BLAS (Basic Linear Algebra Solver, or something, www.netlib.org), being computed with gcc/icc and see the "real" difference.
What about setting up a linux machine and connect X-terminals to it, thus providing multiple users with internet access, but they are on the same machine. Or a windows terminal server. Or ssh in and run applications that are forwarded over X. Or port forwarding.
And, windows 98/ME does this automatically if you have a windows LAN with one computer connected to the internet, doesnt it?
Huh?
Hi!
;)
/boot/new.kernel
/etc/lilo.conf
/etc/silo.conf, do NOT run silo. If you are running some other arch I dont know...
First, I never said Aunt Tillie would ever compile a kernel the old way
A good thing when configuring a kernel (using xconfig) is to change as few things as possible. Do not add drivers to devices you might buy in 6 months. If you get a new device you can just recompile the kernel then. Most devices can be compiled as modules - that is a good idea for devices that are not needed for getting the system up.
Is you saw, I forgot "make bzImage". That command should be given after "make clean" and before "make modules".
A good start (if you usually fail when compiling your kernel) is to start "make xconfig" and exit and save immediately (that is the same as "make config" give default answer to all questions). Try to compile this standard kernel. It might not run your system, but it will tell you whether the compiler is working well. If this test fails you probably have to upgrade the compiler (be careful with gcc 3 though). I prefer Slackware and Debian, and the compilers in these distros usually work well. Yesterday Redhat 7.2 turned out to have a good compiler too (for the purposes of compiling the kernel).
If this test compilation worked then "make clean" and "make xconfig" again. Now change:
-> CPU-model (do not pick a newer one than the one you have)
-> SMP?
-> SCSI adapter (if you have none, turn off scsi)
-> Network adapter (as module)
-> Sound (you can take a bunch of them, as modules of course)
-> You might want ramdisk and loopback device as well (In block devices section, I think).
-> You might want vfat (but probably not umsdos)
When it comes to ISDN and USB I have no experience - I never use it.
Now exit and save and "make bzImage"
The new kernel is arch/i386/boot/bzImage
now:
#cp bzImage
add to
image=/boot/new.kernel
root=/dev/hda1 (where your root is)
label=new
read-only
now:
#/sbin/lilo (i386 only!)
if lilo gave you no errors the kernel is properly
installed. If you are running sparc just edit
...it is a good thing to not be root when compiling the kernel, untar it in your users home directory and compile it there.
If you still fail:
-> What distribution are you using?
-> i386?
-> What is failing?
Good luck!
I mostly run debian (when it comes to linux) nowadays. Why? Becasue of the dselect convenience. I have thousands of pre-configured free packages to choose from.
.deb-files (maybe there is). In fact, freedom for me is not worrying about how the .deb-files were created in the first place.
Did I compile a lot of stuff when I ran slackware? Yes.
Do I know how to build debian packages? No.
Would I be able to build a debian package if I found out I needed to? Probably.
I dont feel I would have more freedom if there was a very simple program that created
Why would an automatic kernel-compiling-wizard give more freedom to users than the opportunity to choose from a set of precompiled kernels?
But of course, I can not argue that more choice is less freedom. Hopefully the tool gets so good everyone uses it.
It could also be a discussion whether it is worth the job to develop such a tool... Or whether it would make anything better...
How much easier can it be?
tar -xzf linux-2.4.17.tar.gz
cd linux
make xconfig
make dep
make clean
make modules
make modules install
...and make it boot...
I mean, If you think that is hard you probably wont be able to give any useful instructions to a kernel configuration program at all... Maybe not even know you need a new kernel...
What is nice with linux (compared to Windows) is that very few things happen "behind your back". The system does not change itself. I find this very comforting.
And modern distributions tend to make it quite easy anyway... I installed Redhat 7.2 from isos today for the first time in over a year. All hardware was autodetected and worked without any tweaking at all (then I felt like compiling an own kernel to play DVDs well, but that another history)
I have used 2.4.* since the very beginning on my firewall (to get IP-tables). arch/i386/kernel/bluesmoke.c is broken for my IBM PC Server, Dual P90 since 2.4.5 or something but just replacing with the old version works fine.
/dev/hda and then shutting down the hard drive). I had problems with this in 2.2 and early 2.4 (perhaps 2.4.10) because for some reason double amount of RAM was required. Allocating 20 Mb of RAMdisk cost me 20 Mb of ram, but when I filled the filesystem another 20 Mb was used. sync did not help. Unmounting and remounting helped... but it was my root filesystem and no good solution.
;)
I am running the entire system in RAM (loading a 96 Mb almost empty ramdisk from floppy, populating the filesystem via rcS from
Conclusion:
If you run old hardware, puts little load on it, and patches the kernel before you compile it you will be quite satisfied with 2.4.*
Oh, btw. I run 2.4.* on a laptop (with more than enough RAM). 2.4.* has been stable and satisfactory since the first release, for me.
I can't help it, but as a non-gamer I find all computers on the market today more than sufficient for my needs. Any CPU that I can buy brand new is enough. Memory is more or less free, and 512 Mb is more than I've ever needed. A 20Gb hard drive makes it. Most graphics adapters can do 1600x1200x85.
I just want something that is "completely" silent. My Mac G4 is not silent. My IBM Thinkpad is not silent. The Sun Blade 100 at my work is not silent.
There was an article a few days ago on Slashdot telling how to build a quiet performance PC. I believe dropping "performance" could make it even more silent.
How do I build a machine (to run my favorite free os) that is completely silent?
AFAIK bash is GNU so vendors may have license problems including it. I think this is why Apple is not including bash in Mac OS X.
I think we are talking about Apple's patent. Not Adobe's.
A patent is valid at most 20 years from the time being filed in the first place (1992 + 20 years). A renewal fee is required every year.
- Hey, I want to buy this computer.
- Ok, do you have license for Hard Drive class B, CPU class A and Full featured NICs?
- What does CPU class A mean?
- That it can boot uncertified operative systems, and run uncertified software.
- Sure, I am a scientist, scientific use you know.
When do you think we will hear this kind of dialog? The scary thing is that most people wont care, and the government will say that "you are not allowed to use any frequency you like with your cellular phone either, so whats the problem".
Black markets for computer hardware...
Am I
[ ] Off topic
[ ] Paranoid
If shit like this is put in computers, will the computer still feel like a universal tool? There is of course a limit when a computer is to restricted to feel like a computer. Most geeks here appreciate computers because they are so flexible.
Hopefully, there will always be a quite large demand for computers like they are today. And even though most people might buy "computers" that are more like a Nintendo with WWW/email/DVD/office, for scientific uses and in the industry, truly flexible and programmable computers will be needed.
Seems like a very lightweight laptop to me... but with neither a NIC nor a HD it probably wont do that job very well...
IF it had a NIC, it could mount NFS, and that would be more cool...
Any "device" that can run a fully featured Unix with network support should be useful for some thing or some persons.
Yes! It can even interface with alien systems (Independence Day)!
Fully carbonized! That means (?) you can develop a program in a mixed Linux/OS X environment, and get Aqua look and feel in OS X. I think it is time to learn Qt ;)
That is exactly what I am saying. But even though that is and continues to be my opinion I have bad surprising experiences weekly. And still I say; "W2k is quite good, despite the fact that i hate MS".
What currently is annoying at our site is that the built in backup program in W2k refuses to do what we tell it, it keeps backing up the same old share all the time, regardless of what we tell it to. (This is so crazy I could not possibly have made it up)
Another annoying feature is this HAL (Hardware abstraction layer). It prevents us, you or anyone from moving a W2k installation between two computers with different HALs, even though you tell Sysprep to check for a better HAL. It turns out that we CAN move an installation from a Compaq Desktop to a Dell if setting the HAL to "Standard", but under no circumstances in the other way.
Not to talk about Outlook Express that:
- Stores EMAIL on the local computer by default
- Stores THAT INFORMATION in the users profile
That is: all user settings, temporary files, and other files are stored in the profile, EXCEPT the users EMAIL which is stored locally and LOST when the user logs out.
One time out of 100 the users roaming profile is not copied back to the server when the user logs out, and the users loses all changes made in the profile, including My Documents.
The idiotic default settings in "folder options", a good practice is to change ALL the values from their default to their opposite.
I could continue in hours...
I work daily in a MS environment, and to motivate myself I think thoughts like "W2k is quite stable anyway" or "It is quite simple install a Bla Bla in Windows". The result is that I am always suprised how bad Windows actually is.
So, even if MS has exactly the plans you describe, and even if I completely believe you, it is hard to actually understand it, to grasp the very idea of it.
You say "its just not gonna happen. I say so to! We dont want it to happen. But lets assume it is about to happen. And lets assume we can stop it - IF we see it coming.
Our good beliefs in MS, our bad imagination, or our opinion, makes us claim it wont happen, and that is exactly what will make it possible!
I just heard on Swedish television that this could possibly be used in microchips in the future... at 20 nK I doubt it... Did the journalists find that out themselves, or has anyone else heard any more details?
I just wonder... exactly what are you going to do with that piece of information? Is it actually useful for any purpose at all? I can hardly come up with an example of less relevant data. Voltage of the motor in a DVD-player?
For people who know a thing or two about cpus, MHz is and has been a useful measure of performance, but primarily in the sense that 333MHz >> 60MHz. If we were often interested in FLOPS those values would have been seen more. Today 1 Hz != 1 Instruction per second...
I guess the kernel developers can use the MHz value for something, but they can probably figure it out too. And of course, I guess that the internal clock frequency of a cpu will be mentioned in a technical specification, available for anyone who might need it along with other information like the function of all pins. And for you to.
Get over it, its like getting used to the metric system ;)
Apart from the containment vessel, no radioactive waste will be produced at all.
The reactor itself will be radioactive in a harmful way for about 100 years (this ultimately depends on what material it is built of). This is much better than the 10000+ years for a fission reactor, and totally acceptable, I think.
An old coal power plant is also quite radioactive, at about the same magnitude as a fusion power plant.