can you damage a processor by running linux as i686 instead of i386
 

Hi,
I had CentOS v4.2 for an i686 Processor running on and AMD Athlon 850MHz. After a while I noticed my processor usage spiking. And I could install packages off the dvd. So I ran:
uname -a
and it was telling me I was infact running the i686 version on a i386 processor.
So I redownloaded the centOS dvd for an i386 processor. And I am unable to get it to start anaconda. And run other bootable cds.
eg. knoppix STD bootable cd, UBCD etc..
So I am a little stumped as to what to do.

Could I inflict physical damage by running thge i686 version not the i386????

I do not think so, the i386 and i686 references the set of instructions that the processors can handle. I would guess that the worst that can happen is that the PC would becomes unstable or hang due to incorrect instructions sent to the CPU.
I would re-format the HDD and start all over.

An Athlon 850 is fully i686 compatible. In this case i386 probably refers to the architecture/platform (like PPC and Sparc also) and the i686 to the extensions/instructions being used.

"and it was telling me I was infact running the i686 version on a i386 processor"
So I think you got that exactly backwards. You're running the i386 version of your distro on an i686 compatible processor. ;)

I agree with Crito, what he said sounds much more likely. A 386 processor wouldn't be able to handle the i686 instruction set.

There is a matter of a rule that says:

OK; this is not always accurate, as you can indeed damage your CPU, GPU or M/B overclocking some stuff so that it burns out. It's difficult but possible.

Perhaps you heart about a virus (doomsday, I think) that 'damaged' the M/B... It wasn''t damaging anything indeed, just erasing the CMOS, which is a programable *erasable* chip.

You can damage the *data* contained in a hard disk, but it would be difficult to damage the hardware itself, unless you do something physical to it, like setting a wrong jumper or throwing it from the th floor.

In the case of a CPU you can hold strictly to the rule I told above. The CPU is not an erasable chip, all it contains is physically implemented. If a code is not designed for a chip it will simply not know what to do with it and will not be able to execute it.

In the case of the i386 arch: i686 includes specialized code for Pentium processors above Pentium-Pro (PII, PIII, PIV... and AMD and similar Cyrix chips). It includes MMX and SSE extensions.
I386, on the other hand means that it only uses the default capabilities of the primitive Intel 386 chips, these functions are present on every PC based machine in the world, it doens't matter whether you use a PIV with Multithreading, and Athlon-XP Mobile, a Cyrix Via C3 or a Centrino (PIII coppermine), i386 can be runt in *any* PC based architecture, even the older ones.

The only thing that could went wrong is in the case you actually use an intel 386 or a intel 486 machine, in which case, using packages for the i686 arch would cause troubles, as many parts are using the MMX, 3DNOW! or SSE extensions and relay on hardware that is not present in the intel 386 chips.

Conclusion: The only way to harm your CPU using i686 packages instead of i386 is trying to do so while putting your PC into the microwave oven, LOL.

It should only cause the processor to throw an illegal instruction exception, should it not?
This won't harm your hardware, but will almost certainly crash the program.

Imagine the potential for virii if it were possible to destroy CPUs from software.
99% of CPUs running windows would be toast by now.

Runlevel0, I understand what you are saying and thanks for the technical explaination. So are you saying that pretty much an i386 version would run on any processor. Using the proper class of the version would more closely exhibit the processors capabilities.

Yes, that's more or less accurate. The only thing which can cause (software) problems is using a K7 package on a non-AMD architecture. Another thing to consider is that the processors themselves can optimize the execution of i386 or i486 code. I was told that indeed Athlons where optimized to use i486 code (I was asking whether extreme-optimizing in Gentoo was really useful). I will spare you the gory details, LOL, but for the insiders: It has to do with -funroll-loop, ROFLMAO.

Ah, and....

Merry Christmas and an *Extremely* happy New Year to all LQ folks !!!!

you should really be using stuff compiled with
-march=athlon
but anything less that that will still work
i686 is 10 years old

waiting with dread the reemergence of -funroll-loops guy

Most posters are incorrect. If the program is compiled for an 80386, it can also include MMX, 3DNow, SSE, SSE2, SSE3. Slackware programs are compiled for 80486 with extra instructions. You may have notice the programs are slightly larger. A program compiled for 80386 will easily be run by any processor above this with out any problems. It will not crash unless the programmer writes poor software. If an 80386 runs 80486 or above instruction set, it will just not run period. A Pentium 4 is an 80768 not 80686, so if programs compiled for a Pentium 4, it will not run on any processor below this. Compiling programs using K7 optimizing settings, can still run on any 80686 compatible processor. It just will not execute as fast.

Software can not ruin hardware. Movies may show that you can but in real life conditions shows you can not.