[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Memory Requirements

> I read about it in an email about what's new in 2.2 that I received.
> I've mentioned it on my local linux mailing list and nobody corrected
> me. It should be relatively easy to check, but I haven't.

Here's the situation with the machine I use at work:

$ cat /proc/version
Linux version 1.2.13 (Linux-FT 1.2) FT-1.2
$ dmesg | grep Memory
Memory: 63804k/65536k available (684k kernel code, 384k reserved, 664k data)

And here's my home machine with my latest hacked kernel:

$ cat /proc/version 
Linux version 2.1.132 (ege@rano) (gcc version #38 Mon Jan 4 21:09:32 GMT 1999
$ dmesg | grep Memory
Memory: 61260k/63488k available (744k kernel code, 408k reserved, 1028k data, 48k init)

I don't see why these shouldn't work with 4 MB, and I don't see why
2.2 should be any bigger. I suppose I could try it; I think there's a
kernel argument to artificially limit the available memory ...

> As for why it should need more than 4meg? Well it's pretty hard to
> install a major linux distribution with less than 8. Features feature, 
> and of course the hardware is cheaper and more available.

Of course it's disc space that determines whether you can install or
not. Insufficient memory usually just makes certain programs, or
cominations of programs, run intolerably slowly.

> I don't write the kernels, I just use them. I guess that means I don't 
> have a lot of say in the memory requirements. I think Linus Torvalds
> uses a quad xeon, I don't suppose he's overly concerned.

I should imagine that lots of people would like it to remain possible
to use Linux to turn an old 486 machine with 4 MB into a router,
firewall or whatever with the latest IP stuff ...

Anyway, if you want to install the Hurd, you probably also want to
develop with it, so you probably want a faster machine with more
memory in order to compile quickly.

In principle the microkernel design of the Hurd should make it
possible to get away with less memory because there is less unpagable
kernel space stuff but it doesn't make much practical difference. For
example, XFCom_SiS alone is currently using 4772 kB on this machine so
I wouldn't be very happy with only 4 MB of real memory however small
my kernel were ...


Reply to: