Text 449, 70 rader
Skriven 2007-02-13 00:23:32 av Maurice Kinal (1:140/13)
    Kommentar till text 448 av Russell Tiedt (5:7105/1)
Ärende: SED - Version
=====================
Hey Russell!

Feb 12 20:48 07, Russell Tiedt wrote to Maurice Kinal:

 MK>> any benefit to anyone, including himself.  Can't live in a bubble
 MK>> forever without somebody taking care of business.

 RT> There be a lot of truth in that as well, ... :-)

For sure.

 RT> Well, there is only one way to know whether it will be good or bad 
 RT> and that is 
 RT> to test that machine in as many ways as possible, until that is done, 
 RT> well ...

I think that is the best strategy.  I'd be surprised to hear that 128M shows
any difference then 64M on that particuar system given the intended usuage.

 RT> My experience is that it is best to load a machine with all the RAM 
 RT> it can 
 RT> handle, especially if one wants to avoid using swap ...

Yes.  However it depends on what you needed swap for in the first place.  The
only thing that seems to use it on the 486 here with 64M is gcc and then it
depends on the version of gcc that determines how large to make the swap space
and what source is being compiled.  For instance glibc is really needy and I
discovered the hard way that a i686 cpu with 128M of ram and zero swap was
simply not enough (gcc-4.1.1).  Adding a 256M swap space took care of that
situation BUT a i686 class cpu with 256M of ram and zero swap was better in
comparison.

Bottomline is that in your situation with the 486 maxed out you'll probably
want a 256M swap space if you plan to use it to take care of ALL it's potential
needs, and ONLY if you plan to throw the latest at it.  Otherwise 32M of ram
looks to be the sweet spot from this angle and that is with two 10/100 networks
onboard going full tilt.  Amazingly it works although it never gets the full
100Mb on either of the networks but that is with a 33MHz cpu.  The 100MHz you
have should see better.

 RT> Leaving room to "grow" is always prudent ... <VBG>

I agree.  Now we need to convince Pascal of that.  ;-)

Stinkin' floppies.

 RT> An AMD, one of the best ... <VVBG>

Yeah, could very well be thee best.  I heard the 133's were prone to
overheating and packing it in before their time.  I cannot confirm that but it
seems possible.  Also Paul told me Intel did make 100-DX4's but I've never seen
one, nor even heard that before today.

 RT> Looks good ...

Still need to test it but it is in place.

 RT> Was wodering about some of those "no"'s myself ...

I am sure the ghostscript related ones are due to the lack of a shared lib I
couldn't build due to the lack of the gtk dependency but some of the rest look
odd to me.  I'll have to check them out to be sure.  fontconfig being one that
seems odd since I have some other stuff that lives with things as is,
fontconfig speaking.

Life is good,
Maurice

--- Msged/LNX 6.2.0
 * Origin: Coffin Point BBS (1:140/13)