lkml.org 
[lkml]   [1996]   [Dec]   [22]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
Date
Subjectre: AMD (fwd)

>Pentium = All Pentium optizmizations
>AMD 5x86 = All 486 Stuff, no Pentium stuff.

Right.

>A Pentium is a Pentium, a AMD5x86 is a 486, just made run to quadrupled in
>speed, no added Pentium instructions or anything.... AMD5x86 can do
>everything a 486 can, but can't do everything a Pentium can, because it
>dosen't have any Pentium instructions. here are answers to your questions, in
>order:

>1) No. AMD5x86 only has 486 instructions. Pentium has Pentium
> instructions.

>2) A AMD5x86 can't do everything a Pentium can, because it doen't have
> any Pentium instructions.

>3) I would say the Pentium is worth the extra $$$. If you get that 5x86
> now, you'll regret it later down the road. If you get the Pentium,
> you'll regret it a lot later down the road :>

There is more to the issue than that. Apparently the AMD586 has a better
cache than any of the 486 CPUs and therefore performs much better than a 486,
especially under high load. A friend of mine who wrote a data warehousing
applocation found that his application ran faster on a AMD586-133 than on an
Intel Pentium 100. The Pentium 100 won all standard bench-mark tests that he
put to it though...
I upgraded one of my servers from a 486-33 with 16megs RAM to a AMD586-133
with 32megs. Before the upgrade the system load was at a minimum of 3 when the
machine was in use, and often went above 5. After the upgrade system load was
down to less than 0.1 for most of the time. I believe that this indicates that
the 586-133 is more than 4 times faster than the 486-33, but this is far from a
scientific test.
If you buy an AMD 586 then you can use a cheap 486 motherboard and only need
a single SIMM. You save money on CPU, motherboard, and memory.

I run 3 Linux servers in 24*7 operation. They all have AMD586 CPUs and all
run as well as can be expected (only problems I get are from bugs in the Linux
kernel such as the socket FIN_WAIT2 problem we all know about).


Russell Coker

\
 
 \ /
  Last update: 2005-03-22 13:38    [from the cache]
©2003-2011 Jasper Spaans