View Single Post
  #1  
Old November 4th 13, 06:49 PM posted to alt.comp.hardware.amd.x86-64
Vince Coen[_2_]
external usenet poster
 
Posts: 4
Default Sabertooth and a AMD 8350 speed issues

Hello Jim!

04 Nov 13 16:11, Jim Beard wrote to All:

The change from single-processor to dual-processor is typically
dramatic. A change from dual-processor to quad-processor is
usually barely noticeable unless the machine is a server under
heavy load and you have the monitoring tools to actually
recognize the effects of tinkering with settings.


The change from quad-core to 8-core will be trivial in most
circumstances. If you have massive amounts of
massively-cpu-intensive processing to do, and rewrite your
applications for optimized cpu-sharing, the results can be rewarding.


I must admit I have not done any CPU intensive work but I did run some
backups/restores from the original 3Tb HDD (E6600 system) to the AMD system
with a 1Tb HDD. Two running at the same time and the load went up to 5.0+.

Looking at both top and htop which seems to give differing output regarding
each core at best only cores 1 & 3 had any real load but well below 50% and
cores 2 & 4 had slight load. I could not see what each core was doing (Is
there a tool for that?). It would be a useful function to run a bash script
against a specific core or get the kernel to distribute the load across the
cores but again I cannot see how one does that BUT the kernel should be
doing all this without user input (as it were).


I will be running thw Windows 7 inbuilt benchmark against both boxes to see
what the numbers are however, I suspect that these tests have a bias in one
direction or another and not just core loadings and throughput etc.
Regardless even if the Windows test shows a strong difference between the
two it does not indicate what is happeing (or not) under Linux.

There has to be suitable tools for dealing with these issues under Linux
without me having to run a JCL tool (al la IBM) on the boxes.

Says, Vince wishfully.


Vince