Dennis Myhand wrote:
 
> Just for some added info, I have Mandrake 8.0 loaded on an FIC 503+, 256 megs ram,  
>AMD k6-2 333MHz, and I also have Mandrake 8.0 loaded on an old, Pentium MMX 166MHz, 
>96 megs of ram, and the cheapest MoBo that NEC could find.  The Pentium machine is at 
>least TWICE as fast as the AMD box, no matter what OS it is running.  That goes for 
>Win98, W2K, NT4.0, Solaris 8, and Mandrake Linux.  My little network
> test field has been a real eye opener to me.  I will go with Intel chips from now on.

In the meantime, something is wrong with your 503+ setup. I put a 503+
with a K6/2 450 in a friend's box, and it is plenty fast. First, find a
benchmark suite you can use on both machines so that test procedures
don't throw off the apparent results. Run a test suite and record
results. Then find out why the 503+ is slow, if in fact it is. It might
be nothing more than insufficient cache for the installed RAM. Drop the
RAM to 64 or 128 Mb and see if the results improve. Your perception of
speed based upon your expectation of speed could be a problem as well.
Your NEC motherboard is probably using an Intel chipset. The 503+ is
VIA. Without the right driver installed, a VIA chipset will provide very
disappointing I/O performance. This could easily be the difference you
see.

Something else to consider is the settings used for overclocking are the
same used for normal clocking. If you take any given K6 chip rated high,
like 500 or 550, and try all the different combinations of multiplier
and bus rate, you might be surprised at the tremendous variation among
different combinations. Some combinations that should be slower are
faster than others that should be faster.

As an example of this, I have an AOpen AX5T-3.1. This board has
ostensible multipliers up to 5.5 and bus rate up to 83. When I bought it
I installed a P55C 200 and used the standard 3.0 X 66. Later on I bumped
it up to P55C 233 using 3.5 X 66 and experienced an expected benchmark
improvement. In this configuration I used it several years. Then a few
weeks ago I installed a K6/2 500 and tried every multiplier & bus rate
configuration possible. The *only* combination that provided the
expected relative performance improvement over the P55C 233 was at 500,
which was achieved by a 2.0 X 83 setting. As it turns out, a K6/2 chip
set to a 2.0 multiplier actually runs at a 6.0 multiplier, which is as I
understand it a function of the chip rather than the motherboard. Using
the in between settings to achieve such speeds as 333, 366, 400, 412,
etc., speeds were as bad as worse than the P55C 233 to marginally
better, to better but disappointingly so. There was a vast improvement
of 500 over 450. Either the chip or the bus just isn't very efficient
with a 75 bus rate on this motherboard.
-- 
"Those who would give up essential liberty to purchase a little
temporary safety deserve neither liberty nor safety." Benjamin Franklin

 Team OS/2 ** Reg. Linux User #211409 ** Rotary ONLY since 1973

Felix Miata  ***  http://mrmazda.members.atlantic.net/ <- Not just a FAQ


Want to buy your Pack or Services from MandrakeSoft? 
Go to http://www.mandrakestore.com

Reply via email to