"Michael S." said:
QZ--the jumpers on a Raptor should be left as they were when they left the
factory since they are SATA drives and the various jumper settings have to
do with testing them as well as special purposes when they are used in an
enterprise situation.
As for the Spread Spectrum setting in the BIOS--I have included a link below
and when you get there, scroll down a little ways to find the information
regarding Spread Spectrum. This was interesting to me since I let it set at
the default which I did with several BIOS settings about which I knew nada.
This may explain why I have been unable to OC by more than 5% -- won't even
boot if I OC more than 5% using the auto setting, and even 5% is unstable.
I did not build this new system to OC it and was aware that this mb was not
famous for its OC'ing capabilities, but out of curiosity, like to tinker.
http://www.adriansrojakpot.com/Speed_Demonz/BIOS_Guide/BIOS_Guide_02d.htm
MikeSp
Adrian's explanation has some merit, but there are a few more details.
There are two kinds of spreading, centered and down-spread. The
centered kind means the clock runs both faster and slower than nominal.
A downspread clock runs at the nominal speed, or it will run slower
than nominal. A downspread clock won't eat into your overclock setting,
because the clock won't run faster than the value you set it to.
Instead, it just reduces the performance you get (i.e. your benchmark
will be ever so slightly slower, due to the _average_ frequency of
the downspread clock being lower than nominal.)
This datasheet, is for a clock generator that supports both kinds of
spreading. I picked this part specifically to demonstrate both types.
http://www.cypress.com/cfuploads/img/products/cy25200.pdf (pg.6)
Downspread is shown on the left column of graphs. In this case the
spread is -4%. Center spread is shown on the right. The clock runs
faster than nominal, then slower than nominal. The average frequency of
center spread equals the nominal frequency. The average frequency of
down spread is less than the nominal frequency. The +/- 1% modulation
would have a 1% impact on your overclock setting, but there is no
benchmark penalty, because "average equals nominal". In a sense,
overall, the center spread hurts you less, because of the 1% part,
versus the 4% of the other one shown in this example.
Spread spectrum is a crock, because the pattern is not chosen to make
other appliances work better - the modulation pattern is chosen to
fool FCC testing procedures. You can disable it, as long as broadcast
TV reception is not being impacted by your computer. There are other
ways in design, to meet FCC, without using spread spectrum.
Spread spectrum should not be used on all clocks in a computer.
There are plenty of interfaces that need a clean clock, so the
computer will have a mix of spread and unspread signals.
As for interference mechanisms, I have the most problems here with
conducted interference. I've found that the EMI filter in my
UPS prevents electrical noise from the PC going down the power
cable and into the wall. And, that kind of noise has nothing to
do with the processor clock - that noise comes from the switching
inside of the PSU and the lack of EMI filtering in cheap PSUs.
The PSU switching can create a herringbone pattern on your TV.
More trivia,
Paul
