CT-7NJL6 w/ barton 3000+ only running at 1300

I left the memory settiings on auto, ostensibly to detect the most
efficient numbers for the RAM. The chips ran fine, but there was a
slight decrease in overall system speed. I may investigate further
and play with the settings.

About the 2900 chip: I had to manually increase CPU frequency in BIOS
(v.1). As shipped, the board set the 2900 CPU at 133 MHz and the
resulting CPU speed was around 1.3 GHz. So, what you read seems
accurate.

I noted that chip is not listed in the manual.
Have you dropped Chaintech a line concerning that? There support
usually gives an answer in a couple of days.
I've noted you kept mentioning overclocked. Are you actually running
the 2900 at 3200?
Since you likely got it from Tiger, I'd be inclined to say ask them,
"NOT".
I've never seen any company thriving that was such a PIA to deal
with.

The manual lists the 3200 Barton, who's core voltage is at or lower
than most slower chips. Have you ever noted a temp issue? Higher
voltage, more heat. But no specs as to current draw. Likely would
have to get from AMD, but I'd bet the draw is higher than for other
chips. Of note would be the figures for 3200 (400FSB) compared to the
3000 (333FSB). Might also be a difference in voltage/current draw for
your RAM sets, and in what should be the maxed/matched performance,
you end up tipping the scale, but that won't fit if the problem is
with the slower.
I have to ask which utility you are using to measure performance? Many
of them no longer work with 9x OS. It might be interesting to run
speedfan simultaneously to see how temps end up. It is the only one
I've found to monitor temps, two sensors, one of them probably on
cpu.

Another possibility that comes to mind of performance, is the north
chip and loaded memory controllers. On this mobo, it looked like the
heat sink had been disturbed after the paste dried, causing voids and
air pockets. I have no idea if certain data related to memory
controllers, such as RAM speed or config ends up stored in registry
or other file. If such factors are important to the drivers, it is
possible that might be stored to speed loading of OS.

I do wish Chaintech had given a detailed change of bios. Seems there
must be some major change to require reloading OS? Way beyond just
detecting Sempron chips. That would seem rather simplistic, to have
the bios interegate the cpu for recognition, and then just continue
with loading. There has to be something else such as how other
functions interact with each other, even if in the same chip(north or
south).
It couldn't be something as simple as the OS reporting the correct cpu
in the device manager, could it? I will browse through the OEMx.inf
files to see if there is a clue.

Norman