That's a tough one. I would definitely try upgrading the BIOS, at least before giving up on it. I did a little Google research and didn't find anything quite like your prob. But here's the A7V33 manual, in PDF form, in case you don't have yours any longer. If I think of anything else I'll be back. Good luck!
http://www.nopanic.it/hw/e1010_a7v333.pdf
---------- Post added at 05:20 PM ---------- Previous post was at 04:58 PM ----------
More info:
Your results are exactly what the manual insert says as well:
http://www.asus.com/pub/ASUS/mb/so [...] a7v333.pdf
Two double sided DIMMs work at DDR333 rates
Three double sided DIMMs work at DDR266 rates
The performance gain is benchmarked here:
http://www.lostcircuits.com/mother [...] 33/8.shtml
In the "buffering disabled" picture, compare "133 1:1 Turbo"
to "133 4:5 222" 1372 vs 1446. (That is as close as I can find
to two comparable sets of conditions. 4:5 is running the memory
at DDR333 while the FSB is FSB266.)
That is 5% more memory bandwidth, which will give 1.5-2% more
application performance. Rather than running asynchronous, you
might be better off running synchronous and trying to overclock
the FSB a bit instead.
Also, get a copy of memtest86 from memtest.org, as it can test
_all_ bytes of the memory. The program is unique, in that it
has no OS, and the program moves itself out of the way, and
tests underneath itself. The download from memtest.org, when
executed, will format a self-booting floppy. The computer boots
from the floppy and testing begins immediately.
The spec for the voltage, at least at the power supply, is
5% on +3.3V. You have a 10% drop, so perhaps verifying the
voltage with a multimeter would be a good idea. The low
voltage can be caused by a bad PSU, or it can also be
caused by a damaged ATX 20pin connector. Try reseating the
connector first, and see if the voltage is any better after.
If not, disconnect the power cable and examine the pins for
signs of heat damage. If the connector has been damaged, then
replacing the PSU would only work for a short time, before
the new connector on the new PSU would get damaged as well.
Unsoldering and replacing connectors like this, is not a lot
of fun, as much heat is required to work with those pins. So,
I hope this is not a case of connector damage.
In most cases, another power supply will fix it, if the power
supply is the source of the problem.
Memory does not use that power directly. Power is regulated on
board. Asus uses some linear regulation designs, so the memory,
for example, could be linearly regulated from 3.3V down to 2.5V.
If your 3.3 drops to 2.9, and the memory voltage is 2.6V, there
is 0.3V drop across the MOSFET, which I'm guessing is still
workable.