Things that could gate off the PS_ON# signal:
1) AGP Warn circuit. I don't see one on the board - usually the
evidence is an "AGP_Warn" logo in the silk screen. I think two
or more transistors are involved. The transistors should be
disabled, in any case, if the video card is removed.
2) The board uses an Attansic.com.tw ATTP3, which a German review
site claims is an 8 pin overtemp detector for the processor.
Although a datasheet isn't available for that part number, the
German site claims it is a fixed version of the ATTP1.
http://www.attansic.com.tw/english/products/pdf/attp1_1.1.pdf
The PS_ON# signal flows through that IC. When you press the power
button on the front of the computer, check pin 3 for a signal
near zero volts (logic 0) coming from the rest of the board.
The output on pin 6 should follow suit and be zero volts, when
the input on pin 3 is at zero volts. If the input and output
don't match states, it means the ATTP? is gating off the signal.
A test you could try, is to remove all components from the board.
With the processor removed, there won't be any thermal diode connected
to the ATTP? . If the board starts then (i.e the fans spin), but
not with the processor in place, it means either the processor
thermal diode is out of spec, or the ATTP? is out of spec.
You could also visually examine the three legged diode next to
CLRTC and the CMOS battery. If a user has done a clear CMOS
operation with the power applied to the system, sometimes the
diode gets burned, and I don't know what would happen without
bias on the Southbridge CMOS well. The dual diode is above the
CLRTC jumper and to the left of the word "CLRTC".
Just a guess,
Paul