The first circuit is the basis for the other two - the middle one adds a 555 configured solely to buffer the output, and the final one uses the 556 dual 555 chip to implement the middle circuit on a single chip. All have similar deficiencies and the output buffer in the latter two isn't necessary or desirable if you take the output from the Discharge pin as I recommended above. I neglected the effect of the 1K resistor R1 between Out and the diodes - it gives a little more flexibility for the choice of timing capacitor but prevents the duty cycle going closer to 100% than about 98%. As you said the fans are excessively loud, I doubt that will be an issue. A 1nF cap for C2 should be acceptable, and is far easier to source than a 910pF or 820pF cap.
The resistor R2 is the pullup resistor and is *SUPPOSED* to be internal to the fan. If you use an external one, and a 12V supply, and connect a fan, you'll also need a 5.1V Zener diode, anode to Gnd, cathode to Discharge to clamp the output and prevent it going significantly above 5V. However if breadboarding without a fan connected you'll need that resistor to see any output on Discharge. I recommend *NOT* adding R2 - its *NOT* compliant with the Intel fan control spec, and isn't needed if yoiu buy a CMOS 555.
Polarised caps without any other indication will be Aluminum Electrolytics. 16V or 25V rated ones would be suitable.
You can check the output at Discharge without a fan connected if you temporarily connect R2 and monitor the output with a multimeter on DC Volts. It should vary between 2% and 98% of the supply voltage. If your DMM does frequency, the same signal should be below 28KHz at both ends of the pot rotation and above 21KHz at its center position. Alternatively if you've got a scope, use it!
See
http://www.formfactors.org/developer/specs/4_wire_pwm_spec.pdf for the actual specs for the Intel 4 wire fan control. You can ignore the requirement for a buffer if driving multiple fans as the 555 Discharge current sinking capability is much higher than that of a typical motherboard chipset fan control output.. I wouldn't expect any issues driving four fans.
Using a PC PSU may be problematic unless you provide a load on the 5V and 3.3V rails. Its the same problem as using one for a bench PSU - you have to add load resistors which get hot unless fan cooled. Also it will have yet another noisy fan.