It isn't possible to increase the total amount of power available to the system - that is a hard limit set by the external AC adapter, the saturation flux density of the inductors in the power supply circuitry, maximum current rating of the switching transistors, etc.
What can be done is to lower the safety margin - i.e. increase the threshold at which the power supply circuitry on the IO will sense the conditions under which it will shutdown. The shutdown capability is a protective mechanism that prevents destruction of the limiting components in a switching power supply, and helps prevent a worst case situation that could result in overheating, thermal runaway, followed by a fire.
It so happens that one of the sense conditions is 'fuzzy' - the voltage across the current sense resistor R302 being 100 mV +/- 20 %. The unit could therefore shut down at as low as 80 mV, or at 0.08 / 0.015 ~ 5.33 A, or at as high as 0.12/0.015 A = 8A. The lower value of 5.33 A could easily be exceeded in practice by a K6-III at a high enough clock or at the high end of its Icc specifications, although the power supply could in theory deliver as much as 8 A. Upon experimenting with the current limits, it was found (by Turbo3) that the original limit could be increased safely by about 25%, from about 0.1/0.15 ~ 6.66A to 0.1/0.12 ~ 8.3A, by substituting a .012 ohm current sense resistor instead of the original 0.015 ohm resistor at R302. This effect can be achieved by removing and soldering another resistor (difficult) or soldering a resistor piggy-back above R302 (easier).
If you short out R302, you are effectively removing the current limit, so there's no safety mechanism to shut it down in the event of a short. The power supply circuitry may easily burn out in the event of a short, when the hard physical limits on the components are exceeded. The first candidates to be destroyed are the switching transistors followed by the inductors.
