The VRM, or "voltage regulation module" for the CPU is comprised of a control and monitoring logic IC (or several), mosfets, chokes, and capacitors, all configured into a PWM buck regulation circuit. The "module" is integrated into the motherboard and converts 12V supplied by the PSU down to ~1.3V for the CPU.
A CPU with a TDP of ~100W, that runs at 1.3V, runs ~75 AMPS. If you're familiar with electronics, then you know that in order to carry that sort of current, we need a WIDE low resistance pipeline. VRM design can vary quite a bit from board to board, but in order to meet the minimum socket specification for power delivery (voltage range, current, voltage ripple, response, and tolerances etc), all CPU voltage regulation modules on modern performance class motherboards for desktop and server applications are made-up of multiple phases of components that work together to supply the required power. Everything from ~3 - 16 phases or more can be found on modern desktop motherboards. Some use phase doubling techniques, some use traditional mosfets, some use lowRDS-on mosfets, some use multi-FET integrated packages.
Your motherboard shares it's VRM design with the 970DE3/U3S3 and 970 Pro2 boards from ASRock. It uses 5 phases, split into 2 separate power planes, 4 phases for the CPU proper, and 1 for the CPU northbridge, which run on independent power planes at different voltages. Each CPU power phase appears to leverage 2 high side and 1 low side mosfet. I do not know if they are traditional mosfets or low-RDS-on mosfets.
When dealing with such low voltage, high current loads, a PWM buck regulation circuit, even made from very low resistance parts, will wind up with ~10-20% efficiency loss. When the CPU is dissipating ~100W, the CPU VRMs are dissipating an additional ~10-20W. That power is dissipated in the form of thermal energy from the VRM components (primarily the mosfets) to their surroundings (the motherboard, and ultimately, to air). The thermals dissipated from the VRMs is not an insignificant issue. Motherboard temps in weak designs working hard can reach ~100C or higher, which has the negative consequence of causing the resistance of the VRMs to rise, which further reduces efficiency. (this can become a vicious cycle that would escalate out of control if not for self-preservation features built into the board, that force the CPU into low power states when the temperature reaches a certain threshold).
In nicer motherboard designs, the load of the CPU is split up among more components and/or better quality components with lower resistance, and then those components are attached to a heatsink to further improve the headroom of the VRM design. The VRM design on your board, combined with the lack of additional heatsinking, is not particularly well suited to the high power demands of the AM3+ socket to begin with. As madmatt explained already, using an aftermarket CPU cooler often winds up reducing the air flow over VRM's, which actually reduces the power capacity of the system.
I do not advise following madmatt's advise concerning the use of a 120mm fan attached to the stock AMD heat-sink. That will not work very well at all as the 120mm fan will actually produce less air velocity within the heatsink itself than the stock fan. With the stock heatsink and fan combo already being borderline for the FX-6300, attempting to use a 120mm fan with that heatsink would just cause CPU overheating problems. (the "exception" to this, would be if you were using a very high rpm 120mm fan, like a delta fan normally specified for use in enterprise cooling applications).