Inquiry about Asrock N68C-S UCC

Personally, I don't believe it is simply the chipset the board uses that makes it unusable with an FX-9590 (as long as it is compatible with a specific processor.)

But there are reasons why in general that cheaper chipsets tend to be used in boards that are not compatible with the FX-9590.

Much more important, as Wardog pointed out, is the rest of the build quality of the board. Particularly the CPU VRM stage, and the AMD TDP rating that the board (VRM stage) supports.

Yes, boards that don't have the 990 chipset tend to be cheaper, and one of the main areas where they are cheaper is the CPU VRM stage.

The VRM transistors, inductors, capacitors, and heatsink are relatively costly as components, and also add to the assembly cost of a board.

The two ASRock 970 chipset boards that ASRock states are compatible with the FX-9590/220W TDP processors are the Fatal1ty 970 Performance and Fatal1ty 970 Performance/3.1. I am fortunate enough to have one of the latter model. It's the board I used in the ASRock AMD board contest.

Both of these boards have an 8 + 2 VRM stage.

The X + Y VRM stage normally means X phases for the processor, and Y phases for the memory.

A "phase" is generally one MOSFET power transistor. Think of each phase like one hose spout, capable of supply a maximum amount of power, or volume of water.

Depending on how much power the processor is using/drawing, various numbers of phases can be active or inactive. That is an efficiency and power saving aspect of the design.

Some mother board manufactures at one time even allowed the user to select the maximum number of phases that would be used (if that was really true) or showed the user how many phases were in use.

Beyond that we get into was is called "split phases", where two MOSFET power transistors are combined to actually create one "phase", but all the transistors are counted in the phase count.

So an eight phase design could be eight MOSFETs in four split phases. Some people consider that cheating and a lesser quality design. It might be from an efficiency standpoint, but as usual this description is simplified.

All VRM transistors are not the same regarding their power capability. That spec is generally ignored when PC builders discuss this, except to quote a part number and the manufacture of the power transistor.

For an engineer designing a random circuit, why use an expensive 100W power transistor when a 10W or 1W capability transistor is all that is needed for an application? Transistor manufactures have hundreds or thousands of different models for various uses. One of the main differences in their specs is power output capability.

So we can easily realize, without knowing the specs of the VRM power transistors in use, how do we judge which design is "better" from a power output aspect? Or even better, how would we measure or verify the output power of a VRM design? Who does that? No one besides the design engineers.

Regardless, the more phases/hose spouts available, generally the more power/water volume is available.

Back to chipsets, a quick review of ASRock 990 and 970 chipset boards:

The 990FX Extreme3 board has a 4 + 2 VRM stage, and is not rated for 220W processors.

The 990FX Extreme6 board has an 8 + 2 VRM stage, and is rated for 220W processors.

The Fatal1ty 990FX Killer boards (two) have an 8 + 2 VRM stage, and are NOT rated for 220W processors.

The 990FX Extreme9 board has a 12 + 2 VRM stage, is rated for, and is called the best choice for use with 220W processors.

Besides the two Fatal1ty 970 boards, the other four 970 chipset boards do NOT support 220W processors.

For the 990 chipset, it is two boards out of five total that are compatible with 220W processors.

For the 970 chipset, it is two boards out of six total that are compatible.

Also notice that two boards with 8 + 2 phase VRMs are not FX-9590 compatible.

I'm still surprised and impressed how relatively cheap the AMD 990 and 970 boards are compared to Intel chipset boards. Note that high end Intel Z chipset boards commonly have 12 + 2 phase VRM designs, for sub-100W TDP processors.

In my Fatal1ty 970 Performance/3.1 board build, I took precautions to insure the VRM heatsink had a fan moving air directly across the VRM heatsink. The heatsink temperature was not greater than ~110º F. But do I know if that board would be fine long term with that processor if the VRM heatsink was not actively cooled? No, but chances are it will work much longer than a sub-eight phase VRM design without a heatsink.

Wait... what is the point being debated?

Bottom line: Simply because an AMD processor is compatible physically with the socket used in a mother board, does NOT mean the board is up to the task of providing the processor with the power it will need in all circumstances.