ASRock Fatality 970 Performance

So glad you asked about this.

But please define a "review" for me? Is it an article published in a PC hardware website, or the comments made by a random person on a retailer's product page for selling a mother board?

The short answer for you is NO, the ASRock Fatality 970 Performance board will not fail or "fry" if you are using an AMD processor whose TDP is 125W.

Compared with ALL AMD AM3+ boards made by ALL manufactures, the ASRock Fatality 970 Performance  is absolutely one of the better AM3+ boards (IMO top 20%), and can even survive using an FX-9590 at stock speeds.

That is, if the user of an FX-9590 in this board provides the appropriate cooling for the CPU and just as important, cooling the CPU VRM/power components on this board. If BOTH of these things are not provided on ANY mother board using an FX-9590, then the chances of the board failing are at best increased, and at worst guaranteed.

I have some experience with this, since I have this board and an FX-9590. This board is one of a small handful of boards using the 970 chipset that has a chance of using an FX-9000 series processor. That is not the fault of the 970 chipset itself, but of the build quality that comes with the price point of a budget priced chipset like the 970. The ASRock Fatality 970 Performance is one of the very rare over-built 970 chipset boards.

Regrettably, AMD made the FX-9000 series processors physically compatible with any AM3+ socket board. Add to that the appropriate CPU microcode in the BIOS, and it will work in a $50 board. In reality the FX-9000 series processors should not be used in a sub-$200 MSRP mother board. Or at least not by the plug and play type of user that does not provide the requirements for cooling such a system.

AMD does have one page of information about the special requirements of the FX-9000 series, and the boards they recommend for use with those processors:

http://support.amd.com/en-us/search/faq/295

AMD is recently touting their new Wraith cooler for the FX series processors. It is an improvement over their earlier designs, but not for use with an FX-9000 series CPU.

It handles my CPU, that's TDP is 125w, with no problems.

If you are going to use a liquid cooler for the CPU, you may want to consider using an extra fan to cool the VRM section, as it's not a true 8+2 power phase design, but a doubled 4+1. The socket may also end up running hot. Note that ASRock states in the CPU support list that you should install a CPU cooler with a top-down blowing design when using a 125 W TDP FX-8120.

PetrolHead wrote: If you are going to use a liquid cooler for the CPU, you may want to consider using an extra fan to cool the VRM section, as it's not a true 8+2 power phase design, but a doubled 4+1. The socket may also end up running hot. Note that ASRock states in the CPU support list that you should install a CPU cooler with a top-down blowing design when using a 125 W TDP FX-8120.

Extra cooling for the VRM stage when using any liquid/radiator based CPU cooler is a MUST if you are over clocking or operating your CPU at high loads for significant lengths of time. That is, if you care about the temperature of the CPU VRM stage.

One advantage of having more phases in a CPU VRM stage is efficiency of power usage. Fewer phases will need to be active at lower power draw levels of the CPU. But simply having a phase (device) active does not mean the CPU is using all the power the device is capable of providing.

Increasing the number of phases in a CPU VRM stage is not without its own issues. A phase/device that is inactive takes time to become active. The ever changing power needs of a CPU occur very quickly, and shutting down a VRM phase as soon as the power demand is below its active range is not done. If the power demands of the CPU cannot be provided immediately, it will throttle or have a computational error. The more phases in a VRM stage, the better its parts and design must be. That translates into more $$.

While the VRM stage we are discussing is said to be a 4 + 1, and not a true 8 + 2, that does NOT mean it is less capable of supplying power to a CPU. It is simply switching two devices/phases on or off at a time, instead of one. The number of power devices is the SAME in both designs.

Given the range of power usage of a processor, it may not even be optimal to have say ten or more phases that allow for one or two to be active when the CPU is at idle. The minimum number that is required to be active may be two or four.

As is always the case, discussing the number of phases without considering the actual parts being used is simplified beyond having any significance. No review has ever truly tested a CPU VRM stage for power output capability over time. It would be interesting to see this but I won't hold my breath waiting.

Yes and no. Cooler VRM or any component is a good thing anyhow. But if the VRM circuit and its heatsink is designed well, the proper CPU air-cooler will suffice.