FX-6300 VRM throttle?

y3llowduk · March 28, 2020

Think i've pretty well concluded this one but any additional thoughts/comments? Areas outside of the red bars is where i've added a little 80mm fan blowing onto the VRM heatsink! (what I assume is the VRM heatsink, right next to CPU?) This is at stock clocks and voltage. Big BeQuiet cooler with CPU barely hitting 45c under load. The VRM heatsink isn't burning hot to the touch under load, i mean it's hot but can be touched.

The situation is even worse at the overclock I was running (4.7GHz @ 1.44v) spiking constantly down to 3.0Ghz. Was wondering why some games were stuttering so bad...

GA-970-Gaming motherboard

y3llowduk · March 28, 2020

Correction, board is a GA-990x

SpaceGhostC2C · March 28, 2020

Likely. Or a combination of things, like not all cores being equally "good" in terms of hitting clocks at a given voltage.

What are you plotting, though? When idle, clocks should fall, and when doing light tasks clocks should be all over the place as individual cores rest while others ramp up or even boost to "turbo" speeds.

You wouldn't be the first facing "power throttling" on AM3+. Your motherboard should be good enough for a stock 6300, though. Still, it could have very conservative power limits set in its bios by default. Active cooling on your VRM is a good course of action, although heatsinks (unless terribly interfaced) would become too hot to touch it VRM temps are an issue. Won't hurt in any case. Other than that, check all power / current controls in your BIOS, LLC, etc, as those will control how much power is allow to flow to the CPU, how stable it would be, and which limits to expand / not enforce. The harder you push, the warmer it will get. While your motherboard isn't the worst, it's not one of the beefiest, "FX-9000-ready" boards. Do not be this guy:

Spoiler

While tuning it, you can run OCCT benchmarks to push all cores and monitor both temps and clocks, to see if the dips are gone. Be sure to stop any test if temps get too warm. If you are lucky, your motherboard will have a VRM temp sensor; if not, you'll be only able to monitor CPU package and CPU socket temps (the latter is typically 10-15C higher than the former. Too high socket temps, or too big a difference between the too, can be an indication of VRM heating up and warming the socket area as a result).

27 minutes ago, y3llowduk said:

i've added a little 80mm fan blowing onto the VRM heatsink! (what I assume is the VRM heatsink, right next to CPU?) T

You have two heatsinks next to the CPU: the rectangular one between your I/O ports and the CPU is the VRM heatsink. The square-ish one between your CPU and the first PCIe slot (GPU) is the North Bridge heatsink.