Originally Posted by miklkit
Different strokes for different folks.
One guy used zip ties. That is at least as professional as anything Gigabyte did with that board.
Two guys cut the mounting brackets and epoxied or glued them on to the long base.
One guy made new metal brackets and bolted them down.
I used no.4 nuts and bolts and a combination of no. 4 and no. 6 washers. I used nylock nuts and nylon no. 6 washers so as to not tighten the bolts too tight and crush the VRMs and also to not short out the board on metal washers. These are all available at your local hardware store for $2.
I've posted these here before but maybe it will be worthwhile to do so again. To those who say it is not a VRM issue, look at that I/O panel.
Thanks for the info!!
While I was waiting to hear back, I wanted to put my computer back together and use it. I decided to zip tie one of the 40mm Fractal Design fans to the top of the default VRM heatsink and re-install it. I have a temperature probe from my fan controller slipped in between the top end of the VRM heatsink fins (the end which says "Ultra Durable"). Anyway, I got it back up and running and decided to run Prime95 Blend and Small FFT tests.
At first I thought the temps were askew (too low), and that maybe the air from the 40mm fan was affecting the temperature probe. But I was monitoring with HWiNFO64, and saw my CPU temp was lower than
prior to adding the 40mm fan. Before, the VRM was reaching 78C about 11 minutes into Prime 95 Small FFT, and CPU temps were about 57 C.
Now with the 40mm fan, running Prime95 Small FFT, about 11 minutes in, VRM temp maxed at 61C, and CPU temps maxed at 53C. I knew if I could cool the VRM a bit, it might affect CPU temps, and it appears it has! I also cool my CPU with a Corsair H80i, where I have its radiator/fans attached to the rear exhaust. Unfortunately, the H80i's innermost fan covers the top of the VRM heatsink, and causes it (I think) to get hotter than if I had a conventional air CPU cooler (hence the aforementioned 78C VRM temp under Prime95 Small FFT).
I tested with HPC disabled, then again with it enabled. Of course, even with LLC = Ultra High and HPC disabled, the CPU was throttling all over the place (like the frequencies competing in an Olympic trampoline contest!). But, with HPC enabled (and the 40mm on the VRM heatsink), I was able to run Prime95 Small FFT for an hour, and not exceed 53C CPU, or 61C VRM on a very stable 4.5GHz OC.
So bottomline, cooling the VRM does make a difference on the Rev 3.0 GA-990FXA-UD3. Know I just have to mull over what approach I'll take in getting the Thermalright HR-09U afixed to the mobo. I'm hoping once I do get it attached, the results will be even better than simpler attaching a 40mm fan to the top of the default VRM heatsink (although that was a nice improvement).
Thanks again for the info!