My current build is in a Thermaltake WP100 with 13 Noctua Industrial 140mm 3k fans(Excessive I know) which idle roughly at 40%. My cooler is a Noctua D-15 w/ 2 Corsair HD140's at max rpm. Z390 Dark.
CPU/RAM: i9-9900k @ 5Ghz override vcore 1.27v -25% vdroop, Ring ratio 46, Avx 3, PWM frequency 800. Ram g.skill royal 4600mhz xmp1 18-22-22-42-1 @1.45v.
Storage: Intel Optane SSD 905p 960gb
GPU: Sapphire R9 Fury Tri-OC
PSU: Asus Thor 1200w
 
The other day I was stability testing with a mix of Aida64 and Realbench. I ran Aida64 w/ stress cpu, fpu, cache and memory enabled for an hour. Then cinebench 3 times, then realbench for half an hour. My max temps across all cores never went past 78c-81c only hitting 81 when the room's ambient reached 70f. So I decided to try and find a more efficient way of fan orientation doing 15 minute pulls with Aida64, trying different brand fans, different locations among the heatsinks. upon the 4th 15 minute pull after 9 minutes Aida64 recorded a hardware failure and stopped the stress test. And ever since then I can hear a low frequency humm coming from around the socket.
 
I tried another bios on default settings, the humm was gone until the cpu went under load. I used an automotive stethoscope to rule out all of the surrounding components including all of the fans in the vicinity.
I've read the hardware failure is just a math error typically with the cache and I know coil whine isn't indicative of an actual problem but I believe it is coming from somewhere in or next to the VRM.
 
I have bad hearing but it is quite noticeable when the computer is at idle. Has anyone experienced anything like this?