I switched to Team Red because the i9-9900KS was overheating its VRMs.
I switched to Team Red because the i9-9900KS was overheating its VRMs.
i9-9900KS is running stock with a minor undervoltage, visible in the VID versus VR Vout. It handles long FFmpeg x265 encoding smoothly, maintaining a steady 120-130c depending on ambient temperature. The CPU throttles to around 4ghz, which is frustrating, as it briefly spikes to 5ghz until the VRM reaches 110c before dropping back down. It never actually gets hot enough to trigger throttling, with the CPU staying under 115c. The system is powered by an ASRock Z390 Phantom Gaming 9, featuring 32GB DDR4-3200 memory at 4GHz.
The simplest method is to install a fan on the VRMs. I added an 80mm Arctic P8 to my VRMs after installing a GPU, which reduced airflow in the case and caused the VRMs to reach 120°C during tests or even caused crashes. Once I added a fan, the temperature dropped to around 89°C. Would trying a fan on the VRMs help?
2-3c drop and that's roughly all of it. I even took off the plastic cover around the IO shield in the hope it would assist, but it only made things take a bit longer before hitting those temperatures. The VRM is definitely running hot—just touched one of the heatsinks for a tiny moment and got burned, which made it tingle for several hours afterward!
I was in my case yesterday tapping the VRMs and it seems to be working fine. It looks like 1.4v might be too high for those VRMs, which doesn’t make sense. I’m pumping the same current through my ASRock B450 without any problems for nearly a year now. I just checked a photo of your board—was it a heat sink or just a shroud? I’d guess it’s a heat sink. What’s your room temperature? It’s around 20°C. Also, someone with more Intel knowledge could give me a better idea—this voltage seems quite high based on what I’ve read (the chip draws 272W before throttling).
93c fits the scenario. Likely a sensor positioned between power stages or MOSFETs. 127c appears to be an outdated value, acting as a static default that remains unchanged. The software might be interpreting the wrong sensor or the motherboard isn’t updating it properly. It represents the highest possible byte value, disregarding the sign—0b01111111 or 0x7F. Possibly it reflects an internal reading from inside a power stage or MOSFET, matching around 95-100°C on the chip surface outside packaging. Simple checks: touch your finger, place it on a heatsink that stays cool for 50-60°C for a few seconds, or hold it until it feels hot (over 80°C). The CPU VRM can function at up to 100°C for extended periods, but it will slow down or shut down if temperatures exceed 105-110°C. Chips under the heatsink are rated for at least 125°C, though most boards try to keep temps below 100°C for long durations. Prolonged exposure beyond this can cause permanent damage due to material degradation.
When idle it maintains around 1.3 volts, dropping to roughly 1.2 to 1.25 volts under load. The power reading isn't reliable because it relies on VID, which is affected by undervoltage. It actually operates at about 221 watts from the VR-Vout, with more precise voltage and power data coming from the VR side. It drops noticeably within half a second before performance issues start and persists for several hours afterward. I prefer not to repeat this situation. At idle it runs between 50-60°C, about 75°C on lighter games, sometimes exceeding 85°C during intense titles like CoD MW, and it spikes above 100 when full CPU usage (such as ffmpeg encoding) is applied.
I have heard that \loop 1 and loop 2 refers to the two banks of VRM next to the socket, so try move the fan around. Also heard of one being placed on the back of the board, which is much harder to verify. but if you take VR Loop 2 as the proper measurement, then it's just hot but not desperate.
Two AUXTIN sensors stopped working, so I couldn't read them. The data comes directly from the VRM. Right now it's hitting 134°C, while my ambient is around 24°C. That's cooler than usual today.