Cyran said: As a 3090 owner this a bit scary |
Certainly does remind me of the Fermi days. While the Nvidia driver measuring averages is quite nonsense, it should also be up to the AIBs to choose their VRMs and set their own trigger point for OCP which really should be the ultimate failsafe. There are VRMs that are certainly designed to handle such loads for long term but there are certain VRMs that are also not designed to either and I think that + shitty OCP limits + driver doing averages is where we see some 3090s get affected more so than others.
I think it all stems from Nvidia not giving AIBs enough time to make proper coolers back during Ampere launch. I remember there were a few articles about it and it was one of the reasons why Asus launched the Strix later than Evga's FTW3 and such. It will be interesting to see the longevity of these cards but I think something like a Strix or Kingpin that have high quality VRMs/PCB designs will last a long time while some like FTW3 or Gigabyte ones are gonna continue to get rekt.
Of course as next generation comes with 500 Watt Lovelace and 400 Watt RDNA 3... Things gonna get even more hectic.
PC Specs: CPU: 7800X3D || GPU: Strix 4090 || RAM: 32GB DDR5 6000 || Main SSD: WD 2TB SN850