I figured out what was going on with NVMe drive. I reinstalled Samsung Magician, ran the test in it and IOPS went wayyy down from 300 000 to 100 000. Played a bit with configuration. Then - and for the very first time - i agree with people on this forum who claim that HPET may cause trouble. So used bcdedit, disabled usedplatformclock and usedplatformtick. Samsung Magician then reported 300 000+ IOPS.
that's much better... guess I need to check mine too, it maybe getting hit some.... but still I don't think mine will ever keep up with yours
I re-tested the power settings towards Kombustor once again. It seems that for 1280x720 was optimal setting -19%. For 2560x1440 its -8%. Here isnt the rise of the FPS so significant, just about 4-5 FPS difference.
However... in both cases it seemed to be CPU bound, as the test is single threaded, and GPU did not went to 1050MHz...
Basically it seems like OS power and performance management kicked in, limited CPU utilization, limited GPU utilization in order to reduce overall FPS. I wonder if something similar can be tested on a Nvidia GPU, because according to several tests, it seems that Nvidia is getting higher CPU utilization...
I know Navi is getting all of the news and hype right now, but I never got around to posting my Fury X stock RE2 benchmarks here so I guess I'll do it now. I ran some benchmarks and posted them on Reddit and made a YouTube showing the performance. This Fury X is still amazing me nearly 4 years after release. With Navi on the way I MIGHT finally bite the bullet and upgrade. Even with the 4GB HBM limitation the card still performs well for the games I play. Put a ton of times into benchmarking games over the past 4 years or so with this GPU and it has been wonderful. Nvidia was an option, but I don't think I want to pay their current prices mostly because I see more longterm value in AMD GPUs at the moment.
I don't know what AMD\ATI did with the Fury X and the drivers, but this thing has been a BLAST at 4K over the past few years.
Due to my 4GB buffer limitations I had ReLive downscale the recording to 1080p, but I kept the bitrate high enough not to effect my FPS that much. I CAN max the game, but Shadows and Mesh needs to be set to High instead of Max due to micro stutter at high resolutions. However, I did NOT notice a difference in image quality even after playing the game on a MAX with a Vega 64.
My old website is down now, but I'll get around to making another one and posting all of my Fury X benchmarks I've ran since Day 1 over the years.