sharkfin61 Posted June 1, 2020 Share Posted June 1, 2020 Bit what is the reason for the gap in fps if cpu and memory are the main difference? Gesendet von iPad mit Tapatalk sharkfin out! Support your local AirTransportWing ! Roal Bavarian Airforce all the way NEW RIG: RYZEN 7 5800X3D~ AORUS GTX 1080Ti ~ AORUS X570S Elite AX ~64 GB Corsair Venegance DDR-4 3600 ~ BeQuiet AIO Silent loop 2 360 watercooled ~ Samsung 890 Pro M.2 (2TB) + 870 EVO (1TB) SSD ~ WIN 10 64-bit ~ AOC 31.5" Gaming 144Hz Display ~ DelanClip@TrackIR 5 ~ TM Warthog no.2 ~Saitek rudder pedals~ 2 TM MFDs on 2nd 27"display ~ Buddyfox A-10 UFC ~ CDU/AAP panel via DCSBios ~ ARC-210 (soldering WIP) Link to comment Share on other sites More sharing options...
Wali763 Posted June 1, 2020 Share Posted June 1, 2020 (edited) Bit what is the reason for the gap in fps if cpu and memory are the main difference? Gesendet von iPad mit Tapatalk Not an expert, but if the GPU is maxed out, overall it dont matter if the CPU just barley can deliver the required data or is crusiing along at 15% load and could deliver data for much more frames. Thats what a bottleneck is all about. At least, as I understand it. I did the test with my current OC (right now 5,1Ghz) and also with stock 4,3 Ghz (if all cores are used). The results were exactly the same. Since it was not mentioned, I turned off TrackIR before loading, so I had a straight look forward, when the Briefing came. FPS are from Afterburner and I also checked the fps when my GPU was overclocked (111% Powerlimit, +100Mhz GPU, +1000Mhz Vram). 1080p, high presets, 4xMSAA Hornet: stock 78fps/OC 85fps Su-25T: stock 94fps/OC 104fps At first I only had 95-95% GPU-load, but after hitting Alt-Enter (which I thought was not needed anymore for a long time) it got the wanted 99%. Edited June 1, 2020 by Wali763 Link to comment Share on other sites More sharing options...
Vertigo72 Posted June 9, 2020 Share Posted June 9, 2020 GPU (and CPU) load really arent very meaningful numbers. They only measure one aspect of the card and modern cards adjust their clocks dynamically so the load tells us even less. Being pegged at 100% doesnt mean the GPU is "the" bottleneck (and seeing CPU load at say 20% says more about how many cores your cpu has than what is bottlenecking it!). The tests posted here really dont mean a lot either when so many different parameters are being changed between them (CPU, GPU, Ram speed, ...). Just because MSAA is set to 4x doesnt mean the framerate will depend solely or even mostly on the GPU. Despite what many people say, based on my own testing, I found DCS really isnt very GPU bound with "normal" settings (and no VR at 200% SS). Performance is bottlenecked primarily by CPU and Ram speed. As you increase eye candy, the GPU may further reduce that, and become a bottleneck in its own right, but even with low resolution and no MSAA or SS, I doubt anyone here gets >2-300FPS (just guessing, havent actually tested yet). Link to comment Share on other sites More sharing options...
Recommended Posts