Hi Guys, Since Ari been advising me to test his DX11 SLI bits, I might as well share the result for those who may need it. This post does not belong in "DX9 vs DX11 performance difference" thread as there is no official support for SLI in DX11 by S397...yet. PARAMETERS : - Silverstone GT track. Race mode, not practice. - Mostly cloudy, to tax the GPU more. Members reported cloudy sky reduce FPS, so cloudy it is then. - GTR500 2013 , this is a more GPU intensive car, compared say to Clio - 20 AI cars, plus 1 of mine = 21 cars. Mine is AI driven for this test. I purposely make my car to be at least 13th - 19th position by starting a few seconds late, so that more cars in front of me, to tax the GPU more. Members reported more cars in view reduce FPS, so I try my best to create the same scenario. - All setting at full/max including shadows, but Anti Aliasing at Level 4, so that the same test can be used for DX9 vs D11 - No Post Effect. Post Effect is no use for now , as a 100 FPS near consistent can't be achieved yet , if at max/full setting and 7680x1440 resolution, car is GTR500 2013 and track is Silverstone GT. - Resolution is triple 2560x1440 monitors or 7680x1440 in multiview angled. - No in game Sync used, I rely on Nvidia G-Sync. - FXAA is off Today's test using SLI bits 0x024020F1 from Ari is promising, it has headroom for further development when S397 works on SLI and submit it to Nvidia for suitable SLI bits. No micro stuttering, but FPS is still too low to be visually pleasing like how a 100-120 FPS would deliver to a G-sync monitor. Nevertheless it is a good improvement considering the GPU loads are down but FPS goes up a bit with SLI bits 0x024020F1 from Ari, compared to 90-100% GPU loads of DX11 bits all zeros / none which is the default by Nvidia for Rfactor 2 in DX11. 7 July test, DX11 1941092 20 July test, DX11 1944770 I am running i7 5960X 8 core at only 3.5Ghz, using Asus own mild OC at BIOS. I think an Intel 4 core CPU with higher clock speed would do better. End of Test.
