That is the whole point of this though, there are so many reviews and benchmarks of modern multithreaded games with the new processors, but nobody tests dx9
You should definitely try xmp but maybe run slightly lower frequencies to make it stable, on my 4770k@4.4 I gained significant performance in raceroom when i upgraded from 1600cl9 to 2400cl11. Had roughly 280 fps in this benchmark with that combo! You have to keep in mind that the memory timing settings are given in # of clock cycles, so 2400cl11 actually has significantly less latency, as the clock cycles are much shorter.
I have tried 1866/2000/2133 and 2400 with different timings and using XMP although I can gain a bit of synthetic performance real world isn't smooth as 1600, it's weird, It's Patriot Viper Black Mamba.
Ok rematch, did a few tweaks, now I'm off to bed, if it crashes I blame you lol ;-) CPU: Intel i7 4790k CPU OC: 4.4 GHz all cores Memory: 16GB DDR3-2400MHz Memory timings: CL10-12-12-30-2T GPU: Nvidia Geforce GTX1080
Grrr 4.5Ghz, I've peaked.....for now.... CPU: Intel i7 4790k CPU OC: 4.5 GHz all cores Memory: 16GB DDR3-2400MHz Memory timings: CL10-12-12-30-2T GPU: Nvidia Geforce GTX1080
I think my main problem are the temperatures, I'm using the stock Wraith Prism cooler which cannot keep the CPU temps under 70°C during heavier load and throttling kicks in, I also had to turn PBO off to be on the safe side. I already ordered a Dark Rock 4 (the Pro wouldn't fit in my case unfortunately), I will repeat the benchmark when it arrives. About the power plans, for me it's the exact opposite, if I choose the Windows balanced plan, the idle voltage stays around 1.4V and idle temps are relatively high, but with the Ryzen balanced plan the idle voltage is <1V as it should be. I'm using the latest AMD chipset driver (07.07.).
Lol.. my I7 8700k would love to be under the 70c. I have a water cooled one but it runs very hot. 85/90c is no exception.
According to AMD, you are supposed to be using the Ryzen balanced plan with the Zen 2 chips, not the standard Windows balanced plan. With Zen/Zen+, it was the opposite.
One thing you can try is undervolting, but you have to be careful, as undervolting with the dynamic clockspeeds does not make the system unstable when you go too far, it just starts throttling performance. I could achieve -0.09v and lowered load temps from about 70c to 65c in stress tests with the dark rock pro 4 Edit: great timing from GN
I've underclocked my System also. Stock is 1,4V at 5 Ghz and far too hot. With a lot of testing I come to 1.28V. Mainboards give way too much Juice on automatic overclocking.
Oops, that can eventually explain, why my max FPS are low compared to the PCs around mine. I've asked myself the question, what I've did wrong with my machine. My Bad. I will have a look this evening again
@Christian Göpfert showed me that his score was about 5-10% higher when running in fullscreen, got me interested so i did the same, also got this 5-10% increase. Would be interesting to see if people with lower than expected scores gain more from running in fullscreen, could be weird Windows issues
The full screen thing isn't available for VR users anyway, so we shouldn't expect gains. For standard display though grab the free performance ;-) NOTE: in game, not for the benchmarks of this thread ofc!
Ladies and gentlemen, behold the dinosaur. CPU: Intel i5-2500 (3.3 Ghz) CPU OC: 3.6 GHz Memory: 8GB DDR3-1333 Mhz, dual Memory timings: CL9-9-9-24 GPU: Nvidia GTX1050 Spoiler: CPU-Z Memory Spoiler: CPU itself