Just tested my 3900X after a ram upgrade to 32gb. The Cpu is running stock speed and the ram is OCd to 3400mhz from originally 3200mhz
TL/DW: UB now (well since Zen 2 came out) heavily skews towards Intel processors and gives them much higher scores even if they are as fast in basically any other workload, this is also reflected by the "reviews" on the respective product pages (11900K = 80% talking about "unfair practices" and AMD)
I see there are one or two other Ryzen 5800x posts, but thought I'd post mine so any one else searching might see the variance. (I saw a 480 MBeams?) I undervolt mine by .150 so it runs cooler - that probably affects boost a bit. PBO is capable of 4.85ghz, mine boosts at 4.75 and usually runs 5-8C cooler. I can't tell the difference when using apps/games. Ryzen 7 5800x @ 3.8ghz (Boost at 4.75ghz) 3600mhz CL16 32gb DDR4 (Seems bananna tries to add 200mhz in it's output) ][ v6 | 0.6.5.4 (64 Bit) built Mon Mar 29 15:29:48 2021 | now: Saturday Sat Apr 10 15:51:25 2021 ][ Windows 10.0 (build 19042), 64-bit | RAM: 31.94 GB | Toolset: 19.27.29114 / 192729114 ][ AMD Ryzen 7 5800X 8-Core Processor | 16 cores | 3800 MHz +----------+----------+ | MBeams/s | % Realt | ---+----------+----------+ 1 | 77.528 | 845.64 | 2 | 146.559 | 799.30 | 3 | 181.052 | 658.28 | 4 | 245.992 | 670.79 | 5 | 289.621 | 631.81 | 6 | 345.363 | 627.84 | 7 | 373.251 | 581.60 | 8 | 288.252 | 393.01 | 9 | 302.552 | 366.68 | 10 | 329.204 | 359.08 | 11 | 343.762 | 340.87 | 12 | 364.584 | 331.39 | 13 | 385.664 | 323.59 | 14 | 411.097 | 320.29 | 15 | 429.796 | 312.53 | 16 | 433.480 | 295.51 | 17 | 321.378 | 206.20 | 18 | 330.109 | 200.04 | 19 | 334.249 | 191.89 | 20 | 341.822 | 186.42 | Max Mbeams/s: 433.480 Mbeams/s BANANAA!!! .______,# \ -----'/ `-----'
All benchmarking software will favor a CPU architecture more or less, as all CPU's do the same things, just in different ways, and different ways are better at different things. The important part is whether the benchmark manages to accurately depict loads that occur in normal gameplay, video rendering, or other tasks, as depending on the product, the activity that stresses it will be different I personally think this guy loves to hate Userbenchmark a little too much. For example, at 8:37, he complains that they praise Intel's iGPU for their encoding performance. Despite the fact that I do now have a dedicated GPU, I think it's reasonable to expect a high quality iGPU in a mid-end processor, and this is what they are pointing out. Nothing's perfect, but this guy is all-too-rough on it. I do love the fact that you changed DL to DW for "didn't watch", I know it's even further off topic, but I appreciate that attention to detail Spoiler I feel a need to point out that while I have an Intel CPU, it's because it was given to me (free!). I like AMD better as a company, particularly because they had the guts to put out a sixty four core CPU with one hundred twenty eight threads.
Man it is tempting to liquid cool this thing to get more out of it. Sitting at 495 with a stock 5900x at the moment. RAM is very slightly tuned 3600c16. I'm seeing about 3.9Ghz single-core and only 3.1Ghz once it hits 20 cars. Wondering if it's just HWinfo but I know this CPU can do 4.9Ghz single-core and ~4.4Ghz all-core.
The issue is that their ranking (and with that also the benchmarks) is totally skewed towards Intel, there is no way a 11400F is overall faster than a 5950X, yet it is in UB's ranking. Anyway, back to something more on topic, here is my laptop beating my desktop with the Banana. The laptop has a 4700U in a 13'' HP Envy x360 with 3200-CL22 memory with the performance mode enabled, so the laptop is running with 30W short term and 25W long term, the 1600 is running stock with 2666-CL18 memory.
When using a CPU with more than 10 cores - 20 threads, you will have a better score if you increase the 20 cars default limit as you get best results with 1 car - 1 thread. So in my case I have 12 cores 24 threads and get the max score at around 24. You can do it by editing the file here : Code: (YourSteamLib)\steamapps\common\BeamNG.drive\lua\console\bananabench.lua change the 7th line to what you need. Code: local defaultVehicleCount = 40 save the file and run the bench again. It will take longer obviously
Can someone help troubleshoot this? I've generally been having performance decline over the past few weeks, unfortunately I don't have a previous bench but comparing my R1700 to posts here it seems to be plenty powerful. rx5700 on the GPU side, it has never given me problems and I'm really hoping it's not the cause of my issues. My normal FPS was 90+ for a long time, since around April/May I've been struggling to get over 60. GPU has not had any decline in clock speed but I've had to cut down my RAM speed and CPU overclock to retain stability. I was used to running 6-7 cars with the auto AI setting and 50-60 FPS. Right now I can spawn 4 or 5 with 35-45 FPS. Is my GPU dying? it seems to hover around 50-70% utilization even with a busy scene and FPS remains low. Seem OK on normal maps, but with AI or a large map the FPS just flops. CPU seems to be powerful enough to run more, and with 16gb of RAM I should be covered. (Cl16 3200mhz ram, currently down at 2667 as it BSODs at faster speeds) So maybe this is a mobo/chipset issue? Really at a loss here... Thank you very much in advance, and I'm sorry for the off topic reply. Hope you're all doing well edit: forgot to mention this is a 3.6ghz all core clock; not the stock boosted 3.6. SMT enabled and no cores disabled
I currently have a similiar issue with my 3900x i cant really spawn more than 4 or 5 cars without loosing fps to a level that is unplayable it seems to me that its a game bug
I know this may be an obvious problem, but have you tried restarting your pc? I had the same problem and that fixed it.
i have had the issue for quite a while and even did hardware changes between it and still have the issue. It worked fine on my i5 6600k but since i jumped to AMD, first with a 1800x and now with a 3900x it wont work
Friend of mine is building a system and waiting on a gpu. In the mean time he's letting me test out the 5800x and b550 combo to see the how it OC's. 1.19v 4.35ghz seems stable on this chip for at least a couple hours. reaches a peak of ~570mBeams/s after several cinebench loops. I'll update more later on.