ACC - 8 CPUs tested on Triple Screens 

Maybe AMS2 deserve a similar video?

I'd like to make one! AMS2 is a bit more challenging for data collection because the replay system has performance flaws. Results from replays are not comparable to real-time gameplay.

Thank you! More to come 😺

My original intent was do to VR analysis with ACC, but I found it very challenging to find graphic settings that worked well. Do you have any recommendations for VR guides to help with setup? I made myself nauseous after a day of testing :(

​@@BenchmarkOdysseys I did play around and found some settings as well as computer config to get best of acc in VR. How can I contact you directly so i could share the screenshots and the google docs with my settings?

@@GUN69_ You can find my email addy under 'About'. Just c&p the text from the ini files and put within email body, gmail may block the file itself. Thank you :)

@@BenchmarkOdysseys ok - FYI I just dropped you the email. I hope it will help you

I'm setting things up now for my next benchmark run, but leaning more towards AMS2!

14700K should deliver slightly better performance to 13700K because the subtle technical changes in specifications (higher boost, 7mb more cache in L2/smart). The uplift should be under 5% though.

Cheers :) I've done some preliminary testing with AMS2, but there are some hurdles that I've yet to overcome. I may have to manually drive against AI for each benchmark run. That will increase the margin of error. I will try and tackle it next month!

Thank you for sharing! I've only done limited testing with Eyefinity on the 6700XT, and it seems to have the same effect as Surround at reducing the spikes.

@@BenchmarkOdysseys worth noting that even on a single screen I can't get ACC to run completely smooth unless I run it fullscreen mode, in windowed it has that mild stutter, without surround or eyefinity you can't run fullscreen. Maybe this is part of the problem?

Interesting. I always run my single resolution tests at Fullscreen. I should specify that in my testing because I know many racers use overlays and opt for Window Boarderless. @@easybreeze1

@@BenchmarkOdysseys I recall now that I believe the issue is that VRR doesn't kick in unless the game is full screen. Personally I can't give up that smoothness although I do miss using overlays. If there is a fix I'd be very happy. I honestly think most just don't realise it isn't as smooth as using fullscreen and engaging VRR

Thanks :) Fair question. I've used the simple CPU-Z bench and Cinebench to compare to other published results. Performance seemed fine. Unfortunately I've run into issues with my MSI AM5 mobo and Gigayte 4080 Super, so I'll probably have to get a another mobo. I will revisit some of these unusual results.

@@BenchmarkOdysseys Looking forward to seeing those results and here's hoping it is the motherboard that's the issue.

All depends on your budget, and other constraints you may have (heat, power, components you are carrying forward, future-proofing, etc). If you're looking for specific advice for your unique situation, you can fire me an email at benchmarkodysseys@gmail.com :)

Thank you :) At 8:40 I briefly chat about iRacing. There is a measurable advantage to Surround vs. regular Triple screens, however, it seems unlikely to be noticeable in real-time racing.

@@BenchmarkOdysseys So you did 👊🏻 thanks

The 4090 is amazingly capable and will push any CPU to its full potential. If you've made it this far with that 12700KF... I'd try to hold out another year for Intel's next gen. Or AMD's 8000 series this summer/fall.

Thank you for the sub and sharing your experience! I thought I was going mad when I got the results... spent half a week redoing all the tests. As far as I know, AMD has not revealed how it's algorithms identify compute operations to run in 3D V-Cache vs. System memory. Maybe some tweaking is necessary by the devs or AMD to prevent the weird disruption to the workflow. TechSpot and Hardware Unboxed have benched ACC with the 5800X3D and 7800X3D at single screen resolutions. They also show the massive lead at 1080p and 1440p to shrink to much less at 4K. With an RTX 4090, however, Techspot, Hardware Unboxed, and Dan Suzuki show examples where the new X3D can 'win' at 4K with ACC. But that card is an outlier, not just in cost, but also with insane technical specifications. Unfortunately their results are misleading to anyone without a 4090... so that's 99% of us!

Video cards are more adaptable to future games because Nvidia or AMD can make optimizations with their drivers. But CPU architecture has to be leveraged by the original programmers when the game or simulator was developed. It's much more difficult (impossible?) for them to modify their game engine for features like Ecores or 3D V-Cache if their code was created before those CPU technologies were available. That is why many sim racers disable Ecores (for better Pcore boosts), and why 3D V-Cache doesn't always produce better results (original code doesn't jive with AMDs algorithms for X3D). And programmers will be reluctant to build game engines around Intel or AMD's unique features because that favoritism will leave half the users in the lurch. That's probably not the answer you were looking for! lol The future in tech is hard to predict, so if you're building a new system today, I'd make sure it's balanced between CPU/GPU so you get the best performance that you can immediately enjoy :)

Triple 1440p is widely considered to be 'the best' configuration because the monitors are much cheaper than 4K, and the decrease in clarity (from 4K) is difficult to notice at 32 or 27inches. We do see a performance gap between the CPUs at Triple 1440p, however, any mid-range CPU from the last 4 years will be able to push an RTX 3080ti to it's full potential. Triple 4K is the most demanding configuration for the video card. My results suggest all CPUs tested are constrained by the GPU at that resolution. The Nvidia RTX 3080ti used in this testing is equivalent to RTX 4070 or Radeon 7900XT [note I have yet to test AMD], and it was able to average +60 FPS with 'good looking' ACC graphics at Triple 4K. These 4K results suggest you will be better served with an 12600K and 4070, rather than 13700K with 4060ti. It's unclear exactly why the 7800X3D under performs at triple screen resolutions. It may be unique to ACC and the testing scenario I developed.

At the graphics settings I used for testing with Spa and 30 cars, there was a 9% advantage to 5800X3D, or up to 19% gain with Surround 1440p when compared to 5600. If you find yourself in multiplayer lobbies with even more cars, that advantage should increase. The question is, do you need or want that extra smoothness, and is it worth the price of the 5800X3D? I expect the 5600 to still deliver +60fps during the intense moments at the start of the race with Triple 1440p and average race FPS of +100 fps. That's pretty good for most drivers!

Thanks :) Glad it's helpful to you. Definitely look into a CPU upgrade. I'd be tempted to drop in an Intel 11700 or 10700 K or KF into that motherboard, but if it's a prebuilt from someone like Dell, that may be difficult.

Right now the plan is to do AMS2 before rF2, but I'll know more in a couple weeks when I start looking at the data :)

@@BenchmarkOdysseys AMS2 would be very welcome as well, thanks. But i'll wait for rF2 data in the future ;)

There certainly is a 'wow' factor with Triple 4K. The clarity of textures and detail within the simulator is impressive. I can only imagine how spectacular triple 43" would look. I think you can find a balance of medium/high settings to achieve 60fps, but you will be limited to the abilities of the 3080ti. In separate testing I gained another 10% performance by increasing GPU and VRAM speed, however, video card power consumption skyrocketed to 400W.

@@BenchmarkOdysseys Appreciate your reply👍

Thank you for sharing your channel :) My first testing of 7800X3D was with Noctua NH-D15, then I tried EK Nucleus 360 AIO when I changed PSU and completed a fresh install of Windows and ACC. 'CPU package' temps were in the mid 60*C range with both coolers. Power consumption rarely peaked above 80W during the benchmark runs. Even though I didn't notice any power or temp issues, I like the idea of trying an undervolt just to see if it influences the results. RAM used was G.SKILL 32GB DDR5 6000 CL30 kit with EXPO enabled.

@@BenchmarkOdysseysThe RAM is nice. Try undervolting, you never know. I did the test on my 5800X3D + 3080 12GB using Surround but the performance was exaclty the same. Something strange with your benchmarks...

Heh! Those are going for about $2,000 USD in my region. That's a steep price to pay for the best! It's most likely that Nvidia launches a 5090 before any other model, maybe even by December 2024.

@@BenchmarkOdysseys when using surround, did you set in game video setting to full size of display and disable triple monitors so that the three are treated as one? I tried this for the improved performance but the side monitors are stretched incorrectly and cannot correct in view menu since triples are disabled in game. I can enable triple monitors in game while surround is enabled in nvidia control center but not sure if this negates the advantage of using surround.

@@M43S7RO You are correct. I used SRWE to force ACC to go triple screens, so even the menu screens are drawn across all three. But I need to run Triple projection to account for the bezels and the angled side monitors. Otherwise the image is distorted. Recorded performance is always better with Surround because the minimum spikes are removed.

@@BenchmarkOdysseys sorry, so you use srwe while surround is enabled so that you can set acc video setting to triple screen? Is so this odd... I can set acc to triples while surround is enabled in nvidia settings and it still works like triples. This is where I'm unclear if acc has to be set to single resolution the size of the surround or if triple screen setting all while surround is enabled is enough to get the performance gain. Sorry for back and forth

@@M43S7RO No prob. ACC doesn't show my Surround 4K resolution in the sim, I use SRWE to make it 11540x2160. Whether I'm doing surround or normal triple screens, I have to select Triple Screens in ACC Options/Video, otherwise I get a 'fisheye' distortion with the side monitors. We need to customize the View Options when in the car to match our personal screen sizes, distance, and angle.

Hrmmm, I doubt it can be done without FSR or DLSS. Maybe a 4090 can push native resolution to peaks of +200fps, but I do not know how stable the framerate will be over duration of a race.

Hey there. I don't understand your question. I have not yet benchmarked ACC and VR because I have been unsatisfied with the experience compared to AC and iRacing.

I have a question just use openxr instead steam vr oculus vr. Max settings with dlss for me 90 fps

Hey there :) I joined an open lobby, and that's the configuration the admin was running on his server.

​@@BenchmarkOdysseys In ACC it's not possible to mix AI and players, I don't really know what you tested here.

Ah. Sorry for the confusion. I just reloaded the replay. Server was [Pommesbude] Spa | Server Ranking & Active Banlist | LFM BoP (RacePlace.de) back in December 1st 2023. 30 drivers. I have a faint memory that someone asked (was it me?) about the grid size, and someone replied that AI will fill it to 30. I guess that was sarcasm or someone else who didn't know, and I ignorantly regurgitated it. That's the only open public lobby I've raced in ACC; otherwise I've only done private leagues or AI races. Thank you for pointing this out.

Cheers mate :) That's a beautiful monitor. I encourage you to really evaluate the gains in "simulation experience" you think you will obtain with Triple 1440p. You gain some field-of-view, but now you have bezels, and the awkward visual of objects moving from one screen to the next. Your hardware can get you 100 FPS, but you'll have to lower some graphics. The 12900K and 3090ti are certainly good enough to wait for next-gen CPU and GPU.

@@BenchmarkOdysseys Thank you for your input!

Interesting! I look forward to testing this theory. I benched all the CPUs with the same NVCP configuration, so at least any disadvantages to my methods were applied uniformly. My preliminary testing showed strange ghost artifacts with Temporal antialiasing (behind my side-view mirrors, behind other cars), and that ultimately led me to using DLSS. How did you remove those glitches at Native reoslution?

@@BenchmarkOdysseys Looking forward custom profile in NVCP test... don't even know what that means : )

@@kristaps2010 It means that you open Nvidia Control Panel (NVCP) and tweak the game setting from there. Stuff like Texture filtering, Tessalation, Antialiasing, Texture quality, Leve of details and so on. Usually you have them set up in a default way when you install the drivers, but often that default "global" settings end up contradicting with the in-game setting you chose. IF you want your in-game settings to take priority, you should chose "Use the application settings" in NVCP. This btw can have a drastic effect on both your picture quality AND performance, depending on the setting you chose.

Thanks :) Can you recommend a MSFS community or discord to me? I'm a little bit out of my element with that sim and want to make sure I capture results that are relevant to them.

@@BenchmarkOdysseysso I don’t have any specific discord channels for msfs but there are couple of facebook groups… msfs is generally cpu limited so it should be interesting video - also VR could be interesting in case people have something similar to 3080 and would like to play it this way. Just make video like the ones you’ve done already for other sims and it should be fine 👌👊