Can the "BEST" CPU survive a simulation soirée?

[@jorgereyes7880]

cant wait for VR testing

[@damianp5888]

+1 on this :D

[@joeyladouceur3483]

+1 ON THIS AS WELL, plus to see a realistic triple 1440 and 4k scenario!

[@marcusr5715]

Would love to see it!

[@seb-kun]

I also wait for VR Testing. tho you can't use FPS as metric for VR. You must show CPU/GPU load and lower is better. This is because most headsets will force vsync to 90/120. Or even better, use openXR toolkit , record-stats feature and look at CPU render thread frame time.

[@triplej1614]

+1!

[@rohitbeniwal8649]

Can’t appreciate enough the effort you put for these benchmarks exclusively for sim racers. Thanks a lot man!

[@paderborner5213]

I'd highly appreciate if you'd do a VR test for the 9800X3D. Your way of benching VR is simply brilliant

[@flxdrv5020]

I have a quest 3 with the 9800X3D, can finally play half life halyx (using airlink) wihtout any cpu stutters!

[@mich5528]

Undervolting a 4090 with a 7950X3D, I get 80 to 120 fps in VR with Pimax Crystal, most settings on highest with FSR… my guess its GPU bottlenecked or am I wrong?

[@funkyd04]

Except he better whip out a 4090 or else its a waste of time

[@e36blur]

@@flxdrv5020my psvr2 had no stutters using a 5950 and 3090… Im still upgrading to 9800x3d 😂 I noticed the difference on my lil bro’s 7800x3d right away, its the .1% and .01% fps that counts

[@LucaRuggier]

@@mich5528 why undervolt if you are gpu limited

[@Carlos.Damiao]

Awesome as always! VR in iRacing is what I’m looking forward to

[@SergejGrabun]

I like how you explain the graphs. I like the speed of your speech. Easy to follow and is informative. Can't wait for VR benches

[@EricK-fr9fe]

Thank you for the detailed explanation. You have the best benchmarking on KZbin imo. You test all scenarios and you don’t just focus on 1080p. Like so many others

[@c.chepre8452]

one of the best cpu tests that ive seen in a while thanks to the inclusion of gpu busy, a sub well deserved.

[@funkyd04]

I've been ignoring almost every "test" of this CPU waiting for your take on this!!!!!

[@brilongenecker]

Same here. I only sim race and dont care about all the other benchmarks or fps games.

[@funkyd04]

@@brilongenecker Exactly 😂

[@Eonymia]

Why the quotation marks around test as if they weren't actual tests?

[@flxdrv5020]

@@Eonymia he test in CPU limited scenarios only, so it really shows the performance difference between each cpu. instead of showing some titles where the gpu is the limiting factore for multiple cpu's, making the comparaison kinda useless

[@funkyd04]

@@Eonymia Because the only thing i care about is iracing performance, which doesn't exactly correlate with whatever game a reviewer tries to test at 1080p. I dream of the day that a computer can finally drive a 25 year old engine at a respectable fps in VR. And i hate that this is the case, to this day.

[@Rauzaire]

I've been looking forward to this type of sim benchmark, good work, it was missing from the benchmark community!!

[@chrisracing81]

Great, thanks for this! Very nice to see LMU benchmarks on your channel 👍

[@jassonco]

Perfect timing for my new AM5 7700X machine, that I have yet to bring online (iRacing mostly), that I built 2 months ago and I was debating on upgrading it to 7800X3D or 9800X3D, thanks for benchmarking!

[@markosz22]

Upgrading from a Ryzen 5600... holy crap even the 0.1% lows are higher than the average of the 5600. Thanks for the review, it's amazing. Providing GPU busy and doing more in-depth analysis is so much better than just posting a few averages that many channels do.

[@justmejustme3042]

Your presentation and quality of production is excellent. And your explanation about CPU bottleneck with that one graph is even better than Hardware unboxed hour long explanation. Subscribed 🎉

[@intercede007]

Thank you so much for the sim racing focused benchmarks!

[@chrissimpson1098]

Oh oh! And I was putting off upgrading from AM4 (5800x3D/4080S) to AM5. Would like to see iRacing triple 1440 benchmark and then it’ll probably be time!

[@faisalalqhtani3942]

Keep up the amazing work actually first reviewer doing gpu usage that's amazing

[@benthecaptain7419]

So excited for this review! thanks for getting to the stuff that really matters... to me!

[@patino001]

Another well put together, succint, informative video. Thanks for the long hours and tiring effort! 👏

[@TheCuboidman]

I lile this video and how it was presented. Even if these games are not my cup of tea (but like other sim. games), they do present another side of cpu testing that i do not see too often. Deservs a subscription, bbye

[@gaminginstilllife9429]

Are you doing VR testing that's like really hard to do cuz I have all these games but I'm really into VR

[@louiienation8497]

Duddeee, that 9800X3D is absolutely a must for triple 1080p. I thinks that’s my new upgrade!! Thank you for these benchmarks !!

[@thyresias194]

Great job 👍thanks you so much for the simracing benchmark 😊

[@GT4tube]

Great video benchmark!

[@leonardo6002]

Great, thanks for the info

[@MadMike991]

By the way, thank you so much for the effort with this stuff. No one else is really doing this in our niche community!

[@nedis4]

Looking forward to those frames on time benches for IR in VR!

[@MadMike991]

I bought one too! Massive upgrade from my 13700 in iRacing. Huge improvement in AI and at race starts. Later in the race I’m limited by my 165 hz monitors, LOL (I’m maxed out since I have FPS capped at 164). Triple 1440 with 9800x3d and 4090.

[@attilalendvai7167]

Super helpful for simracing addicts! Good stuff!

[@celsoskibinski3832]

Parabéns pelo belíssimo trabalho! Há tempo estava querendo ver um comparativo com os principais simuladores de corrida. Seu vídeo veio em ótima hora, principalmente no momento em que penso de construir uma nova máquina. Certamente essa contribuição será muito útil para os pilotos de automobilismo virtual. Celso Luiz Skibinski / Curitiba / Paraná / Brasil Congratulations on your wonderful work! I've been wanting to see a comparison of the main racing simulators for a while now. Your video came at a great time, especially when I'm thinking about building a new machine. This contribution will certainly be very useful for virtual racing drivers. Celso Luiz Skibinski / Curitiba / Paraná / Brazil

[@LucaRuggier]

Hey, first time watching your channel and very nice benchmarking techniques. You don't often see channels diving deep into the statistics of GPU busy, powerlimit% etc. I think the reason the frame times are bad with the 9800x3d in that one bench mark is because the GPU is at full utilisation, instead of sitting around waiting for frames to process. If I remember correctly, one of Nvidia's control panel settings to do with "Low Latency", I think the "Low Latency Ultra" setting, caps the GPU usage at 95% or something similar to alleviate this problem. So the only reason the 9800x3d had this problem was because it was fast enough to shift the benchmark from a CPU bottleneck to a GPU bottleneck.

[@BenchmarkOdysseys]

Interesting idea! I will investigate this :)

[@Naraenda]

thanks for mentioning low latency. will also try it after work!

[@NesterowiczRacing]

Nice video and great graphs :)

[@BenchmarkOdysseys]

Thanks mate!

[@starofeden9277]

so glad to find someone doing real wide screen i use to run the same setups and i bought a 9800x3d also whole new system you not the first person to notice those weird 1% lows..i have seen other channels also seeing it no idea whats causing it as long as AMD gets to see it ..they will bring a new bios/agesa code to fix it let AMD know about it to fix it pls show other games in wide screen setups.. star citizen star wars assasins creed there is many games to show i use to love playing widescreen triple screen setups..i started that back in 2011 i think

[@tomfuegue]

Great video, greetings from Spain.

[@XiaOmegaX]

Still worth testing: Live For Speed. Sure the simulation tick rate is locked to 100hz and it can do that on a P4, but comparing wattage to achieve it is a fun metric.

[@ПредводительПельмешек]

It could be the high gpu utilisation monitoring issue, so called long-short frame. Its obvious that 9800x3d has much more ultra short frames around 3 ms.There is probability, that these frame combinations (3 ms + 12 ms) exists only on monitoring software, if the record not contains any observerable stutters

[@lukepage72]

this is such a good video wow

[@innovirtua]

I'd like to see the 7950x3d included in all the comparisons here. Specifically how that "limit" that Assetto Corsa reaches is extended out any by the way the 7950x3d works.

[@philipgibbon4473]

I've ordered an 9800X3D, and it will be paring with a GPU that isn't released yet, like the RX 9000 series or wait for a completely know line up of GPUs from all competitors in the future.

[@roderickkennedy2192]

You do great work sir, sim racing only benchmarks seem to be thin on the ground. Thank you from a 7700x, rtx 4070 user in the UK, just subbed too, thank's.

[@MegaCm123456]

Please investigate more the AMS2 triples issue and please VR testing as well.

[@Busy_Child]

now we're cooking, glad youtube showed me this video

[@ClaudioBarroso]

Thank you so much. My current system runs on a 8700k and 3060 Ti. Deciding between the 9800x3d and a 14700k + 4070 super. I do some photoshop/lightroom/video editing and office work (WFH). with my PC. Iracing only on weekends.

[@SwingArmCity]

Umm fastest gaming CPU out there, or an unstable one that doesn't hold a candle.

[@NutellaCrepe]

What a monster of a gaming chip. Glad I stayed up to order one on launch day. Worth.

[@payver]

Frame Chasers had the same finding with the 1% lows. But if you didn‘t see micro stutters, maybe the benchmark software has some hiccups with the bew CPU?

[@STKReacts]

walked into micro center and bought one without trouble yesterday, they are STOCKED.

[@CaveWaverider]

It would be interesting to see whether the frametime problems you're seeing with triple monitors on the 9800X3D would also occur in 4K or higher single monitor resolutions or not.

[@sky173]

I went for a 7800x3D and walked out with a new mobo and a 9800x3D. Best thing I ever did for my computer. Combined with my 4070Ti Super, performance is mind-blowing.

[@phatminhphan4121]

Nice review 🎉 i subbed

[@radekzukowski2523]

Hey, maybe check also live runs for iR, ACC to see if the spikes shows up in those?

[@BenchmarkOdysseys]

Totally! I'm also going to try 24H2 as well, maybe that will help?

[@RedSliceGaming34]

Now this is a fun test. Would love to see VR tests. 30 FPS difference on triples for alot of these games is amazing.

[@Riku_9]

Can you do a test for the 9800X3D VS the 9950X3D test when it comes out too?

[@DaveOfRock]

Thanks for your work. Can we have the test with triple 1440p?

[@triplej1614]

Can we please get vr benchmarks with the 9800x3d and 4080 Super? Especially if you can compare against the 7600x

[@mrb552]

Any upcoming 1440p triples testing with the 9800X3D?

[@BenchmarkOdysseys]

Yes, there have been several requests for that. Introducing a GPU bottleneck, however, does very little to help us understand CPU performance. Through my previous videos I have shared many examples of that with Triple 1440p and Triple 4K. But this video has been crazy popular (for my channel), and I understand a lot of people haven't watched those previous videos.

[@mrb552]

​@@BenchmarkOdysseys Thank you. I understand that GPU bottlenecks could exist and would appreciate understanding more about what those limitations look like before plunking down the cash for that new CPU. I'm currently running a 5600X and RTX3080. Thanks again. 🙂

[@ammarbader3068]

If you own Microsoft Flight Simulator 2020, please benchmark it with 9800X3D single and triple (especially at 4K and 1440). I understand the GPU could be a bottleneck, but MSFS is a special case in this regard. Same applies to MSFS 2024, if we can see how it performs at release before any long-term optimizations take place. Thanks anyway!

[@unotoli]

Great review. Wish you had 1440x monitor and VR for all those benchmarks.

[@phillippatryndal4255]

IIRC the person responsible for starting the trend of looking at 1% lows and frametimes etc. in PC benchmarks (CPU or GPU) was one of the original Tech Report crew, (long before it became the cryptocoin site it now is :( ).

[@grayzedynamics]

Nice review. I need to know if the 9950x or 9800X3D is better for not only gaming but streaming and peripherals.

[@BenchmarkOdysseys]

You could wait for 9900X3D or 9950X3D, rumored to have 3D V-Cache on both chiplets.

[@grayzedynamics]

@@BenchmarkOdysseys very true. I just always wonder why I do not see benchmark scenarios for us streamers/gamers. Would help a ton b/c we need the multicores for both and all the peripherals. Thanks Man

[@natanielcarvalho7530]

Thanks for vídeo. My setup for iRacing: Ryzen 7 5800x3D + RX 7800xt + 32gb DDR4, 3 165hz Full hd monitors, What do you think, do I gain more by upgrading to the rx 7900 xtx or upgrading with Ryzen 9800x3D? Sorry for my english google.

[@BenchmarkOdysseys]

Depends on the sim; checkout my videos with 7900 XTX in the title, or LMU and RennSport GPU analysis.

[@DerekMoore82]

I know why everyone only tests in 1080p, but it's really frustrating because I have no idea if I should upgrade or not. I'm using Ryzen 3700x with a 4070 Super on an AM4 motherboard. If I upgrade to a 5700x3d, will I even see a significant difference at 3440x1400 Ultrawide on High settings? Because I'm never gonna play on lower settings or downgrade to a low resolution monitor. And these tests only show data for people with low resolution monitors.

[@BenchmarkOdysseys]

Hey there. Quick answer, yes, 5700X3D is a worthy upgrade over 3700X.

[@DerekMoore82]

@BenchmarkOdysseys Okay, thank you.

[@celtshaun1427]

Yes you are probably best to get another 12 months or so out of your existing platform before jumping onto AM5 but it must be said AM5 prices for hardware is starting to fall and a killer system can be had for $700 if your not needing a new GPU.

[@DerekMoore82]

@celtshaun1427 Well, I know the 5800X3D is the best CPU my current AM4 motherboard can ever get, but the price and availability on that CPU is already an issue, and I worry there good price and availability of the 5700X3D will only be a thing for a little while longer, and so I feel like I might as well max this build out before starting a new build. The thing is, I already get my desired 165 fps in some of game I play and over 100 in the racing sims I play at max settings, and 60 fps in Cyberpunk with DLSS and raytracing, even with just my little 3700x... so I was wondering if I should even bother upgrading, and the tests don't show what happens when you're at max settings 3440x1440 going from 3700x to 5700X3D. But then again, VR is still a struggle with the 3700x, and the tests definitely never show VR performance comparisons, but then again I don't use VR much.. 🤔 I should probably get it while I can for that peace of mind that my motherboard is maxed out to its fullest potential. Because there won't be a way to upgrade the CPU on it in a year or so if I wanted to. I already missed out on getting a 5800X3D when it was on sale for $200, that was a huge mistake.

[@BATTYMATE]

I Think most of the big drop off's are because you're running on old windows 11 23h and not the updated windows 11 24h2 which is made for ryzen 3d vcache

[@BenchmarkOdysseys]

Thank you for input. I've just completed testing with 24H2. I will publish results shortly :)

[@BATTYMATE]

@@BenchmarkOdysseys it should be a night and day change especially when it comes to 1% lows and 0.1% lows as windows 11 23h can cause ryzen to stutter because windows 11 23h load line isn’t in the correct order for ryzen

[@Pillokun]

state of your tune of the system matters alot if u play at low res/low setting ie high fps gaming. u need to tune the ram so giving us a ram setting specification is important as well as higher ram tune will increase perf just like v-cache helps on amd cpus. Ofcourse ram tune helps the x3d cpus as well. I have 7800x3d and on the lga1700 for now I only have left an 12700k(my other 12700k, 12900k and 2x 13900kf are sold) and ram tune is important for the make the cpu wait less for the data ie increasing perf.

[@oroborus2063]

Thank you!

[@rohitbeniwal8649]

How much of a performance jump will I see from a 13600k to a 9800x3d with a 4080 on triple 1440p?

[@BenchmarkOdysseys]

Increasing resolution shifts the burden to the GPU, but by how much? That depends on the graphic quality you have configured. If you are running Ultra settings and your 4080 is pinned at 100% all the time, you may not see any increase in FPS. But if you race against AI, or run big multiplayer races around complex circuits, you could still see a big increase.

[@seb-kun]

ams2/tripple spikes , can't check from a screenshot, but did you check if the spikes are corelated with GPU busy 100% , or just after, or just before ?

[@imanny99]

Great testing. Really appreciate the time this would've taken. I know you show the 9800X3D with worse 1% and 0.1% lows even though the GPU busy is close to being 100% but not quite. Is your GPU busy number an average of the busy time throughout the run, or is it the max GPU busy? If it's an average, then it's possible the spikes are GPU busy pinging off 100% even though the average is lower.

[@BenchmarkOdysseys]

Interesting idea! Yes, the GPU Busy is an average throughout the run. I will investigate further :)

[@tmoshamo]

I really enjoy these benchmark videos. To you or chat.....ive used triple 1080p 144hz monitors and currently a 1440p 120hz 49 inch monitor...looking at the charts, would it be noticeable if max framerate was say 200 and 0.1% lows were 90? Would i be able to tell on a 120hz monitor? Anyways, thank you for another video. Your hard work is appreciated. Peace✌️

[@BenchmarkOdysseys]

Good question: what is the minimum 'noticeable' frame delay? How does that relate to a monitors refresh rate and panel technology? How does triple screens complicate the scenario? Through my experience, my gut thinks it takes a big frame-spike (like +20ms deviation from average). Those micro-stutters with the 9800X3D were about 7ms and I could not notice them on 144hz monitors.

[@seb-kun]

Assetto Corsa 1 Vehicle Physics engine, is running at 333hz.. 1000/3, so that may be the limitiation.

[@dusan78]

The 0.1% and 1% lows results are concerning during the triple screen test. Do you think it could be the Windows 11 problem? Windows 10 seems to work so much better based on what I saw from other reviews. But, you said you dont notice any stutters so results don't match what you are actually seeing on screen. I feel they are crippling the AMD CPUs on purpose in windows 11 to close the gap between intel and amd. I dont know. But, I can't wait for the VR results as it's what I plan to use the CPU for. Once again, great work and much thanks! My CPU is still on backorder.

[@MadMike991]

I have the 9800x3d and in iRacing when CPU limited I have noticed FPS volatility (sudden drops/spikes) in the FPS # on the screen but I don’t notice actual stuttering because the FPS is still in the triple digits. So this does seem consistent with the bar charts. Running triple 1440 with 9800x3d and 4090. So this is sort of what I would call a “high class problem”!

[@dusan78]

@@MadMike991 That's not good. The graphs show even the 5800x3D (which is what I have) has better 1% and 0.1% lows in this scenario. What do you recommend? Are you also on windows 11? I dont like this problem.lol

[@MadMike991]

@@dusan78 @ Yes, Windows 11. But I’m telling you, it’s just a number on the screen, the FPS is already so high that there is no stuttering!

[@dusan78]

@@MadMike991 Oh I misunderstood. So you mean just the reporting for some reason isn't accurate. Not that frame time is actually worse than the previous gen CPUs. Got it. Maybe it will be brought up and fixed in the bios. This is after all first x3D chip of this generation. I still can't wait to see the VR performance. I have the 4090 and 5800x3D and didn't want to upgrade to 7800x3D as it was the whole new platform but, now I think it's the right time. I got the strix x670e-f board as it still all the connections I needed. New x870e boards looks like they went backwards.

[@MadMike991]

@@dusan78 I’m not saying that the FPS number is wrong, just that a drop from 160 to 140 or 120 is not as noticeable as a drop from 60 to 40 or 20. I think it’s a big upgrade from 5800x3d since this video shows that is similar to the 13700 that I had.

[@zedrandomgamer1435]

Look like my other comment didn't pass the system. Thanks for benchmarking raceroom. In raceroom discord others talk about it and validate the result. Some information for future testing:( I can't send link because it won't let comment pass to you.) - The AI have affect but only when you are choosing cars from different classes and different models. The dev of game optimize the game so that single class with 6-7 type of car don't have affect of fps. - you can use to format for testing on 2 weather that have high load. sunset or sunrise for time is the one that have most effect. single class race or multi class race.( officially 2 class race). - There are community race exist that use cars from multi classes ( such as all dtm 2021-2023 + ADAC 2013-2021 + GTR3 ) that also norm. - the worst loading type is running nord in 26hr layout with 100 car from all class and all cars without repeating. It will need lots of ram and vram but no one use it in mp. the casual ones are done by raido in 36 driver in 5-6 classes ( normally 20-25 GTR3 cars type and 1-3 cars from other classes) If you are interest for more info, check my blocked comment. more information are there. thanks again.

[@BenchmarkOdysseys]

Thank you for the info! You can fire me an email if you'd like (benchmarkodysseys / gmail / com)

[@shuflie]

Adjust your settings so you aren't bouncing off the limits of the GPU and see if the spikes in timings are still there.

[@FireStarter23]

Try to benchmark without power monitoring of the GPU. I saw a video, that this causes stutter. Or just uninstalling afterburner etc.

[@BenchmarkOdysseys]

Hello there :) Yes, I also discovered that the hard way and posted a video about it ("MSI Afterburner (default values) to blame.") It's a reminder that: To measure something, is to influence the outcome.

[@tonkatoytruck]

Running a 5600X3D I bought for $160usd with a B550 mobo and 16GB of memory. The deal of the decade. I will upgrade to AM5 once GPU prices drop more on the 7000 series. Thanks for the comparison. It is appreciated.

[@geologik7500]

> once GPU prices drop more on the 7000 series Oh my sweet summer child, bless your heart for thinking that's what is going to happen.

[@tonkatoytruck]

@@geologik7500 i guess you never heard of Black Friday.

[@tonkatoytruck]

@@geologik7500 7900XT is down to $620usd now and Black Friday is right around the corner. Started at $900usd

[@Shenron67]

Hey, for the thing about Assetto Corsa being the limiting factor, maybe you could add the average gpu and cpu wait ? edit: or at least the gpu average wait when testing cpu, might be more accurate with the gpu busy%

[@louiienation8497]

I would love to see iRacing triple 1080p in the rain benchmarks with SSR HIGH.

[@fracturedlife1393]

is this pre 24h2 then? it massively increased my 1% lows, not that they were even a problem

[@BenchmarkOdysseys]

Correct, I have another SSD and I will prepare it with WIn 11 Pro 24H2 for AM5. Depending on the results, I may develop a video, or just make a post of bench jpgs to my channel.

[@RavenZahadoom]

I think your a little wrong with the bad frame times for the 9800X3D on a couple of the games, in both games the GPU was at least 97% busy, (generally speaking 97 to 99% busy means you are GPU bound with the 2% movement down to driver overhead) the dropped frames are more likely to be a GPU issue than a CPU one. This is one of the reasons you should always limit your FPS in game engines that allow this for more consistent frametimes for both GPU bound and CPU bound scenarios when actually playing games vs benchmarking, just to be clear not saying you should have limited the frame rate during benchmarks. For example in Le Mans Untimate the average FPS for the 9800X3D was 224.9, if you limited it to say 200 or 210, I'm confident the 1% lows would match or best the 7800X3D. But ofcourse this is generally true, and new hardware can and will conflict with older software at times. Still as you said at the start, there is a reason to test CPU's at low resolutions to ensure GPU performance is not an impactful issue, and yet here you are comparing 1% lows from a GPU bound scenario vs one that is not. So that result from Le Mans Ultimate with triple 1080p is more of a GPU test than a CPU test.

[@BenchmarkOdysseys]

Thank you for the explanation. What you describe is plausible. Note that AC and ACC saw the 9800X3D at 97% and 100% GPU Busy, and the lows were not an issue there. I'm trying to investigate this further and your insights help.

[@Cloxxki]

Great FPS in Automobilista 2. Is it inferior for visuals, or just a great game for lesser systems as it doesn't need as much processing power to be drivable competitively?

[@BenchmarkOdysseys]

Generally speaking, I think the physics are less accurate, it's certainly noticeable through the wheel and force feedback. The tradeoff is a bunch of inexpensive cars and tracks, and the AI doesn't murder the player like Le Mans Ultimate lol

[@Scalma_91]

Have you done any benchmark with VR?

[@BenchmarkOdysseys]

I've only done a league race last night, ran fine. I want to include VR benching in my part 2 (9800X3D vs 7800X3D vs 7600X)

[@BeachClub_1]

@@BenchmarkOdysseysThis would be appreciated a lot by the VR enthusiasts 👍🏻

[@kaptainemerica1]

You're doing God's work.

[@JAMESuk0007]

You got a subscribed from me sir great info.

[@CamaroWarrior]

I wonder if my RTX 3090 is bottlenecking my 9800x3D, I upgraded my PC from a MSI b550 gaming carbon, 5800x, 32gb DDR4, to the MSI x670e tomahawk, 9800x3d, 32gb ddr5 6200 but kept the GPU (RTX 3090) I ran game benchmarks between both processors with the same GPU, and I game at 1440p and 4k, I hardly saw a difference in frame rate between both processors they were similar or sometimes 5 to 10fps more on the 9800x3D. The only big difference I seen is in BeamNG with the same settings at 1440p, I went from an average 85 to about 130fps plus I noticed there is a small higher GPU percentage usage with the 9800x3d over the 5800x, but everything else hardly even notice a difference 🤔

[@BenchmarkOdysseys]

I can tell you that my 3080Ti performs similar to 3090, so any benchmarks you see where I test with it, are relevant to your GPU. As for the 9800X3D, make sure your BIOS is updated and you enable PBO (Game Boost) and EXPO if your RAM supports it. Also update your AMD chipset!

[@damienlobb85]

@7:10 the minimum fps of the 9800x3d is almost as high as the average fps of the 5800x3d. That's wild!

[@AntoineSammut-n5k]

My PC is going back to where I bought it from in October. A red light came on inside the PC and my three 55” TV’s are black. Do you recommend to change the CPU to the 9800 from the 7800. I’m using a 4090 as well. Why did that red light go on and the screens went black?

[@BenchmarkOdysseys]

Hey there :) Sorry to hear about the computer trouble. Hopefully the seller can sort it out. If you'd like some outside help, you can flip me an email at [e-mail]

[@gannonsgarage3265]

These are rookie numbers people need to subscribe to this guy views dont match subs😎

[@jouniosmala9921]

GPU may limit the framerate on average, but that doesn't remove occasional spikes on CPU demand. Let's put it in perspective. If you run a game at 300fps vs 60fps, and there's 1 second frame every 10 seconds. At 60 fps puts the cut off point for 0.1% lows around worst frame every 13 seconds. But 300fps puts worst frame around every 3 seconds.

[@eliteproduction3005]

Soooooo if you have a resolution that is more than 5760x1080 pixels, having a 9800X3D serves no purpose and you should upgrade your gpu instead, am I right ?

[@BenchmarkOdysseys]

That's a loaded question! If your current CPU is fast enough to push your graphics card to 100% of its abilities at a higher resolution and graphic settings, then yes, it's unlikely 9800X3D can make a difference to increase MAX fps. But if your fps DIPS into low numbers during a race start (with all those cars) or you frequently race with AI, then the 9800X3D could dramatically increase those lows, depending on what CPU you currently have.

[@eltranza3590]

Thinking of buying the new 9800x3d .. I run 1440p triple monitors .. What is "best" combo cpu/gpu for this? Is the 7800x3d pretty close to the 9800x3d running triples with lets say a 4080 super? Id love your thoughts on this. Thanks

[@thomas1988de1]

Thanks for the benchmarks! Could this be the first cpu that enables us to play iracing in VR with dynamic shadows enabled? Not having them was always such a turnoff for me.

[@JustBenching]

Can you share the replay for testing purposes?

[@BenchmarkOdysseys]

I've been working on that. I think distribution would have to be through a Discord server, that's the 'cheapest' way to share, organize, and communicate.

[@Triple_M_Motorsport]

I understand why it's not popular to do triple 1440p/4K benchmarks but it's hard to estimate how big of a difference there will be upgrading from 7600x to 9800x3d in these scenarios. And what would happen while enabling different DLSS scenarios (quality, balanced, performance) in ACC. I've recently changed from triple 4K to triple 1440p and changed DLSS from performance to quality and gained like 20-30fps (7600 and 4070ti). I've been wondering how big of a gap there is and if it's worth to even think about upgrading CPU (can't afford GPU right now). Do you think you will do a follow-up video regarding solely ACC and different (higher) resolutions and DLSS usage?

[@MKIIUltRacer]

Yup was wondering the same. Ofcourse it makes sense to benchmark in 1080p to see the maximum difference between CPU's but many of us are playing in 2k. I've seen the Daniel Owen and Ancient Gameplays explaining why they test in 1080p but the majority of their testing is single player non racing games. An example why showing the 2k numbers can be useful to people thinking about an upgrade is when I went from a 5600X to a 5800X3D with a RX 6800 GPU. I went from having to fine tune settings (to get the best visuals) on the 5600X to reach 120 fps in ACC to just cranking everything up and getting plus 160 fps in 2k with the 5800X3D (34' monitor). That was a huge performance gain at a low cost (in terms of gaming hardware). I'm now on a triple screen 2k setup and mostly play iRacing these days. When I changed to triples I also upgraded the GPU and at the time I only had ACC in mind which works perfectly with my 7900XTX in triples. It does not work very well with iRacing sadly and am at times with downgraded setting strugling to keep above 100fps. I do want to upgrade both GPU and CPU. I'm still on the AM4 platform so upgrading to a 9800X3D is going to be expensive. Normally it would be more sensible to upgrade the GPU first but I'm waiting for the 5000 series nvidia gpu's and since I want to upgrade both anyway, might as wel just do the CPU first. Hope we both learn more about the 2k numbers soon.

[@MKIIUltRacer]

Just found out that Dan Suzika has a video up with the triple 2k numbers but only for the high end CPU's. It is atleast clear that going from a 7800X3D to a 9800X3D is pointless for 2k.

[@Triple_M_Motorsport]

@@MKIIUltRacer he did that in iR but mentioned in a comment reply that in ACC it was 10-15% better. Nonetheless it's comparison between 7800 and 9800 and I'm curious about 7600 and 9800. So different case. Also I'm wondering about the usage of DLSS.

[@MKIIUltRacer]

@@Triple_M_Motorsport I think the best video for now to get an indication on the performance gain fron 7600X to 9800X3D would be the Wardware Unboxed video (kzbin.info/www/bejne/a3isp6WDjN-lb80). It has a 4k benchmark of ACC which is not that far of from triple 2k and for most cases, 4k almost maxes out a 4090. In the video he uses a 7700X vs the 9800X3D and the diference at 4k is almost a 100fps. Ofcourse the boost will be smaller with a 4070TI but I'm sure you will gain a lot by moving to a 9800X3D.

[@Bhgp]

It would be awesome to have this king of benchmark for triples 1440p and 4k. Awesome information.

[@tunaware]

Wouldn't all the CPUs just be the same at those resolutions if we're already gpu limited at triple 1080p? just look for a 4080 super benchmark at that point to see what they will all get.

[@Bhgp]

@tunaware not necessarily. I have for example a 9700k and it sits on the fence regarding the bottleneck. I would love to have a benchmark with triples 1440p with the 4080 paired with the 9700k i.e.. I guess for triples 4k you are right. 👍

[@seb-kun]

using replay on some titles is not the proper way to test as on some titles during a replay, the whole physics of the player vehicle is not calculated but they just replay the position of the car/wheels.

[@BenchmarkOdysseys]

I appreciate the input. In all sims in which I use replays, I validate that it is accurate. Here is an example of the process: I run the race (AI or multiplayer) and collect performance data. After completing the race I benchmark the replay provided post-race. Then I save the replay, exit the sim, restart the computer, and re-benchmark the replay. If it is in with 5% of real-time, then I considering that accurate. Note that real-time race benching (like I did with AMS2 and LMU) introduces greater than 5% performance variability, especially with triple monitor configuration. The more time a dynamic object is rendered through a separate view port (ie: opponent alongside my car), the worse the performance. Also, the more instances my user input changes AI car position (ie: opponent close to my car) the worse the performance. Therefore, race replays are preferred because they generate consistent results.

[@Riyozsu]

There would have been a bigger difference between 7800x3d and 9800x3d if you tested with 4090

[@MKIIUltRacer]

You should buy him a 4090.

[@MegaCm123456]

@@MKIIUltRacer Its a good point/question he is posing. No need for this satiric remark.

[@MKIIUltRacer]

@@MegaCm123456 Yeah maybe I was to salty but I said what I said. Ofcourse the difference is going to be bigger with a 4090 but what is the point? Let's be real here, Most sim racer are not going to use a 4090 to do single screen 1080p. This channel is one of the very few that do benchmarks with usefull data for sim racing. He's using the second best overal GPU you can get at the moment and he has to buy them with his own money. People with a 4080s or 4090 are most likely people running 2k/4k triple monitors or VR and for both cases you're going to be GPU limited. If you really wan't to see it tested on a 4090 than you should look up Dan Suziki's video. He shows that for triple 1440p, the upgrade from a 7800X3D to the 9800X3D is pointless. Have even seen a video (can't remember who it is from) where in triple 1440p it is even pointless to upgrade from a 5800X3D with a 4090.

[@jouniosmala9921]

And Zen5 weakness is probably anytime two CPU threads use same data on same core the other thread can only access it through L3 cache, previously L1 and L2 caches were the place for data sharing. And that's on purpose to prevent one security vulnerability all older CPU:s have. But it's kind of worse than that with Windows scheduler moving threads from core to core in same cluster. Generally optimized game will tell Windows scheduler okay this core is free for any task you want to give it when one thread has done its task. But now it matters a lot more that when those threads start again that they are in exactly same cores as before. And getting the AI threads running short bursts means they jump a lot around.

[@dylanmonahan5909]

Will be interesting to see if this is solved in the 9950x3D.

[@jouniosmala9921]

@@dylanmonahan5909 It's either software fix, that will help every zen5 or it's fundamental to core design and any fix is part of new core design in future generations. Probably best fix would be using something else than threadID as the identification which threads are allowed to touch the data in the cacheline. Basically having some kind of adresspaceID so multiple threads sharing same addresspace would be sharing the addresspace in cache also. However, just fixing Windows scheduler could mitigate the issue. Also it can be mitigated by actual game engine design.

[@dylanmonahan5909]

@jouniosmala9921 What I was referring to was the possibility (unconfirmed rumour) that the 9950x3D (and maybe the 9900x3D) release with 3D V-cache incorporated into both CCD chiplets... though it is probably just a pipedream. I do hope the issues are properly addressed at the software level though, as this will benefit a lot of people, like you said.

[@jouniosmala9921]

@@dylanmonahan5909 Problem I'm describing is issue that L3 cache which size is increased by 3DVcache is lowest level of cache where multiple threads using same core can share data. And any data that is actively used by multiple threads in the same core is going to have L1 and L2 cache misses and always hit L3 cache anyway. Both CCD:s are probably going to have 3DVcache and that doesn't help any issues 9800X3D has, because it's single CCD is going to have large cache anyway.

[@ut2k4wikichici]

Thanks x3d benchtano

[@RelaxSimmer]

I have 4090 and I'm cpu bottled necked in VR with aero varjo

[@B__L]

Complete overkill

[@maed204]

Nordschleife on Raceroom is pretty cpu heavy

[@anonym8951]

Very good videos ❤ thanks I am using a i5 14600kf with a 4070ti on triple 32"@1080p and loving it. Will maybe update with the next big iracing update .. maybe 😀