Another excellent video. Followed by looking at 4080 Super prices (£1200 incl taxes), then 4090s (£1900), then 7900XTXs (£1000). Then wondering about a used 3090, or waiting for the 5080.
The timing of this video for me and the overall quality are insane. I just started considering a triple 4k sim setup, but wanted to do my homework on the hardware needed to pull it off. This is exactly the info I needed. Best of luck on gaining subs! the video quality is there, it just seems to be a relatively small audience compared to general PC gaming
Holy "Radeon" keep up the great vids. Yet again hitting meaningful benchmarks. Waiting forward to you hitting 1K subs! I think you're like 60% there! Regarding your question about comparing each video card that you test vs. what the optimum driver is, great question. I'd say make mention of which driver you're using. And this opens up the door to future videos to update the specs... daunting yes, but that's tech nowadays - which is why I tune into your channel: My one stop shop for performance analysis and benchmarking results for some of the leading simulators! thx again!
Thanks! I don't think it's unreasonable to complete a 'basic' driver comparison once a month or on major revisions. A single racing scenario within each sim, maybe just 4K and Triple 1440p.
The best Sim and graphics card comparison I have seen! You're a legend mate. Yeah, CPU needs to be faster I think for Super :) Getting mine soon with a latest i9. Will be interesting to see how that goes.
Thanks for the Benchmarks! What AMD card are you going to go out and get? Could I ask for you to get the 7900XTX? It’s expensive so I totally get it if you don’t, I just haven’t found any triple monitor benchmarks for iRacing with that card and I’m wondering how it stacks up
Just upgraded from a single 1080 to triple 1440 on a 7800x3d and 7800xt pc. Now looking to upgrade my gpu, likely to a 4080S, thank you for this video. Exactly what i was looking for, comparison wise. Seeing your FPS on different iracing tracks is helpful and let's meet know where you are vs where I'm at FPS wise.
Definitely earned my sub just by testing sim driving with triples etc. Down the road it would be nice to see some testing of a little older hardware (that is more affordable. 🤣) to see what it takes to get to certain performance benchmarks. Also wouldn't mind seeing some ATS/ETS2 testing since that market seems to be expanding rapidly when it comes to equipment (Moza I'm looking at you. 😉) and there is a natural crossover for some that want to race and sometimes want to relax a bit more if it interests you. Overall the content has been excellent!
This is great. I have a 5800x3d and 3090 and play iracing on my Q3. I have a 7800x3d and 4080 Super sitting here that I was getting ready to hook up but now I'm wondering if I should keep the 3090. Looking forward to your next video! Subbed! Please test rain anyway!
Thanks! I will be collecting performance results via SteamVR and a Valve Index because I can use a 3rd party application to collect detailed data. I believe the relative performance of the CPUs and GPUs will be relevant to Quest 3.
@@BenchmarkOdysseys they sure will, thanks! I think I’m most curious if VR will behave like a similar resolution triple-monitor setup or if there is additional processing required (I’m sure I can find this info elsewhere, but seeing it tested with multiple monitors is really the most telling). This memory bus’s impact though is really interesting to see and makes the decision challenging. It seems like newer tracks will lend themselves toward the newer gen of chips a bit better…but yeah I’m looking forward to any sort of conclusions you come to. Thanks again!
Lot of work put into the video. well done. Only a couple things I would suggest. Would be great see a screen shot of what settings in game are actually being used. Far too many folks have differing opinions on what "High" settings are since iRacing doesn't actually offer a true "High" "Medium" and "low" type presets other than the left colum. I think to push the cards as hard as possible maxing out the settings would ensure the cards and cpu are forced to work as hard as possible. Some options like SSAO and Shadow maps are going to really push the system. and secondly, maybe I missed it as I had this running while doing some paperwork. since the average official splits at +/- 25 cars or so on avg. It would be interesting to see replays of official lobbies and then with Ai as Ai is going to be more heavy of course and give us a reflection of the differences in stress. Otherwise great job enjoyed the content!
Thank you for the comment :) In previous videos I showed the in-sim graphics settings, but I didn't like the aesthetic of inserting that into the video. I know, lame excuse, but I think your right. Maybe I'll go back to splash it on the screen. As for 'maxing' out the graphics, I hear your logic and debated it myself. Ultimately I want the results I present to be relevant to users, and that means practical settings. Crushing a 3060ti or 6700xt with intense settings that owner's of that card would never use, doesn't seem super helpful. There's also the subjective experience of what individual drivers value in their graphic settings. I just don't think there's a way to please everyone! heh There isn't much difference between a replay of an AI race or a replay of an online lobby. You're right though, real-time AI gameplay is the harder to render, and discovering performance results for that has value. An ironic challenge with iRacing is the 'quality' of it's AI mimicking the mistakes and randomness of human racers, making each race different. Another challenge with capturing real-time performance from AI races is that the distance of my car to other cars matters. iRacing FPS is sensitive to how close other cars are, and that's a variable I cannot control in real-time. Add in my own driving mistakes or inconsistencies during a lap, and now I'm attempting many more bench runs. I do like the idea of tackling 'AI benchmarking' as a separate, standalone video... but I don't think it's something I can include regularly.
Really interesting, I just bought a 4080 Super as an upgrade to a 3080ti - mainly for DLSS and AV1 encoding for the Quest 3 via Virtual Desktop - it's interesting that at pure rendering, there isn't a huge difference in some circumstances...
iRacing consistency is like that of a Fentanyl addict. It's all over the place. Tracks at 300, tracks a 100. Wat happens on triple in one track doesn't happen on the next...God bless them. I expected a much wider difference. The memory bus difference could be the answer to such a lack of difference in this title. I have a 3080 12GB with a 320bit bus and I thought I was going to sell it for a 4070 Ti Super but I'm actually going to wait for a more meaningful update. Let's see how the 7900 XT/XTX perform.
Haha! Yes, it's really all over the place. I think this is also a reason why user's report different experiences with the same hardware and even graphics settings. Someone may be MX-5 on old 'free' tracks, while another is multi-class on the latest road track. The difference can be staggering.
I have an XTX but I run a single screen. I don't wanna upgrade to triples until I know how it would perform. I have read that it doesn't perform as well as NVIDIA sadly
I've heard the same, unfortunately, user reports are rarely consistent. I'm really looking forward to testing it first hand and sharing my results. Thank you again for the sub :)
*Benchmark Odysseys* I'm glad that you appear to purchase 12/16+ GB GPUs. That would be ideal for someone like myself that plays on 1440p ultra graphics.
Thanks :) When running triple screens, the simulator window is stretched across all three monitors. If the simulator is running at 210 FPS, then each monitor is displaying the simulator at 210 FPS.
Great work. Looking forward to your analysis on VR. I run on a AMD 7 7800X3D, 32 gb RAM, 3080 10 gb system and HP G2. Will it be enough for the coming rain on iRacing? We'll see.
Thank you :) I think rain will be very intense. The 'sneak peak' from the iRacing streamers these last couple days presents misleading performance. Morad and Malone use RTX 4090 which is why they had no complaints while even running Triple screens. Jimmer's on a 3090 pushing a single monitor, and he was getting 60-75 FPS.
Other hardware reviewers (Gamers Nexus, Techpowerup, Hardware Unboxed, etc) have shown the 3060 Ti to perform very similar to the 4060 Ti in traditoinal gaming at 4K. I suspect the same will play out in single 4K with sim racing, however, the 4060 Ti only uses a 128-bit memory interface and that could be troublesome for triple screens.
When you do the VR testing, can you test at multiple SS settings? The new gen hmd's are pushing ~20million pixels with the help of OpenXRToolkit foveated rendering. Would be great to see along side the ~12 million pixels of previous generations.
I've been racking my brain on how to set up the VR testing. There's just too many headsets and variation in specs, with each user having their own 'personal preference' of SS + Hz + in-sim graphics. I've come to the conclusion that the VAST majority will disagree with my choices and they will feel like my results are not relevant to their configuration! Right now I'm leaning towards two graphics configurations, one is potato with everything reduced to near minimums. The the other is steak sandwich with a lot of quality settings. I was going to keep the resolution the same for both tests. Something around the 12 million pixel count. Enthusiast users with Pimax, Aero, etc, have other software technologies that I cannot emulate with my Valve Index. But I can push the super sampling to insane levels of 4508x5008 per eye.
@@BenchmarkOdysseys Oh. No doubt VR testing is exponentially more complicated than monitors. I think what most VR users are interested in though is simply if they're cpu or gpu limited. Myself being one of them as my 10700k/3090 setup is showing my gpu as the bottleneck, but people still just parrot the single thread cpu thing. So clarification on that would be simply amazing. I don't think you need to go down the rabbit hole of assistive tech like OXRTK or eye tracking as that's stuff we can parse out on our own. If time is your friend, I think running 2 resolutions and 2 graphic settings would cover the entire VR community. 10 million pixels will cover all gen 1 hmds including the new Quest 3 which is becoming the ideal starter unit with the discontinuation of the G2. And 20 million pixels covers the new gen Aero/Crystal/BSB and future units. Exact pixel count isn't important, just knowing if there's a point where gpu really does become all that matters. As far as graphic settings go, there's a few community guides I could link you to if needed. They give a great baseline most people use. Then you could turn up everything on the left side to High to see if that really does smash the cpu. I honestly think that would cover the entire community and would be a massive help to most of us. :)
havent read all the comments so not sure if someone mentioned this before: but especially for 1080 and 1440 your cpu is the bottleneck, not the gpu. so those tests have very limited explanatory power in my opinion... i have a 4080 super and a 7800x3d and i get around 140 fps on 1440 on daytona for example ...
Yip, I always point out a CPU bottleneck within the video as well (aka "GPU is idling more at this resolution", or "GPU Busy ratio is 76%"). Daytona @ Night with 50 cars in a tight AI field, is a uniquely difficult compute scenario. And the 7800X3D performs extremely well there, as I previous showed in my "Triple Screens: 8 CPUs Tested with iRacing". I think the 7800X3D is the best chip to buy right now for sims!
I was really surprised at how much of an FPS boost there was! I'd recommend tuning your graphics to get close to the desired performance you want, and when a challenging track/car combo comes up, it's easy to just enable FSR Ultra or Quality and get that smooth experience back.
Triple monitors can be configured in Windows as three separate screens. Three taskbars, three start menus, and you must mark one screen as your 'primary'. If you were to 'fullscreen' an application or simulator, it would only maximize in one of those three screens. Therefore, simulators have to run 'Windowed Bordeerless' to stretch across all three. Nvidia Surround (or AMD Eyefinity) combines all three monitors and treats them as a single desktop in Windows. So one taskbar, one start menu. If you fullscreen an app like a simulator, it naturally stretches across all 3. There are (small) performance benefits to this. Most racers prefer regular triple screen setup because it allows for more user flexibility. Also, lots of racers use 3rd party overlays that require the simulator to NOT be running fullscreen anyways. Some simulators, however, naturally assume the user is racing at full screen. Therefore, I often benchmark at Surround 1080p, 1440p, or 4K because it streamlines my testing procedures.
I wanted to carry forward my results from the 3060Ti vs 6700XT, and those were the settings I used. Future analysis will have High for Sky/Pit, however, I'm on the fence with Shadow Maps. Combined with Tempest weather I think the Shadows Maps are an expensive luxury that few sim racers will deploy.
I really want to test one. I think the Super is the most interesting in the 4070 lineup, but I don't currently have a timeline for when I'll get my hands on one.
I am a little confused, I am running in 1080P, your saying that at 1440P with no other changes, it will have higher frame rate? I am not sure why that happens, shouldn’t it be opposite? My monitors only do 1080, should I buy 1440 monitors (3), or spend more on a better video card than my 6800XT?
Hey there :) Hopefully I did not mistakenly say the wrong thing. The data in the bar charts should resolve any confusion. Increasing resolution (like Triple 1080p to Triple 1440p) will demand more work from the graphics card. But if the graphics card can still complete that work (at Triple 1440p) before the CPU can provide it with more instruction, then it's possible for very little FPS change. This can occur in bizarrely intense CPU-bottleneck scenarios (like first lap around Daytona with 60 AI), or if you pair a fast card with a slow processor, like 4080 Super with AMD Ryzen 7 2700.
This is not new Nvidia slows down GPUs from previous gens when new models are out. For the sake of beeing representative, I'd say just use the latest drivers, because not everyone will use older ones.
Yikes! The cynic in me was thinking the same thing. Just like OS updates to phones. I'm familiar with the gradual decline, but it was shocking to see direct A/B testing with dramatic differences.
I have a 7800x3d and 7800xt. I started iracing in January on a single 1080p monitor and upgraded to triple 1440p this past week. I went from max graphic settings with FPS capped at 150 to med/low settings and barely got 100fps now, some tracks I'm in the high 60s low 70s. No stutters, though, I'd say I average upper 80s in iracing. With that said, I'm not happy with what I've had to lower, graphic wise, to keep my FPS up. For example, Long Beach street circuit, certain buildings don't render in now and it just looks odd without them there. Hence why I'm not searching TH-cam for triple monitor gpu bench marks on the 4080 Super. Will be upgrading sooner than later.
Surround is NVIDIA's mutli-screen technology that makes 2 or more displays behave like 1 within Windows. The AMD equivalent is Eyefinity. There may be a performance benefit to do this, however, there are also annoyances in the user experience (like a Windows taskbar that stretches from the far left monitor all the way across to the right).
question/opinion - currently running cpu i7-9700k ( i know .. little older cpu ) .. my question is. what gpu would pair nicely with this cpu? want to run triple 1440 120fps. cpu is running default.. would it help to OC it to 5ghz and pair with something newer gpu wise? dont really need to run high settings in iracing. just something decent. Thanks
Yes, overclocking helps the 9700K. You don't need a fancy cooler; I ran mine with a Hyper 212 for years, so a single tower cooler is fine. If you go faster than an RTX 3070 or RTX 2080Ti, you'll become increasingly limited by the CPU.
There will be some scenarios (like Daytona) and newer tracks that are challenging for any CPU and GPU, and we have rain coming next week, which I have not tested it yet. If you tune your settings to Medium and maybe a couple Lows, turn on FSR Ultra... I'm confident you can get +100FPS in most scenarios. Just keep in mind, race starts are always the most challenging (which all of my benchmarks are).
@@eltranza3590I have this exact setup - 9700k overclocked to 4.9, 4080S & 1440p triples. I made the jump recently from 2080 and performance is as expected. Not as high as the results shown in this video, but thats due to CPU. If you tweak the settings as suggested by BenchO below you'll be fine with >120
There are a few 'use cases' out there. It works better for some people, and I've certainly measured (slightly) better performance with Surround enabled. But I think for a lot of people, yes, you are right.
@@chrisracing81That is not true my man do more research. Running surround projects one big screen over the entire space, not using it renders each screen separately, this allows for more freedom with rendering from your P.C and results in a nicer picture more often "I can see the difference right away with separate render im not sure how you can't" it is often more taxing on the P.C to run it this way than with surround but not by much if at all. Also might be a fix if you need to run one monotir that is low hz. It will alow the rest of the screens to run at their proper FPS. Surround might not hit FPS but it is another program you simply do not need to run. The Sim runs cleaner without the use of surround the picture renders separately, it does have a performance difference, you get more picture options as well, it is not needed at all, not for many years now that is a fact. Great vid by the way.
Another excellent video. Followed by looking at 4080 Super prices (£1200 incl taxes), then 4090s (£1900), then 7900XTXs (£1000). Then wondering about a used 3090, or waiting for the 5080.
The timing of this video for me and the overall quality are insane. I just started considering a triple 4k sim setup, but wanted to do my homework on the hardware needed to pull it off. This is exactly the info I needed. Best of luck on gaining subs! the video quality is there, it just seems to be a relatively small audience compared to general PC gaming
Amazing comparison! I upgraded my GPU and still keeping the 3080ti as a backup.
Holy "Radeon" keep up the great vids. Yet again hitting meaningful benchmarks. Waiting forward to you hitting 1K subs! I think you're like 60% there! Regarding your question about comparing each video card that you test vs. what the optimum driver is, great question. I'd say make mention of which driver you're using. And this opens up the door to future videos to update the specs... daunting yes, but that's tech nowadays - which is why I tune into your channel: My one stop shop for performance analysis and benchmarking results for some of the leading simulators! thx again!
Thanks! I don't think it's unreasonable to complete a 'basic' driver comparison once a month or on major revisions. A single racing scenario within each sim, maybe just 4K and Triple 1440p.
Would love to see a 7900 XTX comparison. Great videos.
The best Sim and graphics card comparison I have seen! You're a legend mate. Yeah, CPU needs to be faster I think for Super :) Getting mine soon with a latest i9. Will be interesting to see how that goes.
Thanks for the Benchmarks! What AMD card are you going to go out and get? Could I ask for you to get the 7900XTX? It’s expensive so I totally get it if you don’t, I just haven’t found any triple monitor benchmarks for iRacing with that card and I’m wondering how it stacks up
Great!
Rocking a 3080ti for triple 32’s and ill get a 4080super used close to me! Thanks for the vid!!
Just upgraded from a single 1080 to triple 1440 on a 7800x3d and 7800xt pc. Now looking to upgrade my gpu, likely to a 4080S, thank you for this video. Exactly what i was looking for, comparison wise. Seeing your FPS on different iracing tracks is helpful and let's meet know where you are vs where I'm at FPS wise.
Definitely earned my sub just by testing sim driving with triples etc. Down the road it would be nice to see some testing of a little older hardware (that is more affordable. 🤣) to see what it takes to get to certain performance benchmarks. Also wouldn't mind seeing some ATS/ETS2 testing since that market seems to be expanding rapidly when it comes to equipment (Moza I'm looking at you. 😉) and there is a natural crossover for some that want to race and sometimes want to relax a bit more if it interests you. Overall the content has been excellent!
Keep up the great work!
This is great. I have a 5800x3d and 3090 and play iracing on my Q3. I have a 7800x3d and 4080 Super sitting here that I was getting ready to hook up but now I'm wondering if I should keep the 3090. Looking forward to your next video! Subbed! Please test rain anyway!
Thanks! I will be collecting performance results via SteamVR and a Valve Index because I can use a 3rd party application to collect detailed data. I believe the relative performance of the CPUs and GPUs will be relevant to Quest 3.
@@BenchmarkOdysseys they sure will, thanks! I think I’m most curious if VR will behave like a similar resolution triple-monitor setup or if there is additional processing required (I’m sure I can find this info elsewhere, but seeing it tested with multiple monitors is really the most telling).
This memory bus’s impact though is really interesting to see and makes the decision challenging. It seems like newer tracks will lend themselves toward the newer gen of chips a bit better…but yeah I’m looking forward to any sort of conclusions you come to. Thanks again!
Lot of work put into the video. well done.
Only a couple things I would suggest. Would be great see a screen shot of what settings in game are actually being used. Far too many folks have differing opinions on what "High" settings are since iRacing doesn't actually offer a true "High" "Medium" and "low" type presets other than the left colum. I think to push the cards as hard as possible maxing out the settings would ensure the cards and cpu are forced to work as hard as possible. Some options like SSAO and Shadow maps are going to really push the system.
and secondly, maybe I missed it as I had this running while doing some paperwork. since the average official splits at +/- 25 cars or so on avg. It would be interesting to see replays of official lobbies and then with Ai as Ai is going to be more heavy of course and give us a reflection of the differences in stress.
Otherwise great job enjoyed the content!
Thank you for the comment :) In previous videos I showed the in-sim graphics settings, but I didn't like the aesthetic of inserting that into the video. I know, lame excuse, but I think your right. Maybe I'll go back to splash it on the screen.
As for 'maxing' out the graphics, I hear your logic and debated it myself. Ultimately I want the results I present to be relevant to users, and that means practical settings. Crushing a 3060ti or 6700xt with intense settings that owner's of that card would never use, doesn't seem super helpful. There's also the subjective experience of what individual drivers value in their graphic settings. I just don't think there's a way to please everyone! heh
There isn't much difference between a replay of an AI race or a replay of an online lobby. You're right though, real-time AI gameplay is the harder to render, and discovering performance results for that has value. An ironic challenge with iRacing is the 'quality' of it's AI mimicking the mistakes and randomness of human racers, making each race different. Another challenge with capturing real-time performance from AI races is that the distance of my car to other cars matters. iRacing FPS is sensitive to how close other cars are, and that's a variable I cannot control in real-time. Add in my own driving mistakes or inconsistencies during a lap, and now I'm attempting many more bench runs.
I do like the idea of tackling 'AI benchmarking' as a separate, standalone video... but I don't think it's something I can include regularly.
Really interesting, I just bought a 4080 Super as an upgrade to a 3080ti - mainly for DLSS and AV1 encoding for the Quest 3 via Virtual Desktop - it's interesting that at pure rendering, there isn't a huge difference in some circumstances...
iRacing consistency is like that of a Fentanyl addict. It's all over the place. Tracks at 300, tracks a 100. Wat happens on triple in one track doesn't happen on the next...God bless them. I expected a much wider difference. The memory bus difference could be the answer to such a lack of difference in this title. I have a 3080 12GB with a 320bit bus and I thought I was going to sell it for a 4070 Ti Super but I'm actually going to wait for a more meaningful update. Let's see how the 7900 XT/XTX perform.
Haha! Yes, it's really all over the place. I think this is also a reason why user's report different experiences with the same hardware and even graphics settings. Someone may be MX-5 on old 'free' tracks, while another is multi-class on the latest road track. The difference can be staggering.
I have an XTX but I run a single screen. I don't wanna upgrade to triples until I know how it would perform. I have read that it doesn't perform as well as NVIDIA sadly
I've heard the same, unfortunately, user reports are rarely consistent. I'm really looking forward to testing it first hand and sharing my results. Thank you again for the sub :)
*Benchmark Odysseys* I'm glad that you appear to purchase 12/16+ GB GPUs. That would be ideal for someone like myself that plays on 1440p ultra graphics.
New Sub today, id be interested in you doing this with weather effects on iRacing once released. Great content.
Thank you for the sub! I'm currently working on Weather analysis :)
Nice Video!
The FPS at the Benchmark for Triples are for 1 Monitor or all 3 together?
210FPS / 3 Monitors = 70FPS per Monitor?
Thanks :) When running triple screens, the simulator window is stretched across all three monitors. If the simulator is running at 210 FPS, then each monitor is displaying the simulator at 210 FPS.
@@BenchmarkOdysseys thx
Great work. Looking forward to your analysis on VR. I run on a AMD 7 7800X3D, 32 gb RAM, 3080 10 gb system and HP G2. Will it be enough for the coming rain on iRacing? We'll see.
Thank you :) I think rain will be very intense. The 'sneak peak' from the iRacing streamers these last couple days presents misleading performance. Morad and Malone use RTX 4090 which is why they had no complaints while even running Triple screens. Jimmer's on a 3090 pushing a single monitor, and he was getting 60-75 FPS.
Great video Waiting for the 4060ti benchmarke
Other hardware reviewers (Gamers Nexus, Techpowerup, Hardware Unboxed, etc) have shown the 3060 Ti to perform very similar to the 4060 Ti in traditoinal gaming at 4K. I suspect the same will play out in single 4K with sim racing, however, the 4060 Ti only uses a 128-bit memory interface and that could be troublesome for triple screens.
@@BenchmarkOdysseys soo its not recomended for triple screen ?
Thanks. But fow-liage
When you do the VR testing, can you test at multiple SS settings? The new gen hmd's are pushing ~20million pixels with the help of OpenXRToolkit foveated rendering. Would be great to see along side the ~12 million pixels of previous generations.
I've been racking my brain on how to set up the VR testing. There's just too many headsets and variation in specs, with each user having their own 'personal preference' of SS + Hz + in-sim graphics. I've come to the conclusion that the VAST majority will disagree with my choices and they will feel like my results are not relevant to their configuration!
Right now I'm leaning towards two graphics configurations, one is potato with everything reduced to near minimums. The the other is steak sandwich with a lot of quality settings. I was going to keep the resolution the same for both tests. Something around the 12 million pixel count.
Enthusiast users with Pimax, Aero, etc, have other software technologies that I cannot emulate with my Valve Index. But I can push the super sampling to insane levels of 4508x5008 per eye.
@@BenchmarkOdysseys Oh. No doubt VR testing is exponentially more complicated than monitors. I think what most VR users are interested in though is simply if they're cpu or gpu limited. Myself being one of them as my 10700k/3090 setup is showing my gpu as the bottleneck, but people still just parrot the single thread cpu thing. So clarification on that would be simply amazing. I don't think you need to go down the rabbit hole of assistive tech like OXRTK or eye tracking as that's stuff we can parse out on our own. If time is your friend, I think running 2 resolutions and 2 graphic settings would cover the entire VR community. 10 million pixels will cover all gen 1 hmds including the new Quest 3 which is becoming the ideal starter unit with the discontinuation of the G2. And 20 million pixels covers the new gen Aero/Crystal/BSB and future units. Exact pixel count isn't important, just knowing if there's a point where gpu really does become all that matters.
As far as graphic settings go, there's a few community guides I could link you to if needed. They give a great baseline most people use. Then you could turn up everything on the left side to High to see if that really does smash the cpu.
I honestly think that would cover the entire community and would be a massive help to most of us. :)
can you make 4080 super vs 3080ti vr also will be interesting how 4070ti super would be in vr racing
4080S vs 3080ti in VR is my next video! If things go well over the next few days... I'll also have some quick rain analysis too.
havent read all the comments so not sure if someone mentioned this before: but especially for 1080 and 1440 your cpu is the bottleneck, not the gpu. so those tests have very limited explanatory power in my opinion... i have a 4080 super and a 7800x3d and i get around 140 fps on 1440 on daytona for example ...
Yip, I always point out a CPU bottleneck within the video as well (aka "GPU is idling more at this resolution", or "GPU Busy ratio is 76%"). Daytona @ Night with 50 cars in a tight AI field, is a uniquely difficult compute scenario. And the 7800X3D performs extremely well there, as I previous showed in my "Triple Screens: 8 CPUs Tested with iRacing". I think the 7800X3D is the best chip to buy right now for sims!
What software are you using to gather Frametimes+GPUbusy metrics?
CapFrameX, current version is 1.7.2 with 1.7.3 the beta on it's 15th build.
FSR part interested me because i thought it waas just for AMD cards. found on iracing it can be used on nvidia cards. gonna have to try it
It's a bit flickery and depending on the resolution you play you can enable it at Quality. FSR 1.0 is a bit primitive compared to DLSS 3.X or FSR 3.X.
I was really surprised at how much of an FPS boost there was! I'd recommend tuning your graphics to get close to the desired performance you want, and when a challenging track/car combo comes up, it's easy to just enable FSR Ultra or Quality and get that smooth experience back.
I'm new but what is surround 4K ?
Triple monitors can be configured in Windows as three separate screens. Three taskbars, three start menus, and you must mark one screen as your 'primary'. If you were to 'fullscreen' an application or simulator, it would only maximize in one of those three screens. Therefore, simulators have to run 'Windowed Bordeerless' to stretch across all three.
Nvidia Surround (or AMD Eyefinity) combines all three monitors and treats them as a single desktop in Windows. So one taskbar, one start menu. If you fullscreen an app like a simulator, it naturally stretches across all 3. There are (small) performance benefits to this.
Most racers prefer regular triple screen setup because it allows for more user flexibility. Also, lots of racers use 3rd party overlays that require the simulator to NOT be running fullscreen anyways.
Some simulators, however, naturally assume the user is racing at full screen. Therefore, I often benchmark at Surround 1080p, 1440p, or 4K because it streamlines my testing procedures.
@@BenchmarkOdysseysman thanks for this explanation ! that has answered numerous questions I had 👍
Why testing with shadow maps off and sky /pit medium??
I wanted to carry forward my results from the 3060Ti vs 6700XT, and those were the settings I used. Future analysis will have High for Sky/Pit, however, I'm on the fence with Shadow Maps. Combined with Tempest weather I think the Shadows Maps are an expensive luxury that few sim racers will deploy.
Any chance you’re doing a 4070 Super??
I really want to test one. I think the Super is the most interesting in the 4070 lineup, but I don't currently have a timeline for when I'll get my hands on one.
I usually use a 4070Ti (12gb) for PC gaming but it seems that i need to buy a 3080ti just for iracing lol
That card should perform similar to a 3080ti. Do you have issues with it?
I am a little confused, I am running in 1080P, your saying that at 1440P with no other changes, it will have higher frame rate? I am not sure why that happens, shouldn’t it be opposite?
My monitors only do 1080, should I buy 1440 monitors (3), or spend more on a better video card than my 6800XT?
Hey there :) Hopefully I did not mistakenly say the wrong thing. The data in the bar charts should resolve any confusion. Increasing resolution (like Triple 1080p to Triple 1440p) will demand more work from the graphics card.
But if the graphics card can still complete that work (at Triple 1440p) before the CPU can provide it with more instruction, then it's possible for very little FPS change. This can occur in bizarrely intense CPU-bottleneck scenarios (like first lap around Daytona with 60 AI), or if you pair a fast card with a slow processor, like 4080 Super with AMD Ryzen 7 2700.
This is not new Nvidia slows down GPUs from previous gens when new models are out.
For the sake of beeing representative, I'd say just use the latest drivers, because not everyone will use older ones.
Yikes! The cynic in me was thinking the same thing. Just like OS updates to phones. I'm familiar with the gradual decline, but it was shocking to see direct A/B testing with dramatic differences.
7800
Yeah I'm super curious about that card, because the bang-for-buck looks really good.
I have a 7800x3d and 7800xt. I started iracing in January on a single 1080p monitor and upgraded to triple 1440p this past week. I went from max graphic settings with FPS capped at 150 to med/low settings and barely got 100fps now, some tracks I'm in the high 60s low 70s. No stutters, though, I'd say I average upper 80s in iracing.
With that said, I'm not happy with what I've had to lower, graphic wise, to keep my FPS up. For example, Long Beach street circuit, certain buildings don't render in now and it just looks odd without them there.
Hence why I'm not searching TH-cam for triple monitor gpu bench marks on the 4080 Super. Will be upgrading sooner than later.
What is surrond 4k ?
Surround is NVIDIA's mutli-screen technology that makes 2 or more displays behave like 1 within Windows. The AMD equivalent is Eyefinity. There may be a performance benefit to do this, however, there are also annoyances in the user experience (like a Windows taskbar that stretches from the far left monitor all the way across to the right).
question/opinion - currently running cpu i7-9700k ( i know .. little older cpu ) .. my question is. what gpu would pair nicely with this cpu? want to run triple 1440 120fps. cpu is running default.. would it help to OC it to 5ghz and pair with something newer gpu wise? dont really need to run high settings in iracing. just something decent. Thanks
Yes, overclocking helps the 9700K. You don't need a fancy cooler; I ran mine with a Hyper 212 for years, so a single tower cooler is fine. If you go faster than an RTX 3070 or RTX 2080Ti, you'll become increasingly limited by the CPU.
@@BenchmarkOdysseys will that config be OK/ decent for running triples 1440p 120fps on iracing ?
There will be some scenarios (like Daytona) and newer tracks that are challenging for any CPU and GPU, and we have rain coming next week, which I have not tested it yet. If you tune your settings to Medium and maybe a couple Lows, turn on FSR Ultra... I'm confident you can get +100FPS in most scenarios. Just keep in mind, race starts are always the most challenging (which all of my benchmarks are).
@@eltranza3590I have this exact setup - 9700k overclocked to 4.9, 4080S & 1440p triples. I made the jump recently from 2080 and performance is as expected. Not as high as the results shown in this video, but thats due to CPU. If you tweak the settings as suggested by BenchO below you'll be fine with >120
@@150585Alex no cpu bottleneck with a 4080?
Should have just got the 4090 and been done with.
lol
You should never use surround with iRacing there is no need. Input proper distances etc .ini file it looks and runs better on triples.
There are a few 'use cases' out there. It works better for some people, and I've certainly measured (slightly) better performance with Surround enabled. But I think for a lot of people, yes, you are right.
That’s not true dude - even if I don’t use surround right now, but there is absolutely no difference in performance or looking.
@@chrisracing81That is not true my man do more research. Running surround projects one big screen over the entire space, not using it renders each screen separately, this allows for more freedom with rendering from your P.C and results in a nicer picture more often "I can see the difference right away with separate render im not sure how you can't"
it is often more taxing on the P.C to run it this way than with surround but not by much if at all. Also might be a fix if you need to run one monotir that is low hz. It will alow the rest of the screens to run at their proper FPS. Surround might not hit FPS but it is another program you simply do not need to run. The Sim runs cleaner without the use of surround the picture renders separately, it does have a performance difference, you get more picture options as well, it is not needed at all, not for many years now that is a fact. Great vid by the way.
I’m still waiting for that msfs video that you’ve already managed to include in the intro even though I’ve just bought 4080 super and I love it! 🫡👊
Heh, sorry about the tease ;) Congrats on the new card! It's very fast and I steal the 4080S out of the test bench whenever I can
@@BenchmarkOdysseys I can finally play msfs on quest3 with comfort!