@vulcan4d, any specific requests? While interesting, I don't think there is any point buying anything weaker than the 6800XT. I've been considering a 4070 TI Super or 4080 Super, but am hesitant to make the purchase considering the imminent release of the Nvidia 5000 series supposedly early next year. Also, what do you think viewers would favor? Top-end like the 4090 or more "justifiable" options like the 4070 TI Super and 4080 Super. I feel that, like myself, viewers are liking the visual evidence of GPUs they can afford to purchase.
@@CitizenBenchIf you're gonna add more modern Nvidia options, more modern AMD would make sense to balance. 7900 XTX is in 4080 performance territory and a touch less expensive. 🤷🏿♂️
That looks about right. I would have honestly tried out Vulkan though, as it does have some impact on FPS in some areas, but nonetheless good testing here to get an idea.
@Mistasandman, Thanks. Yes, I did some initial testing with Vulkan but ran into serious issues with Nvidia cards (on Game Environment 3.24.2-live.9381373) - on all three Nvidia cards FPS would tank to ~18 FPS or so on my second run. I mean to retest again on the current version when I get some time.
im 18 seconds in and your testing at 1080p ?!?!??!? Makes sense if your trying to eliminate a gpu bottleneck..but your not benching cpus against each, your testing gpus against each other.
@zeroxception, keep in mind I am bench marking neither the CPU nor the GPU, rather the aim is better understand the performance characteristics of Star Citizen. To that end, the aim was to compare the highest possible FPS performance of each GPUs against each other, hence I used 1080P and low settings. By observing the GPU usage during the runs it becomes clear which combinations are GPU-limited and which are CPU-limited. A future video will show a head to head between the 6800XT and 3060 TI at different resolutions - you might find that more interesting.
AMD's dx11 driver performance is notoriously poor where as Nvidias is excellent. Nvidia figured out a way to use multiple threads on the CPU to split up their driver overhead for dx11. If you watch the GPU utilization for both runs you will see the 6800XT rarely hits 99%, usually 60% - 90%. The 3060ti is maxed the whole time taking full advantage of the GPU. Vulkan will make better use of the 6800XT because its inherently multi-threaded, maxing usage. Id be interested to see if Vulkan would have any effect on Nvidia GPU's compared to dx11, I think they would be within margin of error.
@tavis9002, I'll do a comparison to Vulkan soon. I had started a comparison but all my Nvidia cards were crashing to ~20 FPS after a few minutes - so obviously some bug there that I hope has now been fixed.
Would it be possible for you to test on 24h2 next? Win11 and 24h2 has their own problems but for a CPU limited game like SC new version may actually help, a lot (on AMD CPUs of course). There are very few videos of it already and it's promising but regardless, we need to see them in test as fair as possible and your way of benching them seems to be as good as it gets. It almost impossible to bench in SC equally in each run due to server problems and perk of being always online (AC benching doesnt tell much about PTU). I mean couple crashed ships or some people flying around in C2s and Carracks with fully loaded around can tank 5-10FPS and you cant really do anything about it. Even then with more benching available to us, we can at least figure out an average scenario.
@FlesHBoX, I agree. Unfortunately, I couldn't run the 8700G at 6400MT/s - the EXPO profile alone was unstable and I could only get it working at 6000MT/s by using Ryzen Master which refused every attempt to set it higher than 6000MT/s. I might retry that the next time a motherboard BIOS is released.
nice but very scary the cpu at 100 percent cpu core usage at times is not good at all to think this game seems to bring a 9800x3d to its knees i bought one too havnt used it yet would like to see bigger gpus being used i have a 4070ti + 9800x3d still waiting for motherboard to come
Nice work. I am shocked a 3060ti outperformed a 6800xt. In most games The 6800xt would stomp the nVidia offering. Do you think this is because of poor optimization or Game engine preference?
@savagemadness77, to me it just seems kind of odd. My one thought is that Steve from Hardware Unboxed has mentioned that they have previously found bench marking at 720P problematic because it seems like the GPUs can stall from not enough load. That might be the case here because the GPU isn't running at 100% and yet the there isn't a CPU thread running at 100%, which I have seen in Star Citizen. I'm current looking into how the 6800XT and 3060TI perform at different quality settings and definitely as you increase the quality/resolution the 6800XT starts to perform better than the 3060TI.
I think you're compiling lap times on a racetrack that hasn't been built yet to try and determine what tires to use; trying to "understand the performance characteristics of Star Citizen" you may as well run the tests for a week and chart which DAY got the best FPS. 🍻
@stargeeze8427, I can't disagree. And yet, to gain understanding, the first step is always to start measuring something, to try to gain an understanding of the distribution of the results.
I have almost same fps with my rtx3060 12gb Deepcool ak620 7800x3D g.Skill 2x32 ripjaws s5 cl32-38-38-96 @6000 Kingston fury renegade 2tb Msi b650 gaming plus wifi
@lucienlachence, I'd be interested to know what you get on a similar to run to that shown in the video. Are you familiar with CapFrameX, MSI Afterburner, and RTSS (Riva Tuning Statistics Server)? I suspect, based on other videos on memory scaling, that I'm not getting much from the 6400MT/s RAM.
Comparing an APU to a top end CPU is kinda goofy. Why not compare it to a 7800X3D or something similar? Next, there's no surprise that the 3060Ti is outperforming the 6800XT at 1080p. Look at the GPU usage numbers. The 3060Ti is near max and sometimes at 100%. whereas the 6800XT was in the 80+% range. Bump your resolution up to 1440p and you will start seeing the two swap. This was a CPU test so your results on GPUs are useless. Anyone buying a 6800XT and playing at 1080p doesn't understand PCs in general, unless they play competitive shooters only. These are not 'impressive' gains in gaming, going from a $300 CPU to a $500 product.
@Toutvids, unfortunately I don't have a 7800X3D. Also, I wanted to compare between a weaker AM5 CPU and the 9800X3D as that is the upgrade path more people are likely to take. The next time there is a Star Citizen update, I plan to do a comparison between the 9800X3D and the 5800X3D. Previous testing has found the 8700G out performed the 5800X3D but there is a lot of run-to-run variance so I plan to re-benchmark the 5800X3D next time there is a game update or sooner. Why would you say that you are not surprised the 3060TI is outperforming the 6800XT at 1080P? Yes, they are both CPU-limited at 1080P Low settings, but I would have thought given the well known Nvidia driver overhead that the 6800XT would edge out the 3060TI... They are 'impressive' gains compared against the gains you would see if you invested in a more expensive graphics card dollar for dollar. However, the point of this testing is to identify the upper limit of the number of frames each CPU/GPU combination can provide. In the near future, I will be performing a comparison of the 3060TI and 6800XT at various resolutions and quality settings to better understand the difference between those two cards.
@@CitizenBench fair enough, but I would think it is a no brainer to figure there would be a performance increase going from a 8700G to a 9800X3D. But I see that you have quantified that in this video. The 6800XT is better at 1440p gaming more than 1080p. Now if you upped the details to max at 1080p then the 6800XT might be finally running 100%. The 3060Ti was already near 100% most of the time in your tests. My previous card was a 6800XT and at 1440p is was fantastic, I only upgraded because I am running 5120 x 1440p now. Yeah once you start increasing settings and resolutions the 6800Xt will leave the 3060Ti in the dust. Problem will be that star citizen is not consistent enough to use as an actual benchmark.
No, you're right. According to the steam hardware survey far more people are actually using the 1650! ;) If you think most gamers are using a 3080 or better there is plenty of evidence to suggest you are mistaken.
What the fuck? Why on god's green earth would you pair such lowly low class GPUs with that CPU? Isn't the whole point of the CPU is to see the maximum frames possible in a game like this?
@Stardomplay, Great question! I initially started out doing this to answer the eternal question: "What should I buy next to increase my Star Citizen FPS?". To this end, I bench marked every GPU I had available to me on my 5800X3D and was surprised to see that both the 3060 TI and 6800XT were CPU-bound at low settings. At that point, it became clear that the best way forward was to move to the AM5 platform. At this point, I had become very interested in bench marking Start Citizen, and also the latest Cargo update broke my trade route of choice, so I "invested" in an AM5 system with an 8700G as I was wanted to also benchmark the 780M integrated graphics - in preparation for the 9800X3D when it was released. So like many people - I am now left contemplating whether to buy now or wait until the Nvidia 5000 series and AMD 8000 series are released. And if so, do I go AMD or Nvidia? I have been extremely surprised that the 3060 TI is beating the 6800XT, and am wondering whether that is due to CIG starting to use Ray-tracing or whether it just due to me testing at low settings. My next video will compare the 6800XT and 3060 TI at various quality settings at 1080P and 4K upscaled.
@@CitizenBenchisn’t it widely known that SC is heavily CPU or rather memory bottlenecked. There are other SC benchmark channels and it is clear that the 13900/14900k from intel with HIGH memory speeds (tuned) perform the best.
@JaceFunkyFace, did you watch through the entire video? The first two GPUs are GPU-limited (1660Super and 2070), the last three (6600XT, 6800XT, and 3060TI are CPU-limited). I will next be comparing the 6800XT and 3060TI at different resolutions and quality settings to elucidate when we cross from CPU-limited to GPU-limited.
@@CitizenBench Yes, you can see that. The first two GPUs do not budge from 100% usage, whilst the following ones start to go down from 100%. Still the 9800X3D pushes those GPUs hard... Only game I seen that can use 14GB+ VRAM and around 28GB+ DRAM..... Damn crazy....
@@jrlivingspaces 4070TI Super looks pretty decent, much better power consumption too. You'll see a 4090 on your electricity bill and the price is just greedy.
thanks for taking the time to make this video!
@Palmeroide, thanks for you kind words.
This is a nice benchmark video that I never knew I wanted. Very well done!
@Cmusch05, thank you for your encouragement. It's always a bit daunting trying something that is different to the norm.
Now this is how you benchmark Star Citizen. You got a sub and I hope you will be able to get other GPUs to test eventually.
@vulcan4d, any specific requests? While interesting, I don't think there is any point buying anything weaker than the 6800XT. I've been considering a 4070 TI Super or 4080 Super, but am hesitant to make the purchase considering the imminent release of the Nvidia 5000 series supposedly early next year. Also, what do you think viewers would favor? Top-end like the 4090 or more "justifiable" options like the 4070 TI Super and 4080 Super. I feel that, like myself, viewers are liking the visual evidence of GPUs they can afford to purchase.
@@CitizenBenchIf you're gonna add more modern Nvidia options, more modern AMD would make sense to balance. 7900 XTX is in 4080 performance territory and a touch less expensive. 🤷🏿♂️
and here i am using an I7 4770k with an 3080ti
@89HondaCR500R, well the good news is that from my testing it seems that all you need is a CPU upgrade.
how much fps on Lorville with that setup?
That looks about right. I would have honestly tried out Vulkan though, as it does have some impact on FPS in some areas, but nonetheless good testing here to get an idea.
@Mistasandman, Thanks. Yes, I did some initial testing with Vulkan but ran into serious issues with Nvidia cards (on Game Environment 3.24.2-live.9381373) - on all three Nvidia cards FPS would tank to ~18 FPS or so on my second run. I mean to retest again on the current version when I get some time.
im 18 seconds in and your testing at 1080p ?!?!??!?
Makes sense if your trying to eliminate a gpu bottleneck..but your not benching cpus against each, your testing gpus against each other.
@zeroxception, keep in mind I am bench marking neither the CPU nor the GPU, rather the aim is better understand the performance characteristics of Star Citizen. To that end, the aim was to compare the highest possible FPS performance of each GPUs against each other, hence I used 1080P and low settings. By observing the GPU usage during the runs it becomes clear which combinations are GPU-limited and which are CPU-limited. A future video will show a head to head between the 6800XT and 3060 TI at different resolutions - you might find that more interesting.
thanks for the video, amazing.
@pietro_s032, thank you for the encouragement.
AMD's dx11 driver performance is notoriously poor where as Nvidias is excellent. Nvidia figured out a way to use multiple threads on the CPU to split up their driver overhead for dx11. If you watch the GPU utilization for both runs you will see the 6800XT rarely hits 99%, usually 60% - 90%. The 3060ti is maxed the whole time taking full advantage of the GPU. Vulkan will make better use of the 6800XT because its inherently multi-threaded, maxing usage. Id be interested to see if Vulkan would have any effect on Nvidia GPU's compared to dx11, I think they would be within margin of error.
@tavis9002, I'll do a comparison to Vulkan soon. I had started a comparison but all my Nvidia cards were crashing to ~20 FPS after a few minutes - so obviously some bug there that I hope has now been fixed.
Would it be possible for you to test on 24h2 next? Win11 and 24h2 has their own problems but for a CPU limited game like SC new version may actually help, a lot (on AMD CPUs of course). There are very few videos of it already and it's promising but regardless, we need to see them in test as fair as possible and your way of benching them seems to be as good as it gets. It almost impossible to bench in SC equally in each run due to server problems and perk of being always online (AC benching doesnt tell much about PTU). I mean couple crashed ships or some people flying around in C2s and Carracks with fully loaded around can tank 5-10FPS and you cant really do anything about it. Even then with more benching available to us, we can at least figure out an average scenario.
@adilaimdegilx - On Windows 11 24H2? I'll see what I can do.
9800x3d is a brilliant CPu but it's so expensive
Would have been nice to have the 8700G vs 9800X3d results using the same ram configuration, since SC loves memory speed.
@FlesHBoX, I agree. Unfortunately, I couldn't run the 8700G at 6400MT/s - the EXPO profile alone was unstable and I could only get it working at 6000MT/s by using Ryzen Master which refused every attempt to set it higher than 6000MT/s. I might retry that the next time a motherboard BIOS is released.
nice
but very scary
the cpu at 100 percent cpu core usage at times is not good at all
to think this game seems to bring a 9800x3d to its knees
i bought one too
havnt used it yet
would like to see bigger gpus being used
i have a 4070ti + 9800x3d
still waiting for motherboard to come
Nice work. I am shocked a 3060ti outperformed a 6800xt. In most games The 6800xt would stomp the nVidia offering. Do you think this is because of poor optimization or Game engine preference?
@savagemadness77, to me it just seems kind of odd. My one thought is that Steve from Hardware Unboxed has mentioned that they have previously found bench marking at 720P problematic because it seems like the GPUs can stall from not enough load. That might be the case here because the GPU isn't running at 100% and yet the there isn't a CPU thread running at 100%, which I have seen in Star Citizen. I'm current looking into how the 6800XT and 3060TI perform at different quality settings and definitely as you increase the quality/resolution the 6800XT starts to perform better than the 3060TI.
I think you're compiling lap times on a racetrack that hasn't been built yet to try and determine what tires to use; trying to "understand the performance characteristics of Star Citizen" you may as well run the tests for a week and chart which DAY got the best FPS. 🍻
@stargeeze8427, I can't disagree. And yet, to gain understanding, the first step is always to start measuring something, to try to gain an understanding of the distribution of the results.
Although I really appreciate the effort, Star Citizen is not a great title to test performance. Server instability highly affects performance.
@Ricback2, yes I have noticed that but unfortunately Star Citizen is the game that I want to play.
@CitizenBench I wonder how those frames will he impacted in Pyro.
I'm just happy I can finally walk around my Polaris 😊
Why are you running windows 10??????
I have almost same fps with my rtx3060 12gb
Deepcool ak620
7800x3D
g.Skill 2x32 ripjaws s5 cl32-38-38-96 @6000
Kingston fury renegade 2tb
Msi b650 gaming plus wifi
@lucienlachence, I'd be interested to know what you get on a similar to run to that shown in the video.
Are you familiar with CapFrameX, MSI Afterburner, and RTSS (Riva Tuning Statistics Server)?
I suspect, based on other videos on memory scaling, that I'm not getting much from the 6400MT/s RAM.
@CitizenBenc 6400 is maximum only if u go more motherboard will change to 1:2 and u will get less fps
@CitizenBench ill check with that program capframex what i have in terms of fps
Comparing an APU to a top end CPU is kinda goofy. Why not compare it to a 7800X3D or something similar?
Next, there's no surprise that the 3060Ti is outperforming the 6800XT at 1080p. Look at the GPU usage numbers. The 3060Ti is near max and sometimes at 100%. whereas the 6800XT was in the 80+% range. Bump your resolution up to 1440p and you will start seeing the two swap. This was a CPU test so your results on GPUs are useless. Anyone buying a 6800XT and playing at 1080p doesn't understand PCs in general, unless they play competitive shooters only.
These are not 'impressive' gains in gaming, going from a $300 CPU to a $500 product.
@Toutvids, unfortunately I don't have a 7800X3D. Also, I wanted to compare between a weaker AM5 CPU and the 9800X3D as that is the upgrade path more people are likely to take. The next time there is a Star Citizen update, I plan to do a comparison between the 9800X3D and the 5800X3D. Previous testing has found the 8700G out performed the 5800X3D but there is a lot of run-to-run variance so I plan to re-benchmark the 5800X3D next time there is a game update or sooner.
Why would you say that you are not surprised the 3060TI is outperforming the 6800XT at 1080P? Yes, they are both CPU-limited at 1080P Low settings, but I would have thought given the well known Nvidia driver overhead that the 6800XT would edge out the 3060TI...
They are 'impressive' gains compared against the gains you would see if you invested in a more expensive graphics card dollar for dollar. However, the point of this testing is to identify the upper limit of the number of frames each CPU/GPU combination can provide. In the near future, I will be performing a comparison of the 3060TI and 6800XT at various resolutions and quality settings to better understand the difference between those two cards.
@@CitizenBench fair enough, but I would think it is a no brainer to figure there would be a performance increase going from a 8700G to a 9800X3D. But I see that you have quantified that in this video.
The 6800XT is better at 1440p gaming more than 1080p. Now if you upped the details to max at 1080p then the 6800XT might be finally running 100%. The 3060Ti was already near 100% most of the time in your tests. My previous card was a 6800XT and at 1440p is was fantastic, I only upgraded because I am running 5120 x 1440p now.
Yeah once you start increasing settings and resolutions the 6800Xt will leave the 3060Ti in the dust. Problem will be that star citizen is not consistent enough to use as an actual benchmark.
Um bro, please use modern GPU’s. Nobody is rocking a 1660. 3080 should be your starting point.
No, you're right. According to the steam hardware survey far more people are actually using the 1650! ;) If you think most gamers are using a 3080 or better there is plenty of evidence to suggest you are mistaken.
What the fuck? Why on god's green earth would you pair such lowly low class GPUs with that CPU? Isn't the whole point of the CPU is to see the maximum frames possible in a game like this?
@Stardomplay, Great question! I initially started out doing this to answer the eternal question: "What should I buy next to increase my Star Citizen FPS?". To this end, I bench marked every GPU I had available to me on my 5800X3D and was surprised to see that both the 3060 TI and 6800XT were CPU-bound at low settings. At that point, it became clear that the best way forward was to move to the AM5 platform. At this point, I had become very interested in bench marking Start Citizen, and also the latest Cargo update broke my trade route of choice, so I "invested" in an AM5 system with an 8700G as I was wanted to also benchmark the 780M integrated graphics - in preparation for the 9800X3D when it was released.
So like many people - I am now left contemplating whether to buy now or wait until the Nvidia 5000 series and AMD 8000 series are released. And if so, do I go AMD or Nvidia? I have been extremely surprised that the 3060 TI is beating the 6800XT, and am wondering whether that is due to CIG starting to use Ray-tracing or whether it just due to me testing at low settings. My next video will compare the 6800XT and 3060 TI at various quality settings at 1080P and 4K upscaled.
@@CitizenBenchisn’t it widely known that SC is heavily CPU or rather memory bottlenecked. There are other SC benchmark channels and it is clear that the 13900/14900k from intel with HIGH memory speeds (tuned) perform the best.
You are 100% GPU bound, this is test is meaningless…
@JaceFunkyFace, did you watch through the entire video? The first two GPUs are GPU-limited (1660Super and 2070), the last three (6600XT, 6800XT, and 3060TI are CPU-limited). I will next be comparing the 6800XT and 3060TI at different resolutions and quality settings to elucidate when we cross from CPU-limited to GPU-limited.
@@CitizenBench Yes, you can see that. The first two GPUs do not budge from 100% usage, whilst the following ones start to go down from 100%. Still the 9800X3D pushes those GPUs hard... Only game I seen that can use 14GB+ VRAM and around 28GB+ DRAM..... Damn crazy....
cpu make no different in game . save your money to rtx 4090
Or 4080
@@jrlivingspaces 4070TI Super looks pretty decent, much better power consumption too. You'll see a 4090 on your electricity bill and the price is just greedy.
@RillUK, yes but 5000 series is sooooooo close!
@@CitizenBench WHEN WHEN??? lol