haha, who ever knew gpus could make different colours. I actually only noticed it myself last week when I replaced the PS4 with the PS5 and everything was completely different colour wise.
watching in 4K I think the nvidia had richer colors and sharper image most of the time. But in Dirt 5 everything was dark. I have a calibrated monitor and my vision is corrected to 20/12 acuity
It differently looks like the benchmark has time of day different each run. Meaning it could be morning, noon, or evening time each run making the shadows and texture darkness different.
@@zhongxina2614 I don't know the game but it looks to me like a cloud was covering the sun on the right-side video (RTX 3080), I guess that would fall under dynamic "weather".
Averaging out benchmarks still gives a more realistic number for consumers who only play once, as there are many, many people playing and all essentially getting random variation on their one play-through. Averaging is a good tool to even out outliers. Bloatware should never be present in benchmark systems and they should always be checked to run decently anyway.
my thought´s exactly, even if just playing a scene just once, the benchmark will give a mean to what gpu would handle that scene the best at any given scenario.
But a benchmark is meant to be apples to apples, Forza runs through the different seasons repeatably so every benchmark looks the same but is being rendered uniquely
@@spusuf Dirt 5 is clearly is a poor choice for benchmarking, it looks like there was also variation in the weather effects and car positioning between runs too. Tech Yes saying he's not gunna use it anymore is definatly for the best.
Great video you & Steve over at Gamer's Nexus are the only reviewers I really appreciate I get the most information out of you two and I actually learned how to build a computer from one of your videos back in the day
I agree some what At Tech yes you get a fast yet a lot of info which is great for sure. I have started to straw away from GN for reviews on some stuff though. I am finding that Steve over at GN is getting a bit full of himself some times. Things like you all are doing it wrong but we are not. Who is to say the whole review industry is doing it wrong but his way is right. The final straw for me at least was for the AMD 6000 launch it seemed like on review day either he was having a really bad day or he had a butt corn stuck in his butt for AMD. He was trying to find ways to dis AMD's new cards. Where he was saying shit like under performing cooler everyone else was saying the coolers on the AMD cards were top notch and performed great. Heck I think even Mr Yes said the AMD coolers were great. He (GN) was also nit picking on other things as well for the AMD cards. I never even finished watching his 6800 review as I quit about half way through it and came to Tech Yes to get a fair and no nit picky review after all all I wanted to see was the numbers not watch someone spend 30 minutes try to rip the product apart for no reason at all. So for me it is Hardware Unboxed and Tech Yes City as 2 of my go to places for Day one reviews as well as a hand full of written review sites. I think Steve's hair night be whispering to many things into his ears and the fact everyone calls him Tech Jesus may be going to his head. I used to watch his videos all of the time but the last year he seems to be either disliking his YT job or it all is starting to go to his head and he needs to step back and do a realility check and figure out he is not actually the Tech Jesus of YT & there are far more people on YT with vast more no how than him. It may seem like I am picking on him but really I am not. I just hate the whiny attitude he has developed these past couple of years.
I have observed image quality between AMD/Nvidia cards for the past few years, and there IS a slight color vibrancy of AMD over Nvidia cards. I have seen it between Vega 64 / 980ti, and recently 5700XT / 2080ti. Doing a comparison mainly in SDR rather than HDR. HDR looks stunning on the 2080ti@4k 60fps. I even see that same color vibrancy on PS4 (AMD gpu) compared to PC 2080ti. There may be a way to adjust the Nvidia settings by the color vibrancy panel, but I have not been able to match it exactly. The difference is subtle between the two but definitely noticeable when you switch from Nvidia to AMD.
Thank you for this. I've been looking at a lot of comparison videos between the 6800XT and 3080, but almost all of them are devoted to frame rates. The frame rates are quite high for both cards and image quality will probably be a more important factor (right after availability and price). Yours is the only video I've seen that focuses on image quality.
To my eyes though, the Nvidia card seemed to show more detail sometimes and a little sharper. There is a slight difference in color tone too if you look back at Tomb Raider. Lara's skin tone and tree branches.
Tooo awesome I am totally gonna get a 60hz freesync 4k lg 32 inch monitor for Christmas. For when I can get a 6800XT next Spring. Love this generation of gaming tech. :D
Is Preferable to run 3 times the benchmark than just once. Why? Because it will give us the whole image and what truly these GPU can do. Ok, you are indeed true with the 1 run time for the player but a lot can happen in 1 run/benchmark, a fail, something goes wrong and etc. So, multiple repeats of the benchmark help you to avoid these things while a single benchmark simulates better what the player might see. Either way, keep the good job going. My opinion is to keep what you doing but still report if you see a fail or something like that in a benchmark but I think you are doing this already too.
Hey mate, Looking at the nvidia card in dirt there isnt shadows for someone of the things above (looks like ropes) almost looks like it's a different day mode?
I used to use DiRT 3 for benchmarking a few years back and found it quite inconsistent..... the AI just caused so many FPS variations. Like if a car blocked the main car in one place and then the next run it didn't and the score would be different by the end.... so I think it's the nature of a racing sim with AI for the other cars. Probably similar to the difference that people notice in online multiplayer vs bot matches. It's the AI aspect that causes the variations...... I wonder whether the AI is being powered by the GPU, thus causing the variations? Anyways, I can really appreciate the amount of time and effort you put into videos like this and then when you get inconsistencies like that it's a real headache and time consuming to sort out because if you just post inconsistent results then people will just pick holes in it.
The old HD 6000 and 7000 were really good cards and they did put a gorgeous fight against nvidia back in the day. It's really emotional to see AMD putting up a fight again after many years of struggle with this 6000 line up.
Could we please get some testing of the new smart access memory function AMD were talking about when combining the new 6000 series cards with the zen 3 processors? Keen to see if there are any productivity gains to be had from an all red system as opposed to zen3 + 3080?
It varies between games, at the moment it only aids a few titles. That makes sense as the benefits come from when the game knows it doesn't have to manage assets through a 256MB Aperture. You can see the effect in other reviews like Gamers Nexus
Yay! Stuff to watch at work! E: I think that 3 run plan is better if you have the time for it. There are things that can magically go wrong for some reason so 3 runs is pretty good. There is absolutely something odd in the Dirt5 and I hope they havent done stupid tricks to make fps better and try look better. Good thing you dropped that!
There's a new encoder in the 6000 series and OBS hasn't added support for it yet. It is still using the old AMD encoder. Read the OBS release notes. It's not an AMD driver update, it's an OBS update that is required. The 6000 series contains a completely different encoder adding support for new formats.
You dont happen to have power draw comparisons too do you? I see a lot of video showing the 3080s taking 50-60 more watts compared to the 6800xt. Just wondering if so and if it was a factor.
Hey man. Just wanted you to know your channel specifically is what got me into PCs in the first place. The whole price to performance bit. Yeah...I've been drinking a little bourbon while watching, so I feel pretty appreciative, but I wanted to show you some love. lol
@@WoutieX its still the right move to exclude it, if the whole image is covered in shadow on one side and not the other its fair to assume the rendering load is different on each, its easy to dismiss it as just darker on the 3080 side but if all that darkness is shadow that takes extra gpu time to render its outright not an equal test. benchmarks need to be for the most part exactly the same if he is to compare hardware, its one of the reasons reviewers allow for 1 or 2 fps each way and call it negligible, it likely comes down to extra npc's or something in the test from run to run so if the cards come within a couple fps of each other its as good as a draw unless one card is consistently in the lead run after run then its likely just a touch faster in that game.
For me there was one other difference. Its the dynamic range. The rx 6800 xt has a little less dynamic range in instances shown. Weather conditions can account for color but there is no excuse for poor dynamic range.
I can see a difference around 5:00 and before. The fidelity of objects in the distance on the 3080 is noticeably worse. You can see it again at 7:48-7:49, the definition in the stone work on the RTX card looks blurred as you get closer. Neither of those examples are due to the lighting.
@@chrisallen6010 There does appear to be more shadows with the stone part. But I notice the amount of stuttering with some of the footage. This was never fixed in the 5700xt. But I believe that the guy utilized nvenc encoder so maybe that caused it since its an amd card
Interesting colors are all the same across the board, at least with older AMD/ATI cards vs Nvidia you could always see that one of the cards are a bit more vibrant with color? Or is it due to the recording that it gets filtered out if that is the correct term? Or is it another factor that does this?
It's hard for me to say if one run vs. 3 run average is best without knowing how each run performs. Is the first run usually slower than the other runs? If so, does the performance drop change between cards? Showing results for a few games would help think about it more. Thanks for the video. The Dirt 5 case looks strange and it would be nice to see if the devs have anything to say about it.
At 0:16 I can read the Ford emblem on the right of the trunk lid and the word Cosworth on the left. With RUclips's compression I am not certain if that makes a difference. [ I stopped the video at that point to type a comment, not knowing it would be pointed out at 0:29 ] Speaking of the latter time frame I can easily read the smaller print on the license plate generated by the RX 5700 XT: codemasters At 3:50 shadows are darker on the RTX 3080, not certain if anyone would pay attention running a racing game. At 3:57 the sun showing through the windshield with the RX 6800 XT is cleaner, crisper than on the RTX 3080. At 3:55 there are ripples in the water with the RX 6800 and none with the RTX 3080, a glaring omission. At 4:08 there is a great deal of water splash with the RX 6800 XT and none when the Ford on the right hits the water RTX 3080. I stopped commenting at that point. Didn't want to make this comment a short book. [EDIT: I watched it on my computer monitor at 4K, a 55" Samsung Q80T two feet from my face.] And, I did watch your clip from Horizon Zero Dawn in HDR 10bit. It suffered from RUclips compression, it still looked good though. I think doing benchmarks 3x yields more accurate comparisons, rebooting the computer for each run. Thank you for this review.
I noticed the clouds are different in dirt 5. maybe the engine have a random volumetric cloud generator that made it impossible to get a an exact same cloud coverage condition?
Good video. On the benchmarking thing. When running a benchmark test on a game where there is some random factors to exactly what takes place, then doing multiple runs absolutely makes sense sine there is no way of knowing if just one run represents a best case, a worse case or somewhere between those two outer points. So please multiple runs and also letting it be know, how many runs was done and the highest and lowest scores.
Yeah benchmarks like Dirt are great for checking your settings to ensure you are going to get a good gaming experience, particularly given gaming is somewhat random. No good for comparing GPUs/Hardware though. Which is kind of daft because you can achieve both through a consistant Bench.
dynamic clouds system that they somehow didn't sync to be the same in the benchmark. How much actual difference that makes only the developers may know since that's heavily dependant on how the cloud shadows are displayed (culling, batching, etc.).
7:15 how about run benchmark then restart pc, then run benchmark again instead This might be the better middle ground of consistency and first run benchmark Turn off fast boot to flush the memory too
The only thing I can think of is, most games that are tested will look very close because of GPU drivers. The cards are basically drawing the same image, and we are testing the speeds and visual acuity of the games. Call of Duty Black Ops cold war, shadow of War, Ashes of the singularity... theres tons of games that would be really interesting to see these cards run. ALSO! Can you test multi-monitor performance of graphics cards? Get a counter on say, City Skylines, Jurrasic World, and Sims 4, and sims city to see how these cards handle multiple work flows? Its because, I've noticed that switching from Pascal (gtx 1080) to Turing (rtx 2060 super) that the Turing card sucks at playing multiple games at once. I liked having simulations running in the background on different screens and I think this is something people should see maybe.
Has the color range been changed from 16-255 to 0-255 on the nvidia card drivers ? Saw that in old video about image quality that you need to do this so we have apples to apples comparison !
Good point man I'd have to confess I do like all things technology including video games however after RUclips video games are what I spend most of my free time doing.
"PC gamers aren’t real gamers"... Speak for yourself, mate. Every bone in my body is a "real gamer". In fact, while my left hand fingers are typing this very comment, my right hand fingers are playing PC games! Yes, I am also, a "tech enthusiast".
True gamers play consoles they don't care about high fidelity graphics as long as it runs ok on ps4 and xbox one we play singleplayer games more we don't play trash fps games like call of duty fortnite pubg and trash doom eternal mad respect for you
I have a question tomyou or the commenters here. Do you get better shadows with less flickering on one or another card? AMD vs NVIDIA? My xfx 6700xt has a bad cooler and some coil whine and was thinking to get it refunded and get an 3070 Gigabyte gaming oc. When the card runs it is a fast one for my 1440p screen. I unfortunetly seen some stuff in Uncharted that are not that nice. Thinking on an driver issue. Please be kind in response, would love some help and guidence.
Nope it's not processing a lot more shadows, just cloud shadows moving through the circuit. My guess is that cloud shadows don't really obey a common random seed when generated between benchmarks, so they show up in different positions.
For a comparison, the users’ first runs could be less FPS than your first run. The best is to run the benchmark three times and to include the standard deviation of the three runs on your charts. Excel has a feature for adding STD bars on bar charts. Bloatware May decrease performance, but it will, hopefully, be systemic, affecting all tests the same.
You need multiple runs to detect outliers, where for example some OS task degrades benchmark processing. Those outliers ought be discarded not averaged. A first run could be slower due to loading or faster due to better thermals. In general GPU benchmarks are expected to show gfx performance not a general game experience including loads. A good in game benchmark probably pre-loads before it starts recording results. A standard deviation would require a much larger number of runs to determine confidence properly. Just reporting your consistent results on a correctly functioning system suffices, IMO
Will you make a video about ryzen apu-s in 2020 I'm much more interested in overclocked versions something like a ryzen 5 2400g overclocked in 2020 1080p low gaming ... Something like that i would like to see, tho I still enjoy the content that you to offer here on the chanel 👍🏻
About the "one run" or "three run" benchmarking thing, think about it that way: if you play for example an open world rpg, there will certainly be places that you visit more often than once, even more often than three times (witcher, escape from tarkov, wow, ...) even in other titles if you think about it (shooters, racing games, ...) that's why i think that you should take the avg. of three runs, as there is (imo) not really any game where you are going to visit one place only once in your whole playtime
Adaptive sync will adjust the refresh rate to the FPS. Aka for freesync yes, it bottoms out at 35hz/fps usually for freesync panels. Since the 6800xt slaps at 1440p, thats a hell yes. Another point is it will be less cpu intensive then 1080 240-360 hz/fps. The more fps your gpu renders, the harder the cpu has to work in order to keep up with them.so ipc would be key for that. Lastly some games have caps that are lower then 240 fps anyway, doom 2016 tops out at 200 fps.
I liked this video, It may be a driver issue with Dirt5. The bottom line is people have VERY HIGH expectations of a $650 - $700 video card. Yeah, if these cards continue to stay unavailable. Many of us customers will just buy an an LG OLED TV (they are in stock). They cost the same as the high end video cards. I think they will make our current gen cards look even better. Maybe you can do another side by side with current gen gpus on OLED but w/o fps meters at first so we can see with just our eyes, then overlay the fps counter. That way it is the gameplay experience vs. just numbers on a chart. Thanks for all your hard work.
I had felt that there was such a difference before, but recently I have been able to feel that difference personally and notice it more clearly. I had an RX580, I used it for a while, I liked it, but when I went back to my GTX 1060 that I had left idle, I felt that visually it looks more immersive and beautiful. AMD have a more vibrant coloring, but Nvidia has a more immersive shading, the lights and shadows on Nvidia give a more immersive and good look, AMD has this vibrant coloring but it declines in this feature that I mentioned Nvidia has, ending up with AMD giving kinda of an artificial feeling to the visuals even though it has good colors. Good video. I like that you used a extern way to record both.
To me, i noted a couple of areas in the last bench mark with the nvdia showing light or shadow in a couple of different areas. Although for the most part the radeon images looked sharper over all. Nvdia had a history of blurring elements in frames, so i heard, years ago. Reason being for higher fps. If thats is still relevant today every so often do a space bar pause a comparison sided by side screen, see it your way.
@@SirZanZa AMD always was some very little shuttering, always... Nvidia havent these problems. Mby 5000 and 6000 series AMD havent that, dont know, but people are talking about that!
In the dirt benchmark, the dark areas would appear to be from a cloud blocking the sun which if you look at the sky both runs have different cloud formations. The question is will that have a significant effect on the benchmark, especially if you do a three run average. I did see the Radeon card get darker than the Nvidia card at different times, so this effect would have a similar affect either way. One issue with static benchmarks is that those are more prone to manipulation that benchmarks with randomised effects. Rather than throw out the benchmark, it's a good thing to have variables that will average out like what this seems to be.
yeah this is questionable, so many shills these days you can't trust anybody...go look at jayztwocents that clown has his whole top shelf filled with nvidia cards while doing an amd review, what a clown. steve from gamers nexus already said nvidia wants their cards seen in the background when their sponsored shills do amd reviews...here's another youtuber I won't watch...
I would think it's good to know how wide the cards vary in fps. If one run you average 140 fps then the next time it's 115 fps that's a big difference. If it's only 5 fps different no big deal. What card to you think will be a better card in future games?
@Tech YES City bro I have recently build a Desktop , But there was some shortage of 3600 So I had build Using 4650g & rtx 2060 , so is it a bad combo plz reply, by the way big fan love from India 🇮🇳 🧡💚🧡
I don't hink the criticism of dirt 5 is warranted, the brightness of the light shouldn't affect the number of shadows and the computational load, it should only change the variables in the compute. A good benchmark should have a low run to run variance in results, it doesn't need to present the exact same scene, especially if you do multiple runs.
I have been interested in the 6800XT recently and I will admit I clicked on this to see that they're identical, and gloat about the parity AMD has struck with Nvidia, but sadly during the SotTR benchmark I can clearly see a slight stuttering as the background pans and the 6800XT when compared to a relatively butter smooth movement of the 3080. I probably wouldn't have noticed if they weren't side by side. Contrast looks better on the 3080. Granted we're talking a 5% difference in total fidelity. But when they're priced similarly why not get the one that looks better. The comparison helps.
Ah but there is no comparison to a Tech YES Loved GPU pulled out of a scrapyard.
Parts hunt coming soon enough ;)
@@techyescity Tech YES Price-Performance!
Day 13 of replying to marco
The 6800xt showed more frame hitching? Re dirt5 Raytracing can artistically negatively affect lighting unless it is artistically controlled.
@@beingatliberty Did it, though? Isn't that the fault of the encoding PC?
At 2:07
Brian: the image is looking identical to my eyes (in 4k)
Me: Ah yes totally agree (in 480p)
Here I'm watching on 360p sure its looks identical to my eyes ;)
Look at dirt5, in the 3080 there are some areas that are clearly darker than the one on the 6800xt, and i am watching from the mobile
haha, who ever knew gpus could make different colours. I actually only noticed it myself last week when I replaced the PS4 with the PS5 and everything was completely different colour wise.
watching in 4K I think the nvidia had richer colors and sharper image most of the time. But in Dirt 5 everything was dark. I have a calibrated monitor and my vision is corrected to 20/12 acuity
@@chatch15117 also watching this on a 4k 60hz monitor and i fully agree with you.. in my opinion, Nvidia is still king in gpu's.
I haven't seen this analysis anywhere else but here. All other reviews have focused on FPS or similar performance elements. Great job, as usual.
It differently looks like the benchmark has time of day different each run. Meaning it could be morning, noon, or evening time each run making the shadows and texture darkness different.
Instructions unclear. Hit like button instead.
Looks like dynamic weather in dirt5. Clouds are different.
He's talking about the *DETAILS* and color but that has nothing to do with what ur saying.
Did u even watch the video?
@@zhongxina2614 I don't know the game but it looks to me like a cloud was covering the sun on the right-side video (RTX 3080), I guess that would fall under dynamic "weather".
@@zhongxina2614 he literally talks about shadows... And says "ignore the colour".
Wow, you're a bit of a cock aren't you?
@@zhongxina2614 it does has an effect, just look the effect of HDRI on details when you render 3d
Averaging out benchmarks still gives a more realistic number for consumers who only play once, as there are many, many people playing and all essentially getting random variation on their one play-through. Averaging is a good tool to even out outliers. Bloatware should never be present in benchmark systems and they should always be checked to run decently anyway.
my thought´s exactly, even if just playing a scene just once, the benchmark will give a mean to what gpu would handle that scene the best at any given scenario.
Dirt 5 has a day/night cycle. Is it (partly)responsible for the disparity?
But a benchmark is meant to be apples to apples, Forza runs through the different seasons repeatably so every benchmark looks the same but is being rendered uniquely
@@spusuf Dirt 5 is clearly is a poor choice for benchmarking, it looks like there was also variation in the weather effects and car positioning between runs too. Tech Yes saying he's not gunna use it anymore is definatly for the best.
Same with RDR2. It has a day/night cycle, and the performance difference can be upto 5-10 fps depending upon the time of the day.
@@raijin2950 also AMD KILLS Nvidia in this one, so scraping it will give advantage to the latter.
There's also time-of-day variation in Horizon Zero Dawn. No idea if it impacts the performance
Great video you & Steve over at Gamer's Nexus are the only reviewers I really appreciate I get the most information out of you two and I actually learned how to build a computer from one of your videos back in the day
I agree some what At Tech yes you get a fast yet a lot of info which is great for sure. I have started to straw away from GN for reviews on some stuff though.
I am finding that Steve over at GN is getting a bit full of himself some times. Things like you all are doing it wrong but we are not. Who is to say the whole review industry is doing it wrong but his way is right.
The final straw for me at least was for the AMD 6000 launch it seemed like on review day either he was having a really bad day or he had a butt corn stuck in his butt for AMD. He was trying to find ways to dis AMD's new cards. Where he was saying shit like under performing cooler everyone else was saying the coolers on the AMD cards were top notch and performed great. Heck I think even Mr Yes said the AMD coolers were great. He (GN) was also nit picking on other things as well for the AMD cards.
I never even finished watching his 6800 review as I quit about half way through it and came to Tech Yes to get a fair and no nit picky review after all all I wanted to see was the numbers not watch someone spend 30 minutes try to rip the product apart for no reason at all.
So for me it is Hardware Unboxed and Tech Yes City as 2 of my go to places for Day one reviews as well as a hand full of written review sites. I think Steve's hair night be whispering to many things into his ears and the fact everyone calls him Tech Jesus may be going to his head. I used to watch his videos all of the time but the last year he seems to be either disliking his YT job or it all is starting to go to his head and he needs to step back and do a realility check and figure out he is not actually the Tech Jesus of YT & there are far more people on YT with vast more no how than him. It may seem like I am picking on him but really I am not. I just hate the whiny attitude he has developed these past couple of years.
Haven't been around since 2016, ur whole production is just on another level now, good shit
I have observed image quality between AMD/Nvidia cards for the past few years, and there IS a slight color vibrancy of AMD over Nvidia cards. I have seen it between Vega 64 / 980ti, and recently 5700XT / 2080ti. Doing a comparison mainly in SDR rather than HDR. HDR looks stunning on the 2080ti@4k 60fps. I even see that same color vibrancy on PS4 (AMD gpu) compared to PC 2080ti. There may be a way to adjust the Nvidia settings by the color vibrancy panel, but I have not been able to match it exactly. The difference is subtle between the two but definitely noticeable when you switch from Nvidia to AMD.
Thank you for this. I've been looking at a lot of comparison videos between the 6800XT and 3080, but almost all of them are devoted to frame rates. The frame rates are quite high for both cards and image quality will probably be a more important factor (right after availability and price). Yours is the only video I've seen that focuses on image quality.
Yay! The madman is back!
To my eyes though, the Nvidia card seemed to show more detail sometimes and a little sharper. There is a slight difference in color tone too if you look back at Tomb Raider. Lara's skin tone and tree branches.
I have been waiting for a review on this, Thankyou for your hard work testing for us.
Finally got a 5600X, MSI X570 Tomahawk and a Tuf 3080. My body is ready for Cyberpunk 2077.
Delayed again till April 2021
You should of got a 3090 for the 8k experience. Enjoy you sub-par build though.
@@paulfinch1083 LMAO
@@SteveL11 I'm joking obviously. Enjoy your new build. I tried to get a 6800XT on launch day. It was a joke. XD
@@paulfinch1083 Yeah, it's impossible to get these things. So annoying. Keep trying though. Join Stockdrop Discord, it helped me.
I'm so excited for the 6800XT! I just need to find one lol...
Lol
That flag on the thumbnail took me by surprise! Greetings from Greece! 🇬🇷
Tooo awesome I am totally gonna get a 60hz freesync 4k lg 32 inch monitor for Christmas. For when I can get a 6800XT next Spring. Love this generation of gaming tech. :D
Is Preferable to run 3 times the benchmark than just once. Why? Because it will give us the whole image and what truly these GPU can do. Ok, you are indeed true with the 1 run time for the player but a lot can happen in 1 run/benchmark, a fail, something goes wrong and etc. So, multiple repeats of the benchmark help you to avoid these things while a single benchmark simulates better what the player might see.
Either way, keep the good job going. My opinion is to keep what you doing but still report if you see a fail or something like that in a benchmark but I think you are doing this already too.
My take? You are just getting more and more professional. Good stuff, keep it up and I like your mic, you sound good.
Hey mate,
Looking at the nvidia card in dirt there isnt shadows for someone of the things above (looks like ropes) almost looks like it's a different day mode?
maybe he just neet to turn of clouds in settings before benchmarking :)
do a video for 5900x+6800xt vs 10900k+rtx3080
I used to use DiRT 3 for benchmarking a few years back and found it quite inconsistent..... the AI just caused so many FPS variations. Like if a car blocked the main car in one place and then the next run it didn't and the score would be different by the end.... so I think it's the nature of a racing sim with AI for the other cars. Probably similar to the difference that people notice in online multiplayer vs bot matches. It's the AI aspect that causes the variations...... I wonder whether the AI is being powered by the GPU, thus causing the variations?
Anyways, I can really appreciate the amount of time and effort you put into videos like this and then when you get inconsistencies like that it's a real headache and time consuming to sort out because if you just post inconsistent results then people will just pick holes in it.
You did an excellent job on this video imo
The old HD 6000 and 7000 were really good cards and they did put a gorgeous fight against nvidia back in the day. It's really emotional to see AMD putting up a fight again after many years of struggle with this 6000 line up.
Great video! Informative and fair criticism mate, cheers. Liked ans subbed keep em coming
Could we please get some testing of the new smart access memory function AMD were talking about when combining the new 6000 series cards with the zen 3 processors? Keen to see if there are any productivity gains to be had from an all red system as opposed to zen3 + 3080?
It varies between games, at the moment it only aids a few titles. That makes sense as the benefits come from when the game knows it doesn't have to manage assets through a 256MB Aperture.
You can see the effect in other reviews like Gamers Nexus
I would like to see that too. Compared to dlss coming from nvidia I am hoping it competes
i'll just compare this to my 970 at same settings and sob like a kiddie who dropped their sweety in the dirt.
Always run an avg benchmark. One isn't enough. 2 is but 3 is precise
That Red Alert 2 track at the end was sweet 😍
Great video, thank you! Are you going to test, whether cyberpunk runs better on a 6800 or on a 3070? If so, you have my subscription!
Awesome vid... waiting for the next one.... luv the outro music! C&C forever!
Yay! Stuff to watch at work! E: I think that 3 run plan is better if you have the time for it. There are things that can magically go wrong for some reason so 3 runs is pretty good. There is absolutely something odd in the Dirt5 and I hope they havent done stupid tricks to make fps better and try look better. Good thing you dropped that!
There's a new encoder in the 6000 series and OBS hasn't added support for it yet. It is still using the old AMD encoder. Read the OBS release notes. It's not an AMD driver update, it's an OBS update that is required.
The 6000 series contains a completely different encoder adding support for new formats.
I found this video very interesting and informative! great job!
did you upgrade your camera?? video looks a lot clearer.
Which microphone are you using?, it sounds good.
I saw 6800 xt having a lot more stuttering than 3080.
He explains that the stuttering is not due to the gpu but actually the cpu throttling. Neither one of these gpu's stutter at all 4k or 1080p
The 6800 XT is noticeably stuttering a lot more then the 3080. Will a 6800 XT actually stutter like this? or is it a mistake in the recording?
He explains that this is due to cpu throttling and not the card. These cards are 100% capable of running 60fps 4k.
Great work 👍
Something different. Great video. I was wondering why people only compare FPS when there are other things to consider.
Love the sound track in the end ... red alert was awesome
You dont happen to have power draw comparisons too do you? I see a lot of video showing the 3080s taking 50-60 more watts compared to the 6800xt. Just wondering if so and if it was a factor.
Hey man. Just wanted you to know your channel specifically is what got me into PCs in the first place. The whole price to performance bit. Yeah...I've been drinking a little bourbon while watching, so I feel pretty appreciative, but I wanted to show you some love. lol
Man you're truly working from the heart love your work.
Hi mate would like to see number crunching stats for good causes on Boinc !
Under rated program
Im so glad u made this. I have been telling people about the video you did showing dirt
@@WoutieX its still the right move to exclude it, if the whole image is covered in shadow on one side and not the other its fair to assume the rendering load is different on each, its easy to dismiss it as just darker on the 3080 side but if all that darkness is shadow that takes extra gpu time to render its outright not an equal test. benchmarks need to be for the most part exactly the same if he is to compare hardware, its one of the reasons reviewers allow for 1 or 2 fps each way and call it negligible, it likely comes down to extra npc's or something in the test from run to run so if the cards come within a couple fps of each other its as good as a draw unless one card is consistently in the lead run after run then its likely just a touch faster in that game.
In Dirt 5 it Looks like the inconsistant dark colours are due to the could shadows which occur in nature. I saw no other difference !!!!
For me there was one other difference. Its the dynamic range. The rx 6800 xt has a little less dynamic range in instances shown. Weather conditions can account for color but there is no excuse for poor dynamic range.
I can see a difference around 5:00 and before. The fidelity of objects in the distance on the 3080 is noticeably worse. You can see it again at 7:48-7:49, the definition in the stone work on the RTX card looks blurred as you get closer. Neither of those examples are due to the lighting.
@@chrisallen6010 There does appear to be more shadows with the stone part. But I notice the amount of stuttering with some of the footage. This was never fixed in the 5700xt. But I believe that the guy utilized nvenc encoder so maybe that caused it since its an amd card
It looked to me, more was done on the AMD card in Dirt5 not less, which would be consistent with it being a DX12 Ultimate launch title.
kjeep the great work man, great content
Interesting colors are all the same across the board, at least with older AMD/ATI cards vs Nvidia you could always see that one of the cards are a bit more vibrant with color? Or is it due to the recording that it gets filtered out if that is the correct term? Or is it another factor that does this?
It's hard for me to say if one run vs. 3 run average is best without knowing how each run performs. Is the first run usually slower than the other runs? If so, does the performance drop change between cards? Showing results for a few games would help think about it more.
Thanks for the video. The Dirt 5 case looks strange and it would be nice to see if the devs have anything to say about it.
At 0:16 I can read the Ford emblem on the right of the trunk lid and the word Cosworth on the left. With RUclips's compression I am not certain if that makes a difference. [ I stopped the video at that point to type a comment, not knowing it would be pointed out at 0:29 ] Speaking of the latter time frame I can easily read the smaller print on the license plate generated by the RX 5700 XT: codemasters At 3:50 shadows are darker on the RTX 3080, not certain if anyone would pay attention running a racing game. At 3:57 the sun showing through the windshield with the RX 6800 XT is cleaner, crisper than on the RTX 3080. At 3:55 there are ripples in the water with the RX 6800 and none with the RTX 3080, a glaring omission. At 4:08 there is a great deal of water splash with the RX 6800 XT and none when the Ford on the right hits the water RTX 3080. I stopped commenting at that point. Didn't want to make this comment a short book.
[EDIT: I watched it on my computer monitor at 4K, a 55" Samsung Q80T two feet from my face.] And, I did watch your clip from Horizon Zero Dawn in HDR 10bit. It suffered from RUclips compression, it still looked good though. I think doing benchmarks 3x yields more accurate comparisons, rebooting the computer for each run.
Thank you for this review.
Looking forward to that content about benchmark variance. We are seeing a lot of it now with all the new cards out.
Is 4K 60fps good for open world RPG games?
I wanted a rx6800xt but 3080 was available amd only limited stock so I picked up 3080. Both cards are great theres not much difference
I noticed the clouds are different in dirt 5. maybe the engine have a random volumetric cloud generator that made it impossible to get a an exact same cloud coverage condition?
7:34 is it just me or is the distant pillar lighting different?
Also a difference at 7:46
Also at 7:57
Good video.
On the benchmarking thing. When running a benchmark test on a game where there is some random factors to exactly what takes place, then doing multiple runs absolutely makes sense sine there is no way of knowing if just one run represents a best case, a worse case or somewhere between those two outer points. So please multiple runs and also letting it be know, how many runs was done and the highest and lowest scores.
How about another comparison with the e5-1680 v2 vs current cpu's and the new gpu offerings? Very interesting to see the differences.
0:28 why not a static scene?
I think you should do multiple runs because it would help now how much faster the cards are in comparison
Yeah benchmarks like Dirt are great for checking your settings to ensure you are going to get a good gaming experience, particularly given gaming is somewhat random.
No good for comparing GPUs/Hardware though. Which is kind of daft because you can achieve both through a consistant Bench.
So is the Dirt 5 benchmark just inconsistent, or always showing the same shadows when comparing AMD to AMD or Nvidia to Nvidia.
dynamic clouds system that they somehow didn't sync to be the same in the benchmark. How much actual difference that makes only the developers may know since that's heavily dependant on how the cloud shadows are displayed (culling, batching, etc.).
For your benchmarking question: Do both and label as such
7:15 how about run benchmark then restart pc, then run benchmark again instead
This might be the better middle ground of consistency and first run benchmark
Turn off fast boot to flush the memory too
according to Linus (LLT) doing that don't matter. There's not difference, just waist of time
The only thing I can think of is, most games that are tested will look very close because of GPU drivers.
The cards are basically drawing the same image, and we are testing the speeds and visual acuity of the games.
Call of Duty Black Ops cold war, shadow of War, Ashes of the singularity... theres tons of games that would be really interesting to see these cards run.
ALSO! Can you test multi-monitor performance of graphics cards? Get a counter on say, City Skylines, Jurrasic World, and Sims 4, and sims city to see how these cards handle multiple work flows?
Its because, I've noticed that switching from Pascal (gtx 1080) to Turing (rtx 2060 super) that the Turing card sucks at playing multiple games at once.
I liked having simulations running in the background on different screens and I think this is something people should see maybe.
Has the color range been changed from 16-255 to 0-255 on the nvidia card drivers ? Saw that in old video about image quality that you need to do this so we have apples to apples comparison !
yes nvidia full colour depth was enabled!
PC gamers aren’t real gamers. They are tech enthusiasts. Your current benchmarking method looks good.
Good point man I'd have to confess I do like all things technology including video games however after RUclips video games are what I spend most of my free time doing.
"PC gamers aren’t real gamers"... Speak for yourself, mate. Every bone in my body is a "real gamer". In fact, while my left hand fingers are typing this very comment, my right hand fingers are playing PC games!
Yes, I am also, a "tech enthusiast".
wait... So I am not a real gamer?
A 'real' gamer with money buys the best hardware they can to run games on.
True gamers play consoles they don't care about high fidelity graphics as long as it runs ok on ps4 and xbox one we play singleplayer games more we don't play trash fps games like call of duty fortnite pubg and trash doom eternal mad respect for you
Didn't the same thing happened in your stream with it looking darker on the 6800XT?
Lovin the new Tech YES Benchmarks! ❤
Personally can't wait for AIB partner cards
I'm getting the Nitro+ SE
Good luck y'all I wouldn't bank on getting one unless you've got a bot
I have a question tomyou or the commenters here. Do you get better shadows with less flickering on one or another card? AMD vs NVIDIA? My xfx 6700xt has a bad cooler and some coil whine and was thinking to get it refunded and get an 3070 Gigabyte gaming oc. When the card runs it is a fast one for my 1440p screen. I unfortunetly seen some stuff in Uncharted that are not that nice. Thinking on an driver issue. Please be kind in response, would love some help and guidence.
Hi bro . Which gpu you're using now?
Shadow of the tomb raider looks way diff to me. The detail on the textures and font and everything is better on thr 3080
Get eyes checked
Nope it's not processing a lot more shadows, just cloud shadows moving through the circuit. My guess is that cloud shadows don't really obey a common random seed when generated between benchmarks, so they show up in different positions.
Great job. Thx
Thanks for this video, I'll pick up a 6800 xt when they're restocked here in AUS. Probably getting an FE model unless the AIB prices are competitive.
Is there a way to just steam a benchmarks scene data to the GPU, in theory removing the CPU from the test?
For a comparison, the users’ first runs could be less FPS than your first run. The best is to run the benchmark three times and to include the standard deviation of the three runs on your charts. Excel has a feature for adding STD bars on bar charts.
Bloatware May decrease performance, but it will, hopefully, be systemic, affecting all tests the same.
You need multiple runs to detect outliers, where for example some OS task degrades benchmark processing. Those outliers ought be discarded not averaged.
A first run could be slower due to loading or faster due to better thermals.
In general GPU benchmarks are expected to show gfx performance not a general game experience including loads. A good in game benchmark probably pre-loads before it starts recording results.
A standard deviation would require a much larger number of runs to determine confidence properly. Just reporting your consistent results on a correctly functioning system suffices, IMO
@@RobBCactive
That would be ideal, yes. But, he only gave two options. Benchmarking takes a long time. 3 runs is a lot, let along 5 or more runs.
Why scrap a benchmark? Run it lots of times until you get a right aproximation
Will you make a video about ryzen apu-s in 2020 I'm much more interested in overclocked versions something like a ryzen 5 2400g overclocked in 2020 1080p low gaming ... Something like that i would like to see, tho I still enjoy the content that you to offer here on the chanel 👍🏻
About the "one run" or "three run" benchmarking thing, think about it that way:
if you play for example an open world rpg, there will certainly be places that you visit more often than once, even more often than three times (witcher, escape from tarkov, wow, ...)
even in other titles if you think about it (shooters, racing games, ...)
that's why i think that you should take the avg. of three runs, as there is (imo) not really any game where you are going to visit one place only once in your whole playtime
Is there much point in getting a 144hz 1440p monitor with the 6800xt?
Yes it would be a great pairing
you should get the very best 1440p gaming at ultra at high frames - although the 6800 will do it too
Adaptive sync will adjust the refresh rate to the FPS. Aka for freesync yes, it bottoms out at 35hz/fps usually for freesync panels. Since the 6800xt slaps at 1440p, thats a hell yes.
Another point is it will be less cpu intensive then 1080 240-360 hz/fps. The more fps your gpu renders, the harder the cpu has to work in order to keep up with them.so ipc would be key for that.
Lastly some games have caps that are lower then 240 fps anyway, doom 2016 tops out at 200 fps.
I can't find the video where you went to a youtube content creators house and optimized his pc pls help
So amd and nvidia encoders work differently and require different bitrate settings while file size remains the same...
is it just with capturing the gameplay or it hapoen when you are not capturing too?
So just to be clear, the differences in colour and details are just when being recorded or when playing the game live?
Live
@@choatus damn lmao
I liked this video, It may be a driver issue with Dirt5. The bottom line is people have VERY HIGH expectations of a $650 - $700 video card. Yeah, if these cards continue to stay unavailable. Many of us customers will just buy an an LG OLED TV (they are in stock). They cost the same as the high end video cards. I think they will make our current gen cards look even better. Maybe you can do another side by side with current gen gpus on OLED but w/o fps meters at first so we can see with just our eyes, then overlay the fps counter. That way it is the gameplay experience vs. just numbers on a chart. Thanks for all your hard work.
I had felt that there was such a difference before, but recently I have been able to feel that difference personally and notice it more clearly. I had an RX580, I used it for a while, I liked it, but when I went back to my GTX 1060 that I had left idle, I felt that visually it looks more immersive and beautiful. AMD have a more vibrant coloring, but Nvidia has a more immersive shading, the lights and shadows on Nvidia give a more immersive and good look, AMD has this vibrant coloring but it declines in this feature that I mentioned Nvidia has, ending up with AMD giving kinda of an artificial feeling to the visuals even though it has good colors.
Good video. I like that you used a extern way to record both.
To me, i noted a couple of areas in the last bench mark with the nvdia showing light or shadow in a couple of different areas.
Although for the most part the radeon images looked sharper over all.
Nvdia had a history of blurring elements in frames, so i heard, years ago. Reason being for higher fps.
If thats is still relevant today every so often do a space bar pause a comparison sided by side screen, see it your way.
Tell me please...NVidia have smooter image than AMD, right? Becouse all are talking about smoothness in games...
no AMD has better frame pacing resulting in a slightly smoother image. though its barely noticeable if at all
@@SirZanZa AMD always was some very little shuttering, always... Nvidia havent these problems. Mby 5000 and 6000 series AMD havent that, dont know, but people are talking about that!
In the dirt benchmark, the dark areas would appear to be from a cloud blocking the sun which if you look at the sky both runs have different cloud formations. The question is will that have a significant effect on the benchmark, especially if you do a three run average.
I did see the Radeon card get darker than the Nvidia card at different times, so this effect would have a similar affect either way. One issue with static benchmarks is that those are more prone to manipulation that benchmarks with randomised effects.
Rather than throw out the benchmark, it's a good thing to have variables that will average out like what this seems to be.
yeah this is questionable, so many shills these days you can't trust anybody...go look at jayztwocents that clown has his whole top shelf filled with nvidia cards while doing an amd review, what a clown. steve from gamers nexus already said nvidia wants their cards seen in the background when their sponsored shills do amd reviews...here's another youtuber I won't watch...
I would think it's good to know how wide the cards vary in fps. If one run you average 140 fps then the next time it's 115 fps that's a big difference. If it's only 5 fps different no big deal. What card to you think will be a better card in future games?
Probobly rx 6800 xt with its 16gb of vram and plus amd cards do get better and faster over time
LOVE the Red Alert music outro!!!!
@Tech YES City bro I have recently build a Desktop , But there was some shortage of 3600 So I had build Using 4650g & rtx 2060 , so is it a bad combo plz reply, by the way big fan love from India 🇮🇳 🧡💚🧡
I don't hink the criticism of dirt 5 is warranted, the brightness of the light shouldn't affect the number of shadows and the computational load, it should only change the variables in the compute. A good benchmark should have a low run to run variance in results, it doesn't need to present the exact same scene, especially if you do multiple runs.
Why not do a three run average and show the first run numbers too?
I have been interested in the 6800XT recently and I will admit I clicked on this to see that they're identical, and gloat about the parity AMD has struck with Nvidia, but sadly during the SotTR benchmark I can clearly see a slight stuttering as the background pans and the 6800XT when compared to a relatively butter smooth movement of the 3080. I probably wouldn't have noticed if they weren't side by side. Contrast looks better on the 3080. Granted we're talking a 5% difference in total fidelity. But when they're priced similarly why not get the one that looks better. The comparison helps.
great video !
You didn't mention anything about ray tracing benefits or non benefits ?