+8% performance uplift over 7800X3D is actually in line with AMD's own 15 game benchmark results from their marketing material. So AMD at least is being honest this time
If anything, as Steve says, 8% still seems conservative. Last-gen console games, for starters, should be omitted as they clearly aren't stressing CPUs.
"Next up I'm going to run intel CPUs through the same tests"... Listen, Steve, it's a really bad idea to announce you're going to commit a murder on camera.
What you are saying around 11:00 is really important thing I have noticed when switching from 5600 to 5700x3d. I was in the middle of very cpu heavy game/location, and I have found out over 40% gain, when in the tests it's not a thing by average. But when you encounter so specifically cpu demanding scenario, then faster cpu shines greatly.
@@GewelReal switched from a tuned 5800x to a 5800X3D at the time. Definitely very noticeable. Especially in Hunt: Showdown and Escape from tarkov. Depends on the game but in games where you notice it, you really notice it.
Yes, you're...heading in the right direction. I did get very used to your benchmarks showing extracts from sections of the game, but I can appreciate that this is a more efficient and at the same time more informative way to show data from many games. Despite my tongue in cheek comments, I appreciate your work.
@@NickChamberlain980 you should be fine to zen 6, I have 5700x3d and i will wait for zen6, this performance is great but I don't feel like mine is very bad
Do you think that’s because they don’t need to (half) lie to make the product look good this time, or because they learned their lesson this time? I’d like to think AMD learned from their mistakes, even though I highly doubt it at this rate. 😂
@@elibolton6235yea same, although almost every company lies about their product. Side effect of no consequence for them. The degree of lying though, is up to them lol.
The CPU install process itself, at this point, is identical to Intel (AM5 moved to LGA pin layouts, just like intel, with gold plates on the CPU and the pins inside the motherboard) Cooler might need a little bit more work to get it mounted but still overall pretty easy.
just get a good 6000 mhz rated ram kit (2 sticks not 4 sticks) and you should be gucci. AMD are very depended on their RAM cuz of memory training. Don't freakout when it takes awhile to train. It's supposed to. It'll get faster overtime with bios updates.
@@Sleepless4Life G.SKILL Trident Z5 Neo RGB Series (AMD Expo) DDR5 RAM 64GB (2x32GB) 6000MT/s. It arrived yesterday from Amazon 😁 Will try not to freak out. That is definitely different from Intel.
TLDR: Please, include Factorio in your benchmarks again. Factorio is one of the most played games on Steam, specially now that it has released it's biggest update yet. Hitting 118k avg players in the last 30 days (SteamDB). I'm probably the minority here, but I was so happy to see that you were including Factorio in your benchmarks in your previous X3D reviews, since it makes such a huge difference for it vs a non-x3d CPU and almost none/none other channel are doing, specially in a reliable way like you guys do. I was hoping that you were going to include in some video of the 9800x3D, specially now that Factorio has released its biggest update yet, and now there's never been so many people playing it (hitting 118k in October avg from 18k in September, easily keeping itself in the top 30 most played on Steam).
Factorio, Cities Skylines (1 or 2), and other CPU demanding games more than welcome. also 1080x1920 tests are pointless - nobody buys 500$ CPU and plays 1080 low...
Exactly, factorio should be a standard for cpu intensive tests because I just upgraded my 5930k to 7800x3d for this game. If its better I will replace with 9800x3d :)
I think problem with factorio and similar type games is how to test them. State of the save file matters extremely in those games and only 'endgame' state matter yet everyone's 'endgame' is different. There was a case in factorio that to some state X3D chips demolish everything but if you play past that state and build even more you can actually get to the limit of 96mb L3 cache as well which results similar performance with everything else. Those type of games are also usually heavily modified (including cpu optimizations) by the hardcore fan base so vanilla save files are even more useless. I love the genre as well but I, myself really don't find the benchmarking of them useful for me in my own scenarios. It just has scientific value.
Forget it. Look at the game selection in this video and previous videos. HUB game selection is so one-sided, it's probably worst in the whole tech review scene. And it's been this way for years even after numerous arguments.
GPU tests incoming Q1/25?? Laptop pilots' have a video explaining how to use CPU tests to predict fps ceiling so you can plan an upgrade that won't be overly CPU bottlenecked, which you cannot do anyway due to soldering. Jarrod next you'll be pointing out he's using too much thermal grease!! 😮😮😮
@@phatminhphan4121 Yes, for games like Cities Skylines 2 & World of Warcraft. Gained over 120% FPS in WoW going from 13600k to 9800X3D in 4k Resolution. From 50-60fps in the middle of the main city at primetime to 180~ if i unlock(Actually have to cap it now lol).
10:50 OMG finally someone that states this. I've always felt like CPU benchmark should be focused on futureproofing a potential future GPU bottleneck. i see (almost) no point in testing games that are currently GPU limited. it should always be about how fast the cpu can be, because it WILL show those performance gains more and more in the future.
@@EVR1AL Dumb... yet there is a market for e-sports games and more fps = reduced latency. There is a market for it other wise there wouldn't be 240 & 480Hz monitors. Sure there is diminished returns it's a niche market like gaming was back in the 90's. Just because you are old with a reaction time of snail, like me now, time comes for us all.
An update from a 7800X3D to a 9800X3D is basically meaningless (except if your really play on 1080p with a 4090 but in this case should see a doctor for help ^^).
@@jamesleicher Temps are very good! I'm using all stock settings and an NH-D15 cooler without an offset mount in a Fractal Design Meshify 2 case. 45-50c idle, 75c max while gaming, but mostly sits in high 60s and low 70s. Very brief spikes to 85c when I launch tasks and games, but that's expected from a Ryzen chip.
For those looking to compare the 5800X3D, here's the math. Steve's benchmarks for the 7800X3D averaged 23.5% higher FPS than the 5800X3D. So 110% of 123.5% is 1.10 * 1.235 = 1.3585 for a 36% (rounded) uplift for the 9800X3D. You're welcome.
I suppose, you then roughly divide that number in half considering roughly half the overaall PC strength is to the CPU n the other to the GPU. Good stuff! Thanks!
nitpicking here.... he said this is 11% faster on average for CPU-bound games. So 1.11 * 1.235 = 1.37 or 37% faster on average. It's sort of meaningless, though, because you would have to do all of the benchmarks on the same exact version of Windows for consistency. If the Zen 4 and Zen 5 benefit more from the Windows improvements than Zen 3 did, then this calculation is way off.
Honestly, who cares ? I mean most gamers are not using 4090 and most of those with 4070 and up don't play at 1080p low to medium settings. So in real world, the chances to run into CPU bottleneck with any decent CPU are next to none. I really wanted (not really needed) to upgrade my 5600X this year (preferably to intel), but after seeing all the launches (before X3D) I just ordered 5700X3D for 151$ delivered from ali.
Unless playing shooter, compare is useless. Don't get the reason for 2k and 4k testing. Due GPU bottlenecks some game if not all games, may see zero to next to nothing if upgrade from 5800x3d to 9800x3d. They need to include a few data points in 2k and 4k to get really world worth upgrading CPU to 9800x3d. (right now, I play mostly 2k gaming and max setting, I don't think I get any improvement from upgrading from 7800x3d to 9800x3d, Pretty sure every game I play at 2k is bottleneck by GPU for the most part. Even when GPU bottleneck, small gain normally with CPU upgrade, not as much like see when GPU isn't bottleneck.
@@oldRighty1 Hoping this title will perform better on hardware than 2020. It should, according to developers saying it will have much better thread utilization.
I noticed that the first time he listed it the gain wasn't that big, but tested with DirectX11. The second time, with the best gained games in that diagram it was among the better, with close to 20% gain, but this time it noted DirectX12....
8% uplift at 1080p with a 4090. Now play with a more affordable gpu like a 4080 or 7900xtx at a relevant resolution like 1440p or more and now you find less than 1% uplift. As long as you have a 5700x3d you're good to go for the next 2 gpu generations. Dont fall for the inflated hype.
Thank you so much for such exhaustive testing. You know you're setting a new standard for benchmarking when it makes sense to group games into blocks of five for presentation of the data.
If some games were GPU bottlenecked at 1080p Ultra Settings, then I believe it is better to set the settings to the lowest. If CPU performance difference is what we are trying to see, then graphical settings should be set to the lowest in order to prevent the GPU from holding us back.
@@burger-enjoyer you don't want only real world scenarios, you also want *future proof* scenarios that means testing at settings that show what the CPU would be capable of, *if* you had a GPU that could push it as hard as it can go everyone has different systems, so at least showing the uplift compared to previous generations would be enough
@@gasracing5000 it's mentioned previously that testing at 720p is not where most people play at and the 4090 can see a regression in performance at 720p vs 1080p.
@erictayet The first point is a bit mute, as it supports the requests for 1440p reviews, as it is what many "serious" gamers use. (I know the steam survey says 1080p is prominent, but what resolution you game at?) The second point maybe the real reason. Problem now is my head is spining as to why regression at a lower resolution occurs. The joy of scientific analysis, for every question answered, two more arise. The quest for knowledge never ends.
@@gasracing5000 I personally don't think 720p is relevant anymore in 2020s. But I also don't think my playing resolution matter since this is a CPU review and the key point is fps target. I game at 4K HDR native resolution so CPU benchmarks are even less relevant to me for gaming and is the same for many players. Gamers should almost always prioritise GPU upgrade for 1440/2160p gaming. Only gamers who play eSports title and wants high fps that their monitor support should consider a CPU upgrade. I come to these reviews because I'm running a 5900X and I want a more powerful CPU for AI & compute workload while having good gaming performance. Take the Intel 285k for example. It's got awesome compute & AI performance, but its gaming performance is bad, so I'm probably not going back to Intel for this gen. Also, I know that DX12 titles can use up to 8 cores for max performance and my 5900X is only 12C/24T on 2 CCD, so I am at a 5-10% disadvantage compared to a 8C/16T CPU like 5800X3D at 4K. This means my next CPU needs 8 performance cores like Intel 285K or AMD 9800X3D. But since compute is my biggest workload, logically 9950X3D is now my target CPU.
@@ScruffyIMS Looking more closely into the reviews, especially now that there's some that show GPU utilization, we don't see the real performance of 9800x3d yet in these averages. In 100 % CPU bottlenecked situations it goes clearly above 8 or 11 %. The issue is that max graphical settings are so bad for performance that they'll cause bottlenecking even at 1080p.
@@gaav87 Wrong, I’m getting the 9950x3d, I ain’t building a new system just for 40% gains. The 9950x3d will have 2 V-cache dies, they are letting fools overpay for 8 cores before releasing the real top dog
@@gaav87 If you have a very good graphics card and a high quality monitor, it could make sense. I went to a 7800X3D from an 11 year old SandyBridge-E with DDR3 build, a year and a half ago, and with my mid-tier graphics card with a 3k resolution monitor, I saw less than 20% performance boost in some games.
Is there a chance you can benchmark World Of Warcraft? possibly stress test with other people around, like main cities/raid etc.. I normally drop to 50fps with +20 people around using 5800x / 3080ti and would love to know if i should upgrade to 9800x3D. Great vids mate
The trouble with these games is that if there is a benchmark mode they are either not representative for actual gameplay (Final Fantasy) or repeatable because there is no benchmark (WoW).
Sadly, WoW has a very old game engine and cant utilize all the cores properly. I run 7800x3d and 4070S. In populated cities and cpu intensive areas the CPU caps out at 20-30% usage and leaves the GPU running at around 60-75% usage at 1440p. fps runs around 110-140 fps depending on how busy it is. On non cpu intensive scenes however the fps skyrockets past my monitor's refresh rate which is 165Hz. Still tho, you should see a nice performance boost with upgrading to a 7800x3d or 9800x3d, that 3d v-cache is a game changer.
Just get, a 5800x3d it will be 30-40% faster with 3080ti you dont need 9800x3d its only for people with rtx4090 or unreleased rtx5080 and rtx5090. rtx4080 and rx7900xtx is fine on 5800x3d and 7800x3d
If you already have a 7800x3D already it really makes no sense to buy a 9800x3D. You might aswell skip a generation and wait for the Ryzen 10,000 Series.
Wow Thank you for the amount of work that went into this, its much appreciated. I was looking for a review of Cod6 and Fortnite and this delivered. Its nice to just know exactly what games would improve the most.
Insane amount of work and appreciated as always ❤ Would've like FFXIV benchmarks as well but other Steve has that covered. Will be looking to pick a 9800X3D up next year for a gaming only build
Amazing video Steve! What is your thought on testing 720p gaming as a way to extrapolate/estimate future performance, and eliminate any of those odd games that are still gpu bottlenecked?
Hello, just want to post to please consider doing World of Warcraft at CPU benchmarks. It is heavily CPU limited game, milions of people play it, and it's easy to test performance in repeatable way by just doing flight maps from one place to another (there are NPC which gives you automated travel (always predefined route) from place A to B). I'm sure WoW community would be hugely grateful, and it would be certainly more interesting benchmark than a lot of the games like Forza.
Can you add even one or two Path Tracing test on future CPU reviews? On all my CPU testing, the PT is by far the biggest hit on the CPU usage. It just melts the CPU performance on games with already high CPU load (Cyberpunk). You can even run these with low rendering resolution (high crowd density) to show these differences, but use the highest GPU settings + PT. In my tests with 5800x3D, the CPU couldn't handle even console level fps. I'm just afraid that this is going to be a bigger issue with RTX 50XX GPUs that have probably double the RT performance vs. 40xx cards. I can see plenty of scenarios where users are hard CPU bottlenecked with Path Tracing turned on. They might use 4k DLSS performance PT on with RTX 5090/5080 and use 5800x3d, 7800x3D, or 9800x3D. This was a bit funny when I saw my results on CPU test. I wasn't expecting this to be so massive issue.
The valuable lesson here is that if a company can under-promise and over-deliver, they'll receive positive press coverage. Which benefits them more in long term.
@Hardware Unboxed Thank you for breaking up the results into smaller groups makes it eaiser for people with vision issues to see the differences without having to pause every graph! Keep up the great work!
Wtf why do so many people in the comments ask for HIGHER resolutions IN A CPU TEST?!? Resolution is only relevant for GPU testing guys 🤦♂️if anything I want lower resolutions tested to see the actual fps the CPUs can achieve without any GPU bottleneck impacting the results
Plenty of genres not covered by HuB. Survival, city builders and management games. I've got no interest in any of the games HuB has ever used for testing.
@Gazer75 I think MMOs are a good choice because usually, the crowded areas with a lot of players can put a heavy CPU load on the system while the game pulls up the player data. Would be a mostly useless GPU test though, because many of them are ancient graphically. I actually wouldn't hate seeing more city builders and sims myself.
Thanks for doing the work. I think it would be useful to get a couple of games where you test them at higher scalings, it makes it easier to see if at 2k and 4k it really makes an upgrade difference, though I do understand why that isn't the testing range. Really useful to see the difference between the 2 though. Thanks for all the hard work
You need to test more simulation games like factorio, rimworld, cities skylines 2, planet coaster, and a hugely modded minecraft pack with distant horizons loading in the world around you! It's those games that really matter!
Thank you so much for testing War Thunder, I have more than 1000 hours played in that game and finally decided to upgrade my PC. Please 🙏 do not remove War Thunder from your benchmarks. Have a great weekend
@titusrichard4949 my pc is my profile picture. I went with my first all white high end build. Gigabyte aero oc 4090/ 7800x3d which will be given to my bro for his upgrade and I am waiting on my 9800x3d that will hopefully be delivered later this month/ 64 gigs ddr5 g skill cl30 latency 6,000mhz speed which will be back up to the 64 gigs ddr5 silver g skill royal cl28 6,000mhz which I will buy next week/ lian li o11 vision case/ I also ordered all of the lian li wireless strimmer cables with the wireless controller which I should also receive later this month and I will give my bro my current lian li v2 strimmers. Ssd's are the 4tb gen 5 crucial t705 and the gen 4 samsung 990 pro. I am looking forward to a white 5090 and I will most likely go with a zotac or gigabyte and maybe an asus 5090 depending on how much the asus tax will be lol. I will sell the 4090 to put towards the 5090.
Hey i had a maybe unrelated question but i wanted to ask it nontheless, See I have a 4070 paired with a 5600, would it be beneficial to upgrade to a 5700x3d ? I primarly play at 1440p with high settings on fortnite and singleplayer games
yes, if you can get one, get a 5800X3D (which is sadly unlikely) (maybe used? they can't be overclocked anyway). It won't be an insane jump, but there will be a difference. Maybe watch some old Gamers Nexus Benchmark videos so you can determine if you think it's worth the money for you.
at 1440p with a 4070 a 5600 will do fine for most games. a 5700X3D will give you a very small uplift in some games. Especially the 1% lows. However, as soon as you enable DLSS upscaling that 5700X3D will probably give you more frames in some games VS 5600. Also, 5700X3D will set you up for at least one more future GPU upgrade.
@@larsenmats yeah I guess the dlss argument is relevant cuz I always enable it,also I was planning to do a whole new build in late 2025/early 2026 with a 9800x3d and a 5070 ti/5070 super(the 18gb model)
At what time were you running the test? I don't game after midnight so these results may be unrealistic for my use case and I know there is a big time shift due to you being in Australia.
@@justmatt2655 Good call. Also having AIO cooled PC on top of a skyscraper may cause the liquid in AIO to boil faster due to change of pressure which may lead to worse CPU performance.
i doubt you will get much more fps in 4k maybe if you overclock the 9800x3d you might see afew more fps but that depends if its worth it to you for afew odd digits more
I like this new format. I only listen to your videos, and I watch them only when I need to see a specific graph for a game that I play. It is helpful for me that you named almost every game and describe the performance difference between these CPUs. Also, I have a 7800X 3-D, so it was particularly pertinent.
Will be interesting to revisit the games that did not see an uplift when the 5090 launches. Also simulation titles please, so many people play those games for years and years and years that they are severely under represented in review suites.
I would say testing cpus in 1080p is like testing an off road tire off road. Yes it's what it's meant for however 90% of the time people will still be using it on a normal road so you want to see how it performs there
Testing CPU in 4k is as clever as with a 4090. If you genuinely build a 9800X3D paired with a 4090 (or God forbid a 5090) only to play at 1080p there's something very wrong with you.
I would love it if you added an “esports” section to these reviews and quickly showed results of a few popular competitive multiplayer games using realistic competitive settings (e.g., performance mode for Fortnite). I know the market is small but I think there is an opportunity for you to become the definitive esports benchmarker for CPUs/GPUs, which should add to your viewership over the long term. You cover esports titles occasionally but I am talking about a consistent inclusion of these titles at lowest quality settings. Love the channel!
I wonder why so few people actually use CPU intensive games when benchmarking a CPU. Use Stellaris, use Factorio, use Rimworld to show the CPU performance. THOSE kinds of simulation games are the ones where the CPU matters most.
Can I request for 45 game benchmark using max settings at 1440 and 4k? I know this is not how we test CPUs, but I'd like to know if upgrading to 9800x3d will be worth it at the graphics settings I play in. You can even split it into 2 videos, 1440 and 4k for more views 😅
No one can tell you if it's worth it besides yourself. And only after you actually do it and can compare the experiences. I am playing at 4K with a 7900XTX and I am planning to upgrade from 7800X3D to 98000X3D. Why? Because it fits my use case. And because CPUs define so much more than just FPS numbers - how responsive the menus are, how fast the games load, how smooth the scrolling, how well the performance holds in more demanding areas and especially how well older games that are very heavily modded perform overall.
You suddenly realize an 8% uplift is huge when you're already pushing 240FPS. Combine this with the 5090 and 8K, 120hz/4K 240hz Gaming is now a reality. Can't wait to see the power the 9950X3D will bring, we will finally get the perfect CPU that can not only have great productivity OR gaming, we will get the perfect CPU that will have great productivity AND gaming
9950X3D will not be much better. As long as only one CCD have the extra cache you'll still be on an 8 core CPU while gaming. Maybe a 50-100MHz higher boost clock if you're lucky. Then there is the scheduler problem. It is not perfect and in some games you end up using the wrong CCD.
@@Gazer75 The rumours are that both CCDs get extra cache this time. GN's tour of AMD showed a dual 3D chip stress-testing in Task Manager. The engineer said it was far too expensive to be a consumer product at that time. The new layering could make it a reality, although still expensive.
You're CPU isn't going to help much with 4k and up gaming bud, also the 4090 barely does 4k120 why do you think the 5090 would somehow do 4k240, I have a 4090 with 7950x3d I get on average 80 -100 fps on most graphical games with everything maxed
I'd like to echo some other comments here of people wondering about 1440p and 4k benchmarks because those are better real world scenarios for people that are buying high end CPUs and not playing esports titles. Obviously 1080p is going to show the biggest contrast between CPUs specifically, but it would be nice to know how much the high end GPUs at the higher resolutions even really care about CPU performance.
If you want to know how a game will function at higher resolutions then you should watch GPU comparisons. For example, at 1440p or 2160p, both the R7-7800X3D and R7-9800X3D will have the same performance, held back by the GPU.
You won't get that from HUB. They are being paid by AMD to manipulate the test conditions so that this CPU 'looks' better than it is. They have ZERO interest in helping the real-world target audience of this CPU. Unfortunately.
Sure that's true for modern and last generation CPUs. But what if someone is running, say, a 3600x still. Would be useful to know at what point you're going to see gains from a CPU upgrade vs just dropping in a shiny new GPU.
Thank you guys for your hard work and the extensive benchmarks that you are doing.Whenever I want to know how a new release is performing the only place I look is here because I know I will get the best non-biased review on the internet. Keep up the good work and thank you again 🙇
Nearly everyone is glossing over the fact that the 9800x3d is an overclocking MONSTER with massive gains potential. Any Joe Schmoe with a basic AIO can overclock it to 5600mhz all-core, and for the tinkerers who are willing to delid it, we've already seen 5750 all-core clocks being achieved pretty easily on day 1 with just a basic OC. That is a 500mhz increase, so that 11% uplift in modern games could easily be closer to 16-18%, and even more extreme uplifts to the 1% lows.
@@L0Lyeahright He literally explained at the beginning that if you want to know why he does that and why basically any other reviewer who reviews a CPU uses 1080P you have to look in the video description. In the video description there are 3 videos that go more into depth as to why. In short, 1080P is currently the best possible way to show how much more performance a certain CPU has over another while minimising GPU usage during a particular benchmark. Again if you want a more detailed explanation, look in the video description of this video :)
@@emiel255 We know why 1080p is used. The problem is that ONLY 1080p is used anymore, when we used to get 1440p and 4k results as well. Also, 1080p is too high a resolution, based on their logic, and why is a graphics card even installed in the system to begin with for this testing?
@@L0Lyeahright Yes, that is true, but I think the reason they don’t do lower resolutions (like 240 or 720P) is because basically no one plays on 720P anymore but still plenty of people use 1080P. I think it’s just the minimum resolution that most people play now nowadays so that’s why it is better to use that as a CPU benchmark standard. Otherwise we could basically say let’s go down to 240P or something like you basically said. I’m not sure about this so if anyone knows better let me know And I understand that you would like to see 1440P and 4K but it it will show nothing of valuable data (HUB and other reviewers have explained and showed this numerous of times) which causes the 1440P and 4K data to be meaningless as it will say absolutely nothing about how much extra performance the CPU has over other models. You will see only a meaningful difference at 1080P. You should definitely check out the HUB video called ''Why Reviewers Benchmark CPUs @ 1080p: Misconceptions Explained''. It is a very good video
Unfortunately the 1080p rtx4090 test skew the results for everyone with a 4090 and plays at 4K For you to be able to quickly just do 2-3 games at 4K would add at least a reference point of “hey, if your gaming at 4K you will see a .5% increase in most games and there is no reason to upgrade your CPU” Because some people will see results like this and assume there is some kind if not similar performance to be had. Lots of people have money to blow but neither the understanding or time to actually know what effects the system as a whole.
Respect for seeing through HUB's bullcrap manipulation. As you realise, HUB already knows that 4K results show little/no improvement over the 7800X3D. Why did they not say this to the viewers? - simple, they were paid by AMD to test ONLY at 1080p because 4K results would be embarrassing for them. As you already understand, HUB's exclusive 1080p results here are from cloud Cuckoo land because NO ONE with a 7800X3D or 9800X3D plays at 1080p, obviously. In the end, this testing configuration serves to mislead the uninformed while pissing off every owner of these actual CPU's.
if they have money to blow and no time to educate themselves on basics of what they're blowing their moneys on, then it's not the reviewer problem. On top of that I'm sure there're plenty of people who will eventually do pointless 4k benches for CPUs anyway
What reviewing TechTubers should do is. Benchmark at 1080p Benchmark at 1440p Benchmark at 2160p And tell us what the percentage gains are for each resolution, this way a lot less desktop people will be mad at your one sided benchmarks. Why is Linus the only person showing quick 1440p and 4K resolution, can you TechTubers explain that to the public?
@@ScruffyIMS Still we don't have 1440p and 4K benchmarks and the percentage gains for those resolutions. 1080p is not the all and only CPU benchmark to show the percentage gain for the CPU. On that ground you could go 720p with all low settings and show the world 1000 FPS in games and +50% FPS gains. That will paint the CPU as +50% faster than the previous gen, which is an odd ball as only 1080p benchmarks.
@@N0N0111The FPS limit of a CPU is the same at 4k, there is literally nothing to test. 720p runs weird on the 4090 and some settings run on the CPU. 1080 ultra is the only way.
@N0N0111 yes, we have those numbers. You have to look at a Gpu test ;) Your Gpu only does 60 fps in 4k in your chosen game? All the Cpus achieving 90+ are fine. Your Cpu doesnt care about resolution, and you want to test the part you are testing right now at the limit. And think about Steve, he can't do all these tests again x10 for different Gpus. Not everyone has a 4090, I'd guess a smaller part of the audience. The tests would be meaningless for everyone without one. Why does this only (mostly) come up with Cpu testing? You don't see all the cries for 1080p testing with a Ryzen 3600 for the Gpu benchmarks. It's easy to find out if you are Gpu limited, only you know what you play with which settings. If you are, don't bother with the Cpu upgrade. But I want to know which Cpu is faster
I know this is a big ask, but there is still huge World of Warcraft playerbase but nobody really benchmarks the game to see how the new hardware handles it. Would you consider adding WoW in some capacity? ❤
@19:00 on their video titled: "AMD Ryzen 7 9800X3D Review, An Actually Good Product!" you can find the gaming power consumption, and @8:30 for synthetic benchmark
No point doing benchmarks at anything higher than 1080...both are GPU intensive. You always test CPUs at 1080p. At 1440 and 4k you just don't want you CPU to bottleneck the GPU.
@@5ean5ean22 “At 1440 and 4k you just don’t want your CPU to bottleneck the GPU”. That is the exact reason why we should test games at 1440p and 4k. Some CPUs can bottleneck your GPU. That is the reason we test at 1080p. 1080p is where the greatest bottleneck exists. If we test a new CPU at 1080p and see a 5% gain, we can conclude that the new CPU is less of a bottleneck compared to the other CPU. At 4k and 1440p there are plenty of games that provide a 5% difference. Of course there are some games that have no difference and are GPU limited. I don’t disagree that 1440p and 4k are more GPU intensive but there still are some significant gains to be made with some CPUs.
@@A_GWAKwork_OrangeI understand the whole concept of bottlenecking and how higher resolutions are more GPU limited. I don’t disagree that the gains are small. But single digit gains like 5% or 8% or sometimes even more exist at 4k and 1440p in many games like Borderlands 3, MSFS, and Stellaris. Of course some games will show absolutely no gain because they are GPU limited. My point is, if an 8% difference is significant at 1080p(ie. The advertised improvement from 7800x3d to 9800x3d and 1080p) then why is a single digit improvement at higher resolution not significant? Again, I understand plenty of games will show such small differences that it doesn’t matter, there are plenty that do.
Thanks for bringing back MSFS2020.... Hoping to see MSFS2024 next month in charts 😉 Also nice to see War Thinder again, will be interesting if that will show up in RT charts as they are now adding RT to the game
What I found amazing is the performance in Stellaris simulation time in Gamer Nexus video. For people that plays that game is a no brainer upgrade. Huge improvements that translates into wasting much less time waiting for the game to process things and actually playing.
Basically yeah. I am glad I didn't wait to build my new system (first in 8ish years). Since I did the math and in Aus, it's about 44-45% more expensive to get the 9800X3D Vs what I paid for the 7800X3D a few months back. Good timing I think 😆
Why all the test are in 1080p... seriously... People who spend this kind of money don't play in 1080p! we want to see 4k results. If you're going to play in 1080p, why even bother with this cpu and a 4090... fcs...
Thanks for all the tests at 1080p, I understand it more gpu limited at 1440p and 4k, and 1080p shows the differences better between the cpus, i still would like some 1440p testing as well
@@sabretoothauit’s ludicrous, I’d just look at cinebench and 3dmark if I wanted non-real world scenarios of how the chips perform. Buyers of this chip are not gaming at 1080p which makes any justification irrelevant
Can't wait to pair a 9800X3D with a 5090 Ti instead of buying a car
based. cars are trash and games are fun.
9800X3D and 5090 Ti. My dream combo.
If you are in that position, maybe you should work more instead of gaming...
@@TheRealEtaoinShrdlu I will get an RTX 5090 with a 9800X3D. The 5090 Ti seems unlikely to happen since there was no 4090 Ti.
I've personally stopped buying food to buy this next month. I should also go to the hospital and pay to get my crushed knee cap fixed, but F that ..💪🏿
+8% performance uplift over 7800X3D is actually in line with AMD's own 15 game benchmark results from their marketing material. So AMD at least is being honest this time
They also tested with a 4090 this time.
When the truth is enough to destroy your enemy, facts are the best propaganda.
@@daedalus6433 As they should
If anything, as Steve says, 8% still seems conservative. Last-gen console games, for starters, should be omitted as they clearly aren't stressing CPUs.
@@neoaliens8635 Yeah, it was very dishonest using a very bottlenecked GPU for the normal 9000 series
"Next up I'm going to run intel CPUs through the same tests"... Listen, Steve, it's a really bad idea to announce you're going to commit a murder on camera.
😂bro lmao
😂😂
He's gonna be standing again
@@soulsbourne Gonna be standing on stilts
Oh that’s cool the new 14900k beats last gen’s 285k
that was an insane amount of games to benchmark. appreciate the hard work
What you are saying around 11:00 is really important thing I have noticed when switching from 5600 to 5700x3d. I was in the middle of very cpu heavy game/location, and I have found out over 40% gain, when in the tests it's not a thing by average. But when you encounter so specifically cpu demanding scenario, then faster cpu shines greatly.
My problemis that my 5600 is running at 4.9GHz with heavily tuned B-die. Would going 5700X3D still be a good idea?
Yes@@GewelReal
@@GewelRealyes
@@GewelRealjust switch to Ryzen 5 7500f it will offer better performance ❤
@@GewelReal switched from a tuned 5800x to a 5800X3D at the time. Definitely very noticeable. Especially in Hunt: Showdown and Escape from tarkov. Depends on the game but in games where you notice it, you really notice it.
Great format for head to head comparisons, Steve, really good choice here I reckon
Yes, you're...heading in the right direction.
I did get very used to your benchmarks showing extracts from sections of the game, but I can appreciate that this is a more efficient and at the same time more informative way to show data from many games. Despite my tongue in cheek comments, I appreciate your work.
Day 1 of asking for a 5800x3d version of this benchmark
Add that 5700X3D in there too!
Indeed, wondering if its finally time to move on from AM4 but given I don't game at 1080p the uplift probably isn't worth it
@@NickChamberlain980 you should be fine to zen 6, I have 5700x3d and i will wait for zen6, this performance is great but I don't feel like mine is very bad
don't they have 5800x3d vs 7800x3d , just add the diference , don't be dumb .
As an 5800X3D owner I second this.
Wow, AMD was honest about the performance uplift. I'm surprised, but I guess they don't need to lie when the results are genuinely impressive.
Do you think that’s because they don’t need to (half) lie to make the product look good this time, or because they learned their lesson this time? I’d like to think AMD learned from their mistakes, even though I highly doubt it at this rate. 😂
@@elibolton6235yea same, although almost every company lies about their product. Side effect of no consequence for them. The degree of lying though, is up to them lol.
Is +8% really that impressive after nearly 2 years? I feel we've had so many bad CPU launches lately that we are now overhyping mediocrity.
@@JimmyChino55in the case of 9800X3D, yes it is. There are games that have 20% of uplift. Everything depends on what you are playing.
impessive at 1080p sure, at 1440p 4k, not so much.
Finally leaving Intel. Getting the 9800 on Nov 19. Watching videos how to install AMD CPU's. I feel like this is my first build again. So pumped!
The CPU install process itself, at this point, is identical to Intel (AM5 moved to LGA pin layouts, just like intel, with gold plates on the CPU and the pins inside the motherboard)
Cooler might need a little bit more work to get it mounted but still overall pretty easy.
@ yeah, memory seems to be a bit more finicky from what I have seen. Also, AMD machines can boot slower was reading how to improve that in the bios.
just get a good 6000 mhz rated ram kit (2 sticks not 4 sticks) and you should be gucci. AMD are very depended on their RAM cuz of memory training. Don't freakout when it takes awhile to train. It's supposed to. It'll get faster overtime with bios updates.
@@Sleepless4Life G.SKILL Trident Z5 Neo RGB Series (AMD Expo) DDR5 RAM 64GB (2x32GB) 6000MT/s. It arrived yesterday from Amazon 😁 Will try not to freak out. That is definitely different from Intel.
@@Cris1Mac I read, and found, that disabling the integrated GPU has a big impact on boot and program loading time.
TLDR: Please, include Factorio in your benchmarks again. Factorio is one of the most played games on Steam, specially now that it has released it's biggest update yet. Hitting 118k avg players in the last 30 days (SteamDB).
I'm probably the minority here, but I was so happy to see that you were including Factorio in your benchmarks in your previous X3D reviews, since it makes such a huge difference for it vs a non-x3d CPU and almost none/none other channel are doing, specially in a reliable way like you guys do. I was hoping that you were going to include in some video of the 9800x3D, specially now that Factorio has released its biggest update yet, and now there's never been so many people playing it (hitting 118k in October avg from 18k in September, easily keeping itself in the top 30 most played on Steam).
Factorio, Cities Skylines (1 or 2), and other CPU demanding games more than welcome.
also 1080x1920 tests are pointless - nobody buys 500$ CPU and plays 1080 low...
Exactly, factorio should be a standard for cpu intensive tests because I just upgraded my 5930k to 7800x3d for this game. If its better I will replace with 9800x3d :)
I think problem with factorio and similar type games is how to test them. State of the save file matters extremely in those games and only 'endgame' state matter yet everyone's 'endgame' is different. There was a case in factorio that to some state X3D chips demolish everything but if you play past that state and build even more you can actually get to the limit of 96mb L3 cache as well which results similar performance with everything else. Those type of games are also usually heavily modified (including cpu optimizations) by the hardcore fan base so vanilla save files are even more useless.
I love the genre as well but I, myself really don't find the benchmarking of them useful for me in my own scenarios. It just has scientific value.
I don't play Factorio or even like games like it and it was always one of the results I was most interested in, so +1 from me
Forget it. Look at the game selection in this video and previous videos. HUB game selection is so one-sided, it's probably worst in the whole tech review scene. And it's been this way for years even after numerous arguments.
2:08 After watching your recent 2 hour podcast and your recent tweets, I'm wondering where the 4K games are???
4k? Pfft. We want 8k benchmarks because reasons.
GPU tests incoming Q1/25?? Laptop pilots' have a video explaining how to use CPU tests to predict fps ceiling so you can plan an upgrade that won't be overly CPU bottlenecked, which you cannot do anyway due to soldering.
Jarrod next you'll be pointing out he's using too much thermal grease!! 😮😮😮
4K with a GT1030
Testing a cpu at 4K 😂 ?
@@phatminhphan4121 Yes, for games like Cities Skylines 2 & World of Warcraft. Gained over 120% FPS in WoW going from 13600k to 9800X3D in 4k Resolution. From 50-60fps in the middle of the main city at primetime to 180~ if i unlock(Actually have to cap it now lol).
10:50 OMG finally someone that states this. I've always felt like CPU benchmark should be focused on futureproofing a potential future GPU bottleneck. i see (almost) no point in testing games that are currently GPU limited. it should always be about how fast the cpu can be, because it WILL show those performance gains more and more in the future.
Games capable of 200 frames shouldn't be tested at all, that's so dumb, he spent time testing War Thunder, Rainbow six, CS2, instead of Witcher 3
@@EVR1ALIt's not relevant how many frames a game renders. It's all about a percentage difference between CPUs.
If you're on AM5 you don't need to future proof your CPU as upgrading your CPU in the future will not require a new motherboard or RAM
@@stark_2991 yea with such thinking can test half life 1 and CS 1.6 and quake 3 and team fortress 2 genius galaxy brain
@@EVR1AL Dumb... yet there is a market for e-sports games and more fps = reduced latency. There is a market for it other wise there wouldn't be 240 & 480Hz monitors. Sure there is diminished returns it's a niche market like gaming was back in the 90's.
Just because you are old with a reaction time of snail, like me now, time comes for us all.
Super happy with my 7800X3D.
Got one in the beginning of 2024, and I don't plan on swapping it for years to come. :)
hows the temps?
An update from a 7800X3D to a 9800X3D is basically meaningless (except if your really play on 1080p with a 4090 but in this case should see a doctor for help ^^).
@@christianschmidt6110exactly
7800X3D is still strong. Especially if you play at greater resolutions where the differences in fps are a lot smaller.
@@jamesleicher Temps are very good! I'm using all stock settings and an NH-D15 cooler without an offset mount in a Fractal Design Meshify 2 case. 45-50c idle, 75c max while gaming, but mostly sits in high 60s and low 70s. Very brief spikes to 85c when I launch tasks and games, but that's expected from a Ryzen chip.
For those looking to compare the 5800X3D, here's the math. Steve's benchmarks for the 7800X3D averaged 23.5% higher FPS than the 5800X3D. So 110% of 123.5% is 1.10 * 1.235 = 1.3585 for a 36% (rounded) uplift for the 9800X3D. You're welcome.
For double the cost?
I suppose, you then roughly divide that number in half considering roughly half the overaall PC strength is to the CPU n the other to the GPU.
Good stuff! Thanks!
nitpicking here.... he said this is 11% faster on average for CPU-bound games. So 1.11 * 1.235 = 1.37 or 37% faster on average. It's sort of meaningless, though, because you would have to do all of the benchmarks on the same exact version of Windows for consistency. If the Zen 4 and Zen 5 benefit more from the Windows improvements than Zen 3 did, then this calculation is way off.
Honestly, who cares ? I mean most gamers are not using 4090 and most of those with 4070 and up don't play at 1080p low to medium settings.
So in real world, the chances to run into CPU bottleneck with any decent CPU are next to none.
I really wanted (not really needed) to upgrade my 5600X this year (preferably to intel), but after seeing all the launches (before X3D) I just ordered 5700X3D for 151$ delivered from ali.
Unless playing shooter, compare is useless. Don't get the reason for 2k and 4k testing. Due GPU bottlenecks some game if not all games, may see zero to next to nothing if upgrade from 5800x3d to 9800x3d. They need to include a few data points in 2k and 4k to get really world worth upgrading CPU to 9800x3d. (right now, I play mostly 2k gaming and max setting, I don't think I get any improvement from upgrading from 7800x3d to 9800x3d, Pretty sure every game I play at 2k is bottleneck by GPU for the most part. Even when GPU bottleneck, small gain normally with CPU upgrade, not as much like see when GPU isn't bottleneck.
Appreciate the inclusion of MSFS 2020. Thanks for all the testing.
Will be interesting to see the 2024 benchmarks coming out very soon!
@@oldRighty1 Hoping this title will perform better on hardware than 2020. It should, according to developers saying it will have much better thread utilization.
yes a very intensive cpu limited game is perfect. wish more would test it
He really needed 64GB of RAM for that test.
I noticed that the first time he listed it the gain wasn't that big, but tested with DirectX11. The second time, with the best gained games in that diagram it was among the better, with close to 20% gain, but this time it noted DirectX12....
Terrific video, so much information packed in so little time. I bet you worked really hard getting all those numbers. Thank you my good man!!!
Nice to see AMD being honest with their data
they only do whats beneficial for them
Yep unlike the rdna3 bs data
Buying my 7800X3D for ~330€ before Zen5 NDA was lifted was a pretty good deal after all. But the 9800X3D is impressive.
8% uplift at 1080p with a 4090. Now play with a more affordable gpu like a 4080 or 7900xtx at a relevant resolution like 1440p or more and now you find less than 1% uplift.
As long as you have a 5700x3d you're good to go for the next 2 gpu generations. Dont fall for the inflated hype.
Thank you so much for such exhaustive testing. You know you're setting a new standard for benchmarking when it makes sense to group games into blocks of five for presentation of the data.
Thanks mate :)
If some games were GPU bottlenecked at 1080p Ultra Settings, then I believe it is better to set the settings to the lowest. If CPU performance difference is what we are trying to see, then graphical settings should be set to the lowest in order to prevent the GPU from holding us back.
But some settings actually use more cpu load. So some settings need to be on. Like gta v with its density settings
use 720p or upscaling in that case to make sure a cpu limit is reached
But what's the point? I want real world scenarios, not benchmark settings that with results as pointless as synthetic benchmarks scores.
@@burger-enjoyer you don't want only real world scenarios, you also want *future proof* scenarios
that means testing at settings that show what the CPU would be capable of, *if* you had a GPU that could push it as hard as it can go
everyone has different systems, so at least showing the uplift compared to previous generations would be enough
That's tricky because some quality settings do rely on the CPU. Also, it's nice to see a scenario that is at least somewhat realistic.
Also, this is out of the box cpu comparison, the fact that 9800x3d is overclockable boosts these differences even more.
Exactly, we need a Max OC Benchmark also
@@Cryptonomous So that on average 50% will be mad because their silicon isnt capable of the results?
checkout jaytwocents recent video. He becnhmarked it stock vs overclocked. The gains seems solid/interesting
Ooooooooh this is what I kept missing!
@@joaogabriels.f.5143He didn’t do any manual overclocking though. Roman (Der8auer) did some manual overclocking and delidded the 9800X3D.
The hero we don't deserve, but the hero we need.
We deserve it, long over due
@@mort996 No
The hero we can't afford.
@@garwynrosser8907 more like afford
Wow!!! 8%!!!!
2:15 shots fired i repeat shots fired!!
Shame it doesn't help people who write comments before watching or skipping everything except for comparisons or conclusions.
I think he is talking to me... lmfao.
So why not 720p rather than 1080p?
My simple brain says this would tax the cpu even more.
@@gasracing5000 it's mentioned previously that testing at 720p is not where most people play at and the 4090 can see a regression in performance at 720p vs 1080p.
@erictayet
The first point is a bit mute, as it supports the requests for 1440p reviews, as it is what many "serious" gamers use. (I know the steam survey says 1080p is prominent, but what resolution you game at?)
The second point maybe the real reason. Problem now is my head is spining as to why regression at a lower resolution occurs.
The joy of scientific analysis, for every question answered, two more arise. The quest for knowledge never ends.
@@gasracing5000 I personally don't think 720p is relevant anymore in 2020s. But I also don't think my playing resolution matter since this is a CPU review and the key point is fps target.
I game at 4K HDR native resolution so CPU benchmarks are even less relevant to me for gaming and is the same for many players.
Gamers should almost always prioritise GPU upgrade for 1440/2160p gaming.
Only gamers who play eSports title and wants high fps that their monitor support should consider a CPU upgrade.
I come to these reviews because I'm running a 5900X and I want a more powerful CPU for AI & compute workload while having good gaming performance.
Take the Intel 285k for example. It's got awesome compute & AI performance, but its gaming performance is bad, so I'm probably not going back to Intel for this gen.
Also, I know that DX12 titles can use up to 8 cores for max performance and my 5900X is only 12C/24T on 2 CCD, so I am at a 5-10% disadvantage compared to a 8C/16T CPU like 5800X3D at 4K.
This means my next CPU needs 8 performance cores like Intel 285K or AMD 9800X3D. But since compute is my biggest workload, logically 9950X3D is now my target CPU.
what is the avg 1% low uplift? that seemed to be a pretty strong point in some games for the 9800X3D
It means less stuttering where the same picture is displayed more than once. It is a strong part of the X3D cpu's.
It's also 8% for all games, and 11% for the reduced selection
@@ScruffyIMS Looking more closely into the reviews, especially now that there's some that show GPU utilization, we don't see the real performance of 9800x3d yet in these averages. In 100 % CPU bottlenecked situations it goes clearly above 8 or 11 %. The issue is that max graphical settings are so bad for performance that they'll cause bottlenecking even at 1080p.
Stellar work Steve - thanks for this content!
The 9800X3D is awesome, but will keep going with my 5800X3D for now until I can grab it at a good price point on a black Friday deal or something.
Obligatory "wheres 5800x3d" comment
Who cares all 5800x3d owners including me swaping to 9800x3d xD its 40-60% faster
At 4k@@gaav87
@@gaav87not at 4k… lol I’m on an 5700x3D still not much reason but definitely a way better chip
@@gaav87 Wrong, I’m getting the 9950x3d, I ain’t building a new system just for 40% gains. The 9950x3d will have 2 V-cache dies, they are letting fools overpay for 8 cores before releasing the real top dog
@@gaav87 If you have a very good graphics card and a high quality monitor, it could make sense. I went to a 7800X3D from an 11 year old SandyBridge-E with DDR3 build, a year and a half ago, and with my mid-tier graphics card with a 3k resolution monitor, I saw less than 20% performance boost in some games.
240 out of the 1837 post are asking for 1440p! And then there are the request for 4k!
You GOTTA do this whole thing again when 5090 comes out, 9800x3d seems gpu lited
and even more clowns will come out with rumble about testing at 1080 lol
@jermygod well they're clowns as you said, hardware canuks(idk the spelling) tested at 720 low as well and 9800x3d was more better there
After months of standing and a little bit of laying on the coach, Steve is now sitting again. Balance has been restored.
Is there a chance you can benchmark World Of Warcraft? possibly stress test with other people around, like main cities/raid etc.. I normally drop to 50fps with +20 people around using 5800x / 3080ti and would love to know if i should upgrade to 9800x3D. Great vids mate
Yes you should upgrade to the 7800x3d or the 9800x3d both will be huge jumps.
The trouble with these games is that if there is a benchmark mode they are either not representative for actual gameplay (Final Fantasy) or repeatable because there is no benchmark (WoW).
Sadly, WoW has a very old game engine and cant utilize all the cores properly. I run 7800x3d and 4070S. In populated cities and cpu intensive areas the CPU caps out at 20-30% usage and leaves the GPU running at around 60-75% usage at 1440p. fps runs around 110-140 fps depending on how busy it is. On non cpu intensive scenes however the fps skyrockets past my monitor's refresh rate which is 165Hz.
Still tho, you should see a nice performance boost with upgrading to a 7800x3d or 9800x3d, that 3d v-cache is a game changer.
Just get, a 5800x3d it will be 30-40% faster with 3080ti you dont need 9800x3d its only for people with rtx4090 or unreleased rtx5080 and rtx5090. rtx4080 and rx7900xtx is fine on 5800x3d and 7800x3d
So if your not planning on getting rtx4090 or 5080 (in future) just get 5800x3d
If you already have a 7800x3D already it really makes no sense to buy a 9800x3D. You might aswell skip a generation and wait for the Ryzen 10,000 Series.
so glad I went for am5, not with intel 13/14th gen and the new one
Wow Thank you for the amount of work that went into this, its much appreciated. I was looking for a review of Cod6 and Fortnite and this delivered. Its nice to just know exactly what games would improve the most.
This guy is amazing! Thank you for reviewing everything in so much detail!
Going to wait for 10800X3D with my 7800X3D 🙂
New memory controller + all improvement from 9800X3D and better Zen6 IPC = best AM5 cpu for games and bright socket finish.
maybe in 8 years i will replace my 7800x3d bought on may
*11800X3D
AMD always make nonsense CPU with even number.
Don't think there will be a 10800X3D. If AMD keep the desktop naming the 10000 series will be the Zen 5 APUs and the 11000 series will be Zen 6.
@@Hanginthere4100Or maybe a new naming as Intel did? 😅
The Grey Knight holding the 9800X3D is the best thing ive seen all week.
Insane amount of work and appreciated as always ❤ Would've like FFXIV benchmarks as well but other Steve has that covered. Will be looking to pick a 9800X3D up next year for a gaming only build
Were you looking for FFXIV or FFXVI?
@@cheesyvoid3143 FFXIV, FF14. GamersNexus has it on their roster of gaming benchmarks
Just got a 9800X3D from Newegg. I’m so hyped. 😂
What are your specs?
😂😂😂
Same they said Monday for delivery but it comes today!
I bought one as soon as it became available at MSRP. Two days later, it's 100eu more expensive yikes
@remsterx i bought my 9800x3d from amd on Amazon and as of now I don't get mine until nov 21. 😭
Amazing video Steve! What is your thought on testing 720p gaming as a way to extrapolate/estimate future performance, and eliminate any of those odd games that are still gpu bottlenecked?
Hello, just want to post to please consider doing World of Warcraft at CPU benchmarks. It is heavily CPU limited game, milions of people play it, and it's easy to test performance in repeatable way by just doing flight maps from one place to another (there are NPC which gives you automated travel (always predefined route) from place A to B). I'm sure WoW community would be hugely grateful, and it would be certainly more interesting benchmark than a lot of the games like Forza.
Can you add even one or two Path Tracing test on future CPU reviews? On all my CPU testing, the PT is by far the biggest hit on the CPU usage. It just melts the CPU performance on games with already high CPU load (Cyberpunk). You can even run these with low rendering resolution (high crowd density) to show these differences, but use the highest GPU settings + PT. In my tests with 5800x3D, the CPU couldn't handle even console level fps.
I'm just afraid that this is going to be a bigger issue with RTX 50XX GPUs that have probably double the RT performance vs. 40xx cards. I can see plenty of scenarios where users are hard CPU bottlenecked with Path Tracing turned on. They might use 4k DLSS performance PT on with RTX 5090/5080 and use 5800x3d, 7800x3D, or 9800x3D. This was a bit funny when I saw my results on CPU test. I wasn't expecting this to be so massive issue.
The valuable lesson here is that if a company can under-promise and over-deliver, they'll receive positive press coverage. Which benefits them more in long term.
Impressive deduction skills.
I think you should take a more parental approach to explaining why you test at 1080p.
“Because I said so.”
HUB have dived into the gutter these days. All they seem to do is shill for money.
Did you consider grouping the results by Game engine type?
@Hardware Unboxed Thank you for breaking up the results into smaller groups makes it eaiser for people with vision issues to see the differences without having to pause every graph! Keep up the great work!
That was a lot! Thanks a lot for doing it, though I'm sure you must have had a blast while finally being excited about something after so long.
9800XD. Because its funny how much better than intel it currently is.
Wtf why do so many people in the comments ask for HIGHER resolutions IN A CPU TEST?!?
Resolution is only relevant for GPU testing guys 🤦♂️if anything I want lower resolutions tested to see the actual fps the CPUs can achieve without any GPU bottleneck impacting the results
720p has some complicated problems apparently. You will have to wait for the 5090 1080p ultra tests.
Had to check the comments, and you are right, and many are even calling themselves "enthusiasts" - enthusiasts of RGB leds maybe lmao
Would like to see popular MMOs like World of Warcraft and FF14 added to the list for testing; they can be quite CPU demanding at highest settings.
Gamer Nexus covers FF14. But I agree on more MMO coverage for these benchmarks. Although I guess they technically did include Throne and Liberty.
Plenty of genres not covered by HuB. Survival, city builders and management games.
I've got no interest in any of the games HuB has ever used for testing.
@Gazer75 I think MMOs are a good choice because usually, the crowded areas with a lot of players can put a heavy CPU load on the system while the game pulls up the player data. Would be a mostly useless GPU test though, because many of them are ancient graphically.
I actually wouldn't hate seeing more city builders and sims myself.
wow lol. that shit is 30 years old. its not cpu demanding its just shit old tech
>20% faster than Intel.
Thanks for doing the work. I think it would be useful to get a couple of games where you test them at higher scalings, it makes it easier to see if at 2k and 4k it really makes an upgrade difference, though I do understand why that isn't the testing range.
Really useful to see the difference between the 2 though. Thanks for all the hard work
You need to test more simulation games like factorio, rimworld, cities skylines 2, planet coaster, and a hugely modded minecraft pack with distant horizons loading in the world around you! It's those games that really matter!
Thank you so much for testing War Thunder, I have more than 1000 hours played in that game and finally decided to upgrade my PC. Please 🙏 do not remove War Thunder from your benchmarks. Have a great weekend
What are your pc specs?
7900XTX/9800x3D, ordered, waiting for it, 64GB 6000Mhz, CL30/ 4TB 990 PRO
@titusrichard4949 nice high end build.
@MrAnimescrazy, thank you, what are your specs?
@titusrichard4949 my pc is my profile picture. I went with my first all white high end build. Gigabyte aero oc 4090/ 7800x3d which will be given to my bro for his upgrade and I am waiting on my 9800x3d that will hopefully be delivered later this month/ 64 gigs ddr5 g skill cl30 latency 6,000mhz speed which will be back up to the 64 gigs ddr5 silver g skill royal cl28 6,000mhz which I will buy next week/ lian li o11 vision case/ I also ordered all of the lian li wireless strimmer cables with the wireless controller which I should also receive later this month and I will give my bro my current lian li v2 strimmers. Ssd's are the 4tb gen 5 crucial t705 and the gen 4 samsung 990 pro. I am looking forward to a white 5090 and I will most likely go with a zotac or gigabyte and maybe an asus 5090 depending on how much the asus tax will be lol. I will sell the 4090 to put towards the 5090.
Hey i had a maybe unrelated question but i wanted to ask it nontheless,
See I have a 4070 paired with a 5600, would it be beneficial to upgrade to a 5700x3d ? I primarly play at 1440p with high settings on fortnite and singleplayer games
Yes
u really dont have much upgrade path for 5600.
5700x3d is your best bet.
yes, if you can get one, get a 5800X3D (which is sadly unlikely) (maybe used? they can't be overclocked anyway).
It won't be an insane jump, but there will be a difference. Maybe watch some old Gamers Nexus Benchmark videos so you can determine if you think it's worth the money for you.
at 1440p with a 4070 a 5600 will do fine for most games. a 5700X3D will give you a very small uplift in some games. Especially the 1% lows. However, as soon as you enable DLSS upscaling that 5700X3D will probably give you more frames in some games VS 5600. Also, 5700X3D will set you up for at least one more future GPU upgrade.
@@larsenmats yeah I guess the dlss argument is relevant cuz I always enable it,also I was planning to do a whole new build in late 2025/early 2026 with a 9800x3d and a 5070 ti/5070 super(the 18gb model)
If you look closely, the video starts at 0:00
Awesome benchmarks. How much performance difference at 1440p and 4k? Is there any benefit at those resolutions?
The new format for the head-to-head comparisons is great! Keep up the amazing work! 😊
Imagine having a 4090 and 9800X3D to play at 1080p.
You test TWO Assassin's Creed games, and TWO Horizon games but no Alan Wake 2? No Dragon's Dogma 2???
Alan Wake 2 is a GPU bottlenecked game
At what time were you running the test? I don't game after midnight so these results may be unrealistic for my use case and I know there is a big time shift due to you being in Australia.
I'd also like to see testing on some tall buildings because it might affect airflow with reduced air pressure!
@@justmatt2655 Good call. Also having AIO cooled PC on top of a skyscraper may cause the liquid in AIO to boil faster due to change of pressure which may lead to worse CPU performance.
@@WereCatStudio yes exactly these are the things people just aren't testing
An amazing piece of work, as always from you guys! Is there an option to add the 1% low fps differences to the summary graphs?
I know it's for the sake of comparison but people playing these games aint playing on 1080.
Can we just get a battle of the X3D chips including the 7600x3d, please sir.🙏
so for 4k gamers no need to upgrade from a 5800x3d at least untill 6090/7090.
Yeah. Zero reason.
Probably not. Once cpu starts hitting 90% plus it will be time to upgrade.
@@rogerk6180 that is not how this works.
i doubt you will get much more fps in 4k maybe if you overclock the 9800x3d you might see afew more fps but that depends if its worth it to you for afew odd digits more
I'd like to know if it offers any benefits for Helldivers 2, sadly I can't find anything on that
Great work on the 1080p CPU LIMITED testing
I like this new format. I only listen to your videos, and I watch them only when I need to see a specific graph for a game that I play. It is helpful for me that you named almost every game and describe the performance difference between these CPUs. Also, I have a 7800X 3-D, so it was particularly pertinent.
My 5800x non 3D giving me the “I’m tired boss”. Been holding off moving from am4 but this is really making me considering it.
Plz add percentage change next time in your graphs. I know u saying it, but its much easier to recognize. Anyway you guys are awesome!
10:00???
Will be interesting to revisit the games that did not see an uplift when the 5090 launches.
Also simulation titles please, so many people play those games for years and years and years that they are severely under represented in review suites.
Testing a CPU in 4k is like testing a tire but the car its mounted to is only allowed to go 30.. never reaches full potential
I would say testing cpus in 1080p is like testing an off road tire off road. Yes it's what it's meant for however 90% of the time people will still be using it on a normal road so you want to see how it performs there
@@phillyd2018 someone has to tell them that buying an "offroad tire for a normal road" is the best way to waste money.
Testing CPU in 4k is as clever as with a 4090.
If you genuinely build a 9800X3D paired with a 4090 (or God forbid a 5090) only to play at 1080p there's something very wrong with you.
@@selmevias1383 Even a $190 Ryzen 7600 can reach +90 FPS in all games tested. At 4k it is the same limit, resolution does not use CPU resources.
I would love it if you added an “esports” section to these reviews and quickly showed results of a few popular competitive multiplayer games using realistic competitive settings (e.g., performance mode for Fortnite). I know the market is small but I think there is an opportunity for you to become the definitive esports benchmarker for CPUs/GPUs, which should add to your viewership over the long term. You cover esports titles occasionally but I am talking about a consistent inclusion of these titles at lowest quality settings. Love the channel!
Please show the comparison against the 14900K at 1440P & 4K also if you can. And overclocked comparisons also are appreciated. Thx Steve ❤
You guys are just the absolute gold standard for hardware testing, keep up the great work!
Holy shit, 45 games benchmark. You da goat
All that work for only 1080p results...
@@L0Lyeahright less demand and fomo if they show 1440p or 4k results which may compete with 1700X lol
I'm more anticipating the 9950X3D
...those never will work. :(
@@GameBacardi Unless they do the unthinkable and put extra cache on both CCDs
@@GameBacardi it was literally already leaked 😂
I wonder why so few people actually use CPU intensive games when benchmarking a CPU. Use Stellaris, use Factorio, use Rimworld to show the CPU performance. THOSE kinds of simulation games are the ones where the CPU matters most.
Can I request for 45 game benchmark using max settings at 1440 and 4k? I know this is not how we test CPUs, but I'd like to know if upgrading to 9800x3d will be worth it at the graphics settings I play in. You can even split it into 2 videos, 1440 and 4k for more views 😅
No one can tell you if it's worth it besides yourself. And only after you actually do it and can compare the experiences.
I am playing at 4K with a 7900XTX and I am planning to upgrade from 7800X3D to 98000X3D. Why? Because it fits my use case. And because CPUs define so much more than just FPS numbers - how responsive the menus are, how fast the games load, how smooth the scrolling, how well the performance holds in more demanding areas and especially how well older games that are very heavily modded perform overall.
You suddenly realize an 8% uplift is huge when you're already pushing 240FPS. Combine this with the 5090 and 8K, 120hz/4K 240hz Gaming is now a reality. Can't wait to see the power the 9950X3D will bring, we will finally get the perfect CPU that can not only have great productivity OR gaming, we will get the perfect CPU that will have great productivity AND gaming
9950X3D will not be much better. As long as only one CCD have the extra cache you'll still be on an 8 core CPU while gaming. Maybe a 50-100MHz higher boost clock if you're lucky.
Then there is the scheduler problem. It is not perfect and in some games you end up using the wrong CCD.
@@Gazer75 The rumours are that both CCDs get extra cache this time. GN's tour of AMD showed a dual 3D chip stress-testing in Task Manager. The engineer said it was far too expensive to be a consumer product at that time. The new layering could make it a reality, although still expensive.
You're CPU isn't going to help much with 4k and up gaming bud, also the 4090 barely does 4k120 why do you think the 5090 would somehow do 4k240, I have a 4090 with 7950x3d I get on average 80 -100 fps on most graphical games with everything maxed
Im getting this cpu for my 1440p 360 hz rig, hoping to upgrade from a 3080 when the 5080 or maybe higher comes out
Not 8k
I'd like to echo some other comments here of people wondering about 1440p and 4k benchmarks because those are better real world scenarios for people that are buying high end CPUs and not playing esports titles.
Obviously 1080p is going to show the biggest contrast between CPUs specifically, but it would be nice to know how much the high end GPUs at the higher resolutions even really care about CPU performance.
If you want to know how a game will function at higher resolutions then you should watch GPU comparisons. For example, at 1440p or 2160p, both the R7-7800X3D and R7-9800X3D will have the same performance, held back by the GPU.
You won't get that from HUB. They are being paid by AMD to manipulate the test conditions so that this CPU 'looks' better than it is. They have ZERO interest in helping the real-world target audience of this CPU. Unfortunately.
Just look at the benchmarks of your gpu of choice at 1440p or 4k. That is the fps these cpu's will maxmout at.
Not rocket science.
Sure that's true for modern and last generation CPUs. But what if someone is running, say, a 3600x still. Would be useful to know at what point you're going to see gains from a CPU upgrade vs just dropping in a shiny new GPU.
So watch a GPU benchmark video instead of a CPU one?
Sure is better at 1080, but at 1440? 4k? I don’t think everyone plays at 1080 nowadays, would be cool to see how it goes with the other resolutions.
Thank you guys for your hard work and the extensive benchmarks that you are doing.Whenever I want to know how a new release is performing the only place I look is here because I know I will get the best non-biased review on the internet. Keep up the good work and thank you again 🙇
I’d love to see the 5800X3D also in this graph.
Allow us to track the generational progress in this AMAZING product stack.
Nearly everyone is glossing over the fact that the 9800x3d is an overclocking MONSTER with massive gains potential. Any Joe Schmoe with a basic AIO can overclock it to 5600mhz all-core, and for the tinkerers who are willing to delid it, we've already seen 5750 all-core clocks being achieved pretty easily on day 1 with just a basic OC. That is a 500mhz increase, so that 11% uplift in modern games could easily be closer to 16-18%, and even more extreme uplifts to the 1% lows.
Hardware Unboxed do the most thorough CPU benchmarking of all the TH-cam channels out there. Great work guys.
Thorough? I only saw 1080p results...
HU are the best.
@@L0Lyeahright He literally explained at the beginning that if you want to know why he does that and why basically any other reviewer who reviews a CPU uses 1080P you have to look in the video description. In the video description there are 3 videos that go more into depth as to why.
In short, 1080P is currently the best possible way to show how much more performance a certain CPU has over another while minimising GPU usage during a particular benchmark. Again if you want a more detailed explanation, look in the video description of this video :)
@@emiel255
We know why 1080p is used. The problem is that ONLY 1080p is used anymore, when we used to get 1440p and 4k results as well. Also, 1080p is too high a resolution, based on their logic, and why is a graphics card even installed in the system to begin with for this testing?
@@L0Lyeahright Yes, that is true, but I think the reason they don’t do lower resolutions (like 240 or 720P) is because basically no one plays on 720P anymore but still plenty of people use 1080P. I think it’s just the minimum resolution that most people play now nowadays so that’s why it is better to use that as a CPU benchmark standard. Otherwise we could basically say let’s go down to 240P or something like you basically said. I’m not sure about this so if anyone knows better let me know
And I understand that you would like to see 1440P and 4K but it it will show nothing of valuable data (HUB and other reviewers have explained and showed this numerous of times) which causes the 1440P and 4K data to be meaningless as it will say absolutely nothing about how much extra performance the CPU has over other models. You will see only a meaningful difference at 1080P.
You should definitely check out the HUB video called ''Why Reviewers Benchmark CPUs @ 1080p: Misconceptions Explained''. It is a very good video
Unfortunately the 1080p rtx4090 test skew the results for everyone with a 4090 and plays at 4K
For you to be able to quickly just do 2-3 games at 4K would add at least a reference point of “hey, if your gaming at 4K you will see a .5% increase in most games and there is no reason to upgrade your CPU”
Because some people will see results like this and assume there is some kind if not similar performance to be had.
Lots of people have money to blow but neither the understanding or time to actually know what effects the system as a whole.
Respect for seeing through HUB's bullcrap manipulation. As you realise, HUB already knows that 4K results show little/no improvement over the 7800X3D. Why did they not say this to the viewers? - simple, they were paid by AMD to test ONLY at 1080p because 4K results would be embarrassing for them. As you already understand, HUB's exclusive 1080p results here are from cloud Cuckoo land because NO ONE with a 7800X3D or 9800X3D plays at 1080p, obviously.
In the end, this testing configuration serves to mislead the uninformed while pissing off every owner of these actual CPU's.
if they have money to blow and no time to educate themselves on basics of what they're blowing their moneys on, then it's not the reviewer problem. On top of that I'm sure there're plenty of people who will eventually do pointless 4k benches for CPUs anyway
What reviewing TechTubers should do is.
Benchmark at 1080p
Benchmark at 1440p
Benchmark at 2160p
And tell us what the percentage gains are for each resolution, this way a lot less desktop people will be mad at your one sided benchmarks.
Why is Linus the only person showing quick 1440p and 4K resolution, can you TechTubers explain that to the public?
2:00
@@ScruffyIMS Still we don't have 1440p and 4K benchmarks and the percentage gains for those resolutions.
1080p is not the all and only CPU benchmark to show the percentage gain for the CPU. On that ground you could go 720p with all low settings and show the world 1000 FPS in games and +50% FPS gains.
That will paint the CPU as +50% faster than the previous gen, which is an odd ball as only 1080p benchmarks.
@@N0N0111The FPS limit of a CPU is the same at 4k, there is literally nothing to test.
720p runs weird on the 4090 and some settings run on the CPU. 1080 ultra is the only way.
@N0N0111 yes, we have those numbers. You have to look at a Gpu test ;)
Your Gpu only does 60 fps in 4k in your chosen game? All the Cpus achieving 90+ are fine.
Your Cpu doesnt care about resolution, and you want to test the part you are testing right now at the limit. And think about Steve, he can't do all these tests again x10 for different Gpus. Not everyone has a 4090, I'd guess a smaller part of the audience. The tests would be meaningless for everyone without one.
Why does this only (mostly) come up with Cpu testing? You don't see all the cries for 1080p testing with a Ryzen 3600 for the Gpu benchmarks.
It's easy to find out if you are Gpu limited, only you know what you play with which settings. If you are, don't bother with the Cpu upgrade. But I want to know which Cpu is faster
I know this is a big ask, but there is still huge World of Warcraft playerbase but nobody really benchmarks the game to see how the new hardware handles it. Would you consider adding WoW in some capacity? ❤
Thanks for the video! I would have loved some data for strategy/management games too (Stellaris, Civilization, Cities Skylines, etc)
They dont want to show 4k or 1440p because the gap isnt there.
But 1% low continues best in 1440 and 4k
Cause that's a gpu benchmark. Go back to linus
Thanks for this! Considering moving from my 14900k to a 9800x3D, would’ve been nice to see difference in power consumption though as well
@19:00 on their video titled: "AMD Ryzen 7 9800X3D Review, An Actually Good Product!" you can find the gaming power consumption, and @8:30 for synthetic benchmark
The have already done this in the previous video.
Day 1 of asking for 1440p and 4k benchmarks, because almost half of us don’t play at 1080p.
that doesnt makes sense
No point doing benchmarks at anything higher than 1080...both are GPU intensive. You always test CPUs at 1080p. At 1440 and 4k you just don't want you CPU to bottleneck the GPU.
@@5ean5ean22 “At 1440 and 4k you just don’t want your CPU to bottleneck the GPU”.
That is the exact reason why we should test games at 1440p and 4k. Some CPUs can bottleneck your GPU. That is the reason we test at 1080p. 1080p is where the greatest bottleneck exists. If we test a new CPU at 1080p and see a 5% gain, we can conclude that the new CPU is less of a bottleneck compared to the other CPU. At 4k and 1440p there are plenty of games that provide a 5% difference. Of course there are some games that have no difference and are GPU limited.
I don’t disagree that 1440p and 4k are more GPU intensive but there still are some significant gains to be made with some CPUs.
@@A_GWAKwork_OrangeI understand the whole concept of bottlenecking and how higher resolutions are more GPU limited.
I don’t disagree that the gains are small. But single digit gains like 5% or 8% or sometimes even more exist at 4k and 1440p in many games like Borderlands 3, MSFS, and Stellaris.
Of course some games will show absolutely no gain because they are GPU limited.
My point is, if an 8% difference is significant at 1080p(ie. The advertised improvement from 7800x3d to 9800x3d and 1080p) then why is a single digit improvement at higher resolution not significant? Again, I understand plenty of games will show such small differences that it doesn’t matter, there are plenty that do.
@@5ean5ean22 This is a valid PIONT!
Why You need better CPU, when cannot utilize because GPU limit?
Thanks for bringing back MSFS2020.... Hoping to see MSFS2024 next month in charts 😉
Also nice to see War Thinder again, will be interesting if that will show up in RT charts as they are now adding RT to the game
What I found amazing is the performance in Stellaris simulation time in Gamer Nexus video. For people that plays that game is a no brainer upgrade. Huge improvements that translates into wasting much less time waiting for the game to process things and actually playing.
who's playing with a 4090 and this CPU at 1080p ?! Do tests at least in 1440p !
Soooooooooo keep my 7800x3d?
Absolutely. I'm also keeping my 5800X3D.
Definitely. 7800X3D is way more efficient
Basically yeah. I am glad I didn't wait to build my new system (first in 8ish years). Since I did the math and in Aus, it's about 44-45% more expensive to get the 9800X3D Vs what I paid for the 7800X3D a few months back. Good timing I think 😆
Why all the test are in 1080p... seriously... People who spend this kind of money don't play in 1080p! we want to see 4k results. If you're going to play in 1080p, why even bother with this cpu and a 4090... fcs...
shouldnt you talk to them instead though instead of hurting me lol, logic isnt easy to comprehend
ive never thought abt such things, and quite surprised its normal in your environment
Thanks for the War Thunder Benchmark!❤ Plus... 9800X3D came to Papa😅!
Thanks for all the tests at 1080p, I understand it more gpu limited at 1440p and 4k, and 1080p shows the differences better between the cpus, i still would like some 1440p testing as well
Then you didn't learn anything from the numerous videos stating why they use 1080p.
@@sabretoothauit’s ludicrous, I’d just look at cinebench and 3dmark if I wanted non-real world scenarios of how the chips perform. Buyers of this chip are not gaming at 1080p which makes any justification irrelevant
@@sabretoothau i know why it's less important, but it's still important to me