Love to see my Q&A question turned into a whole video! I agree with other people in this comment section in that this opens up a whole spectrum of comparisons related to "what if you spend more" or "what if you spend less" or "what's the cost per frame of these setups".
I'm also one that's asked for this several times, but Steve thinks it's ridiculous. THIS is the info that's so rarely seen because all reviewers only use top end pairings to "remove bottlenecks" without realizing knowing WHERE those bottlenecks are is VERY relevant information to anyone building on a budget. Knowing what's the fastest with no budget is easy. Just buy the top part. I don't need to watch a review to tell me that. I guess we're just supposed to _guess_ where those bottlenecks are when you're building a realistic system or wondering if you need to replace your 12600k if you're thinking of buying a 4080 super.
Skewing as much of your budget toward the GPU as you can get away with has been common wisdom in the PC building community for a long time now, but it's good to see periodic testing to prove the point anyway.
A good cpu is going to age better in your system than a good gpu. It's easier to turn down settings and still get good performance where if you're cpu bound you're pretty much screwed. And it's easier to upgrade the gpu later.
I think it's a mistake, the GPU is the easier component to upgrade. With a worse CPU, when in 3 years you'll get your new 2-gen-newer GPU you'll have some problems.
This is one of the most useful videos i've seen in a long time. Congratulations! I wish you would do these kind of comparisons with the most common GPU+CPU combos.
@jakacresnar5855 I originally wanted to do that, but I want to play a game called ark Ascended, which is a remake of Ark Survival Evolved. It's probably the most beautiful game out there IMO, but it's very unoptimized. I am talking 4070super 40fps native 1080p unoptimized. It's enjoyable with frame gen, but If I play at 1440p my fps will even with frame Gen constantly be
That's the main advantage of the 3D cache. Even if there is no improvement in average FPS, there is a great increase in 1% lows. This is true even at 1440p and 4K.
That was also a noticable thing when 5800x3D was released. x3D's also helping a lot with truly bad coded games. On jedi survivor, truly terrible coding on this game, at least with 5800x3D I got much less micro stutter than my friend with an i7 13700, while we both have a 3080 graphics card and game installed on a gen2 NVME. And this although benchmarks show similar results on 1% lows, the experience is smoother on 5800x3D.
Could have been relevant to include one or two example results of the 7600+4070 and 7800x3d+4070ti for reference to see what the additional savings vs spending would also do here. Might be relevant especially in the cases where the result was veered to more gpu or more cpu limited.
GPU pricing may change to the worse or the better depend on supply & demand, some political move also play a big part of it. So it's better to get as high GPU while you can. CPU pricing will always go down so you can upgrade later.
By that point a 5070ti or whatever will be out that matches a 4090. There's an argument to be made for going 7800x3d + 4070 super. The upgrade 4070 super to 5070ti or whatever after 2 yrs. With how upgradable AM5 is, the best value would be 7600 to 9600 and 4070 super to 5070ti(assuming it comes close to a 4090)
That's what I did with AM4 - ditched my i5 4690k/Z97-I Plus (Asus mobo) when Ryzen 3600 had been out about 6 months - yes I kept that older system for years as it played my games just fine & all new Intel chips were only slight improvement - So I bought Asus ROG STRIX B450-I Gaming mobo/Ryzen 3600 & moved my trusty GTX 1070 (EVGA with thermal pad mod) over to the new system, upping the GPU to a 1080TI a few months later as it was great price off of eBay - did another upgrade a year later to 5900X/RTX 3080TI - these kinds of upgrades were unheard of with Intel platforms 😮 I did completely disregard Hardware Unboxed review on this Motherboard around VRM heating issues - I'm running 12 core 24 Thread CPU & a power hungry GPU (350+w) & this board is completely fantastic 🤩 - I do have the GPU on a riser so it's not radiating heat to the mobo (modded Gamemax Brufen C1 case). I will definitely be moving to AM5 but unsure about A620 or B650 as I tend not to overclock anymore (CPU/GPU are fast enough at stock for me - only Benches I overclock).
@@warsshanonly if Nvidia follows it's previous cadence. But there was a leaked Nvidia roadmap in 2023 that showed they planned on releasing 5000 series in 2025. And rumors late 2023 said Nvidia was gearing up and would be ready to respond if AMD released new video cards in 2024
That was my strategy, except I rolled over the 3080 from my previous platform. Well and truly still good enough at 1440p for what I mostly play. Significant FPS and particularly 1% low jumps from the 11400f/DDR4 platform
I would have liked to see results for at least one or two popular strategy/city-builder games, like Anno 1800 or City Skylines, since these games are really demanding on both CPU and GPU.
Same. I actually thought they were going to, since they specifically called city builders out in the beginning. It would be super helpful to have tons of games ranked by how GPU vs CPU bound they are. Baldur's Gate 3 is another one I've seen called out as surprisingly CPU bound. MSFS as another.
@@Chilith yeah, cuz i mostly play cpu bound singleplayer games , , i willing to combine ryzen 7 7800x3d with rx 6600 non xt which is very stupid , unless you play mostly cpu bound games whether its singleplayer or multiplayer
If you can get one where you are it certainly makes sense to really follow the min max strategy. 7600 prices are too high to really consider it a min, but I don't think the price gap between 7500F and 7600 would let you buy higher up the Nvidia stack anyway. Might let you get a 7900XT though.
Yes but if people still have R5 5600 the upgrade should be overthought. 7500f and 7600 are only considerably faster in 1080. if u have 4070 ti that makes no sense .
@@Greenalex89 Yes that's true which is why the 5800X3D is good for an upgrade. However it's a bad idea for a new system due to the price and lack of upgrade path.
I would've appreciated it more, if you would've included the combinations of 7600 + RTX 4070 and 7800X3D + RTX 4070 Ti in this comparison too. I think is could've shown better the CPU or GPU "bottlenecks" in some titles.
This has been super illuminating, I'd love to see more comparisons like this moving forward as they directly address our real-world decision processes.
I have a R5 3600 with a GTX 1060. I plan to upgrade to a 4070 Super when it releases💀But I am also planning to upgrade to a R7 5800X3D not too long after. My CPU still feels fine for now. It is definitely my GPU that is holding me back from enjoying modern games
i had a r5 3600 with a 1660super, recently upgraded the gpu to a rtx 4070 still with the same r5 3600. works okay as the most demanding games are usually gpu intensive, but a cpu upgrade would definitely be worth it if i can get a good deal somewhere
@@FATBOY_. most games it doesnt, i upgraded to 1440p about a month after i got my rtx4070 so i played on 1080p with that combo for a while and it was mostly fine. honestly its kind of hard to say with the amount of shitty optimised games this year. it'll definitely be worth it to upgrade my cpu at some point but its not a must and my experience now is great already
I’m thinking of pairing the 7800X3D with a 4070 super, after some time upgrade the gpu to the new generation. This way because the 7800X3D is a powerful chip It probably won’t bottleneck the future generation of GPU’s and you’ll get 5+ years of usage. Hopefully 🙏🏼
Finally, thank you guys, I was thinking the same since the Q&A question and trying to find relevant testings with little success. I would really love to see these kind of comparisons in the future, maybe cross test them as well so both GPUs with both CPUs. I know that a RTX4090 will get you the best CPU results when testing but for some the 4070 and the Ti is more than we can afford and if someone is planning to spend this kind of money I think tests like these are perfect to help us get the best for the price. Cheers!
I'd like to see HUB branch out into doing more hardware reviews which I think is the wheelhouse of this channel. Maybe cases? When there are no new cpu's, gpu's and motherboards to review it seems like there is just the Q&A's with Tim and Steve.
People don't get into cases that much. They're usually GN's lower viewed videos. They matter, but they're not exactly exciting. I'd rather see more of this mid range pairing when new releases slow down. Judging by these comments I'm not alone.
Yeah, I just used cases as an example. I love this channel but it seems like the ratio is too highly tilted towards the Q&A stuff lately. I rather see reviews, benchmarks etc...@@zodwraith5745
I just think there are too many Q&A's lately. Steve, your reviews and benchmarks are the best. That is what I enjoy. I can get podcast type stuff elsewhere. I mean all this in a constructive way. Not as any kind of criticism. @@Hardwareunboxed
I would very much like a video about long term choices when picking pc parts. This always has been in the back of my head for a long while. It's easy to pick today how to buy pc parts, you buy what's best for your money. I've been building PCs for a long while and nowadays I sometimes don't necessarily buy the best bang for buck at the time of purchase because I think about the long term cost. I'll use a concrete example; way back I had to pick from a i7 4770k or the i5 equivalent for haswell. At the time I went with the i7 even if it didn't really made sense in gaming benchmarks. I've had this cpu for 7 years or so and I'm pretty sure it lasted one GPU upgrade longer because I had hyper threading. So the 100$ difference saved me from buying new ram, mobo and cpu for a few more years. Same thing with the power supplies, I've had many friends upgrade their 550-650w power supply, not because they aren't working, but because a modern intel gpu and 3000 series gpu was pulling too much for the wall. A decent power supply can last a decade, might as well invest a bit more so you aren't forced to upgrade while just halfway through it's life span. I'm very happy with my overkill 850w power supply at the time, it's a part I can keep for a long time. Another example, I still have a nhd15 that I got in 2015, so many of my friends cheaped out and had to upgrade many cheap aios or air coolers I'll still use this cooler in a decade probably.. worth the 50$ premium I paid for vs buying a cheaper cooler and having it be replaced because of the lack of forward compatibility. Picking a cpu has quite a bit of implications, because you are also picking a platform. In this example, you've got am5 first generation, meaning that there's an upgrade path without buying a new motherboard and ram in a few years, I wouldn't mind cheaping out in the CPU department knowing I could get a much better CPU latter. If I was picking up an am4 cpu right now, I would get a higher end model knowing there's no cheap upgrade path without changing ram/mobo, might as well get something that might last me one more GPU upgrade. It's hard to reason about it because you also have to consider the selling price, but most of the time the obvious option might not be the right pick if you look at from a longer timeframe.
Excellent video. This pretty much exemplifies _why_ I've always argued that reviews of mid grade parts need to be paired with realistic counterparts. Or at the very least remember to throw in 1-2 benchmarks to remind people this. Most often your reviews _DON'T_ remind people with realistic budgets that the min/max of CPU over GPU is usually the most beneficial. Having the 4090 have such a huge performance gap to the rest of the pack just amplifies this problem further. It's easy to see where someone with mediocre knowledge of building a PC would see the massive difference in the 7600 and 7800x3d when paired to a 4090 compared to the much smaller differences between a 4070 and 4070ti paired with a 7800x3d, then incorrectly assume "Well, obviously I'll see more gains with the faster CPU at the same price." And this is when the 4070 and 4070ti are still relatively powerful GPUs for what many people can afford these days helping the x3d look even better. The problem gets compounded even worse as you move into lower tiered GPUs which would be quite likely with current prices. Do the same comparison with a 4060 and 4060ti that many people buy and you'd rarely be CPU bound even in competitive shooters. This would show in most games that CPU is completely irrelevant when paired with the most popular GPUs and using settings above 1080 low, as long as you're at least using a modern CPU. While I understand the _theory_ that you need to remove any bottleneck to amplify differences in parts being reviewed, that's not how real people build real systems unless they're building a $3500+ system. This is why I think it would be beneficial to show at least 1-2 "don't overspend on the wrong shit" benchmarks to show real world pairings. It's insane the idiots I've built systems for that insist on an overpriced X3D because they saw the reviews just to pair it with a 3060 or 4060.
In reality the opposite is true, you should always remove bottlenecks when testing a component, you want to test that component in isolation as best you can. Reviews aren't designed to convince you to upgrade, if that's how you view them you're doing it wrong. Reviews are there for when you've decided it's time to upgrade and they should show you exactly how competing parts really compare. If there are two CPUs that cost $200, why would you want them compared with a low-end GPU? Those GPU bound results are only testing the GPU and tell you nothing about how the CPUs compare, one CPU could be 40% faster than the other, but you'd never know. Any heavily GPU bound CPU testing is extremely misleading, as is any CPU bound GPU testing.
@@Hardwareunboxed Which would be exactly _why_ I said include 1 or 2 benches with real world pairings to remind people to build balanced systems, not omit bottleneck removing pairings all together. I was agreeing with the point of the video all along, Steve. You of all people should know just by some of the BS you see in comments that people need to be reminded more often about building balanced systems. Especially if they're watching as a buying guide and not just us enthusiast fanboys. You're absolutely correct that not removing bottlenecks is misleading when reviewing parts, but _not_ showing where those bottlenecks will occur is just as misleading when someone is trying to spec out an _entire_ system. Not just the individual part. That's why this video was more valuable than you think. There's a lot of people out there that don't have the understanding of the basics like you and I do. And they sure as hell haven't gone through as much hardware as we have and just automatically know where those bottlenecks occur.
You're asking for the impossible, as this video has just demonstrated. Who are you to determine what 'useful bottlenecked' data people should be presented with? What does that even look like? What hardware are we using? What games are we testing? What are the quality settings and resolutions? You are working under the assumption that all gamers play the same games, the same way, when in reality that's not the case at all. Of course we can test a 7800X3D and a range of other CPUs with a low-end GPU and show them all rendering 60 fps, how is that useful? I assume you've watched this video: ruclips.net/video/Zy3w-VZyoiM/видео.html
@@Hardwareunboxed _You're_ working under the assumption that I'm saying all your testing needs to change, which I never did, and I haven't said anything that was incorrect. I simply stated it would be nice to get a graph once in a while to show _why_ you build balanced systems, because a lot of people don't seem to get that. Which you yourself HAVE done a video like that in the past and I praised you for it. Ironically your theoretical test at the end WOULD be useful because it tells people if you can't afford a 4 figure GPU don't waste your money on an overpriced CPU. You yourself constantly recommend the i5s and R5s. You think this video is ridiculous but from my point of view it carries a message that is rarely heard: Don't buy more hardware than you need. You just PROVED that by showing the second you're not playing at 1080p in competitive shooters that GPU is more important. You chose 4 very relevant parts and proved WHY "it depends" is on your shirt. You keep getting these questions _because_ it's rare anyone gets to see mid range pairings. We know you can't test every configuration out there, but don't attack the viewer for wanting relevant information for realistic builds, when you provide more than enough info that's irrelevant to anyone building within a budget. Almost no one asks me to build them a "money doesn't matter" system anymore because it's gotten horrifically expensive, but TOO many people insist on an X3D so they can pair it with a 3060. That's a problem and at least now I can link them this video.
Would be funny that RTX 4070 Super will release later and pairing with the 7800X3D will be the perfect 3rd option/answer in this comparison. If 4070 Super is indeed close to 4070 Ti performance.
This was a great video. I actually just assumed almost nothing is CPU limited anymore (and functionally yeah above a certain fps it's just academic) but I was shocked that the 7800x3d4070 system won anything. I have a ryzen3600/rx6800 and I wonder sometimes what would be the best upgrade path from here, and since I do play VR maybe the additional frames would make something like a 5600x3D a worthwhile purchase.
New CPU won't help VR. There's still something with Radeon that causes the 1% low issues. AMD admits it but haven't fixed it for years now. I can't use my AMD system for VR so it stays on one of the Nvidia systems.
As I said in the community post. It 100% depends on the games. AAA is better with the beefier GPU but there are plenty of other styles of game where the CPU will matter more. ARPGs, especially late game like Path of Exile get very CPU bound, simulation titles like ACC, iRacing, Flight Sim, Cities Skylines can be very CPU bound for both frame rates and the games tic rate. Strategy titles can also be very CPU dependent for tic rate and simulation speed. Then there are MMOs which also get CPU limited in large fights where you actually want to maintain a reasonable fps. So while this is a pretty good video the conclusion was predictable purely based on the titles tested and to do this kind of test justice really needs a far broader suite of games that cover all the bases so people can actually make use of it to inform buying decisions.
This is a really good video. You should do this more often to help people understand how old cpu's work with new gpu's and whether or not they need to upgrade CPU or GPU first!
@Eternalduoe just suggested that too. At what CPU does the frame rate stop getting better. Couple that with asking at what CPU is it better to upgrade the GPU instead. This would result in the most perfect balance between CPU and GPU.
This is exactly why I bought a 4070Ti and 7600. At the time, it was the best value and will hopefully be an upgradable platform for a long time. Not thrilled about the price of the 4070Ti, but that's a different story for a different time. I consistently get 5.3GHz on the CPU with the BeQuiet Dark Rock TFs and my system is whisper quiet. No regerts.
I recently put together a 4070 and 7800 X3D build and I'm pretty happy with it. I had initially got the card first for my older system but decided to update my whole system and switch from Intel to AMD as it was aging. It's nice that I can keep the powerful CPU and always upgrade to a better GPU in the future.
Am planning to upgrade to the same specs soon as well, currently rocking an old i7-4790K + RTX2060. I'm working as a video editor, and have finally save up enough for an upgrade :') And since I'll probably do 60% work/productivity and 40% game on my pc, I hope I won't regret with this decision 🙏
I feel like the 7800x3d combo will have more utility once a single GPU upgrade comes into the equation (in the event that the entire system isn't done at the same time).
@@Grena567 I think typically speaking your average user is less likely to perform a CPU upgrade than a GPU upgrade, even with the longer tooth support we enjoy with current AMD platforms.
Yeah. That 35% advantage in Avatar and 27% in SWJS at 1440p is pretty impressive. 1440p gaming at Max settings is clearly very GPU demanding. And 4070 is not powerful enough for stable 60 fps experience.
When I moved apartments recently, I decided to save space and consolidate my two PCs into one SFF pc using the latest parts from each build. I saw basically no change in performance going from a i7-11700KF + 3080 12GB to a i3-12100F + 3080 12GB. I ended up switching to a 12600K on sale, but I'm not even sure it was worth it since that 12100F was such a little powerhouse.
That was great, Steve. This is useful info for determining where to spend limited funds. Stir into the mix a 7700 and the new 4070 Super for those seeking a middle ground. In the same ballpark cost-wise.
I watched my i7-12700K destroying the Ryzen 7 5800X3D for emulation by 51%. It chills at very low CPU usage for anything minus any console emulators and UT2004 (singlethread bottleneck, barely more than 300fps).
More performance now: skimp on CPU (although here skimping is a big word, the 7600 is a great CPU) More performance over the lifespan of the platform, assuming 2 or 3 years upgrade cycle: better CPU, you swap the 4070 for a 5070 or 6070 and you can keep the rest of the platform untouched. Still I "built" 3 PCs so far and every single one of them picked the better CPU than GPU option simply because the 3 people using them (me included) use the pc to also do very CPU intensive work (e.g. python and AI classes), so the 5800X3D for me and yhe 7800X3D for thebother 2 was basically a necessity (I got a 3060, the other 2 a 4070 each)
I really appreciate this type of video! One thing i would love to see explored is where the ceiling in GPU performance for a particular CPU is. E.g. For a 7600, which GPU generally maxes-out the performance of that chip? I envisage it as 4070/3080 Ti/4070 Ti/7800 XT/ 7900 XT comparison. Or something similar...
You can get an idea of this by looking at the different frame rates per resolution. You can see that generally the X3D system wins at lower resolutions and the Ti system wins at higher. Also the 1% lows are better with the X3D. Something is going to be the bottleneck. I had a chap with a GTX 1060 and a gen 4 i5 at a time when new GPUs were super expensive. He wanted a new GPU so I upgraded him to a RYZEN 5 3600 whilst still using his original GPU. He was delighted with the improved frame rate and smoothness. His old CPU was not driving his GPU hard enough. There is not so much of a performance ceiling but a more effective way to spend your money.
I just bought a 4070 super fe and was planning to buy ryzen 7600 to go along with, then came across this video! Helped quite a bit, thank you very much..
You guys should really add benchmarks of games that are *actually* CPU bound to these videos. Grand Strategy games like Hoi4, Vic3, Civ6, etc. would be really useful to also have in benchmarks of these types.
These games are CPU heavy but you don't need 300 FPS like in CS2. 60 FPS is 100% ok. So in reality these games are very little CPU limited. 60 FPS can achieve even the Ryzen 5 5600.
@@andrewcross5918 Which means what? Even fast pace competitive shooters like Fortnite or COD have only 30Hz tick rate. Even the Ryzen 5 3600 can achieve that.
If you look on paradox forums you'll find benchmark threads for their games. I recently looked at the victoria 3 benchmark thread and late game the 7600x is about 50% slower than a 7800x3d ( about 3 mins to simulate a year in 1900 with the 7600x versus 2 mins on the 7800x3d ).
these are the videos people need. You might want to include the results of the best combo as well. In this case 7600 + 4070 ti // 7800x3d + 4070 // 7800x3d + 4070ti. That way people can see how much they are actually compromising and it gives them a path to follow in their choice between bang for buck or ultimate performance.
Now I would have liked the 5800x3d thrown in as a bonus, paired with the 4070Ti. I hope you guys will be able take a look at the 5700x3D now that it has been announced officially.
@@GLDragon93 Valid point actually. Plus AM5 and DDR5 is more expensive, though the 80€ saved on a 7600 could probably cover the cost difference and even it out (unless you're already on AM4)
imo that used to be the case - but until the gpu manufacturers stop pricing in the fps gains you get as extra cost, I doubt this rings as true as it used to
The GPU market is dogsh*t. I can't find any good price perfomance pairing for my i7-12700K. A friend had luck and he got the 3080Ti for 700USD (the 4070Ti sucks, 800USD for 192 bit memory bus).
@@beachslap7359agree. I think this was a rare dud for not covering that angle. Both cards will age similarly while I expect the CPU will age well. Better to put that money towards the CPU now and have the tidier upgrade path.
My 7800x3d and 4070 was surprisingly amazing for 4k. I got it with only 1440 at first and decided to push 4k, of course I still plan to go for 4080 super soon because as amazing as it is, it is still struggling at times.
I use a 4070ti + 5800x3d and most performance insufficiencies come from being CPU bound i would say. An average and max fps bar is not really an optimal way to measure this btw, because as soon as you become CPU bottlenecked you will feel it, the frametimes become uneven and it will be stuttery. This is much more noticeable than a GPU dipping a bit, especially on a vrr monitor.
World of Warcraft and MS Flight Simulator are my most played games, both are almost always limited by the CPU, MSFS massively. Also Ray-tracing games like the Spider-Man games for example. And if not CPU limiting, the VRAM becomes an issue first, but the GPU clock speed is almost never the problem. @@capdave5
Since linux are getting better and better with the time thanks to valve and the steam deck I would love to see a comparison between linux vs windows and how it preformed on different GPUs
Agreed! The linux drivers for amd gpus in particular are very fast and often even faster than on windows. I recommend nobara for linux newbies and arch based distros for more advanced users. All of my games are running on linux. And hopefully valve pushes SteamOS for pc gamers soon as well. It is quite amazing how many possibilities one have to tune your rig.
One thing that was IMHO maybe missed was : -IF you plan on getting a GPU now and replacing the GPU in say .... 3 years but keeping the platform as it is, then an argument could be made for the stronger CPU IMHO but that's also IF you're perfectly happy having a little bit less fps right now..... And lets keep in mind that nobody has a crystal ball that would perfectly predict whether or not the 7800X3D will be obsolete in 3 years time but I find that possibility very very small IMHO AND also it is true that in theory there should be next generation of CPU possible for upgrades on this platform nevertheless I still think upgrading a GPU is just much easier than upgrading a CPU
In my opinion always get the best cpu you can, you won't have to upgrade it for a long time. And you can extract maximum performance from your gpu. Especially when you decide to upgrade in 2-3 years
I'd probably get the 7800X3D either way, as I feel like a lot people would upgrade their GPU before their CPU anyway, but I guess it really does depend on what games you play.
I have a 7950X3D + 7900XTX at 1440p 165Hz combo despite only playing single player and RTS. I need the cores for software compilation but opted for the premium over the 7950X "because why not", and I'm actually rather amazed in how many visually demanding games the XTX can still make the 7950X3D sweat. I'm using launch scripts to tie the games to the CCD they respond the best to (which still leaves 8 cores for background compilation, which is surprisingly stutter free) and yet still Cyberpunk with RT reflections set to highest, and everything else at highest raster settings, the GPU load is only 92-94% while the main game thread is pinning a core to 100%-flatline. Granted, RT in it's current iteration is fundamentally flawed, as having to "bother" the CPU for every single ray-hit is dumb as fuck, but still, that is some brutal CPU utilisation.
I personally don't care about playing at ultra settings even in single player games. High-medium looks good to me and I'll even turn on DLSS. However I do value having relatively high and consistent framerate (90 FPS 1% lows). So I would lean towards buying a better CPU
Very useful, would have loved to see the numbers of how the two higher and two lower compared to these listed numbers in the video as comparison (from previous video data).
Just a couple weeks later and we already need an updated "4070 Ti Super + 7600 or 4070 Super + 7800X3D" now that they're taking over the old model's price points. 😅
I have a 5800X and really thinking if I even need the 4070 Ti Super at this point I'm at 1440p, usually set my settings to high and am happy with 90-120 FPS max. Looking at these results, I won't even be close to taking advantage of what the 4070 Ti offers, so not even talking about the Super variant. And in Europe the cards will be hella expensive. Any opinions?
I like the premise of the video, but feel the best way to compare which combo is "best" would have the combos being compared against a baseline, which in this case would be a 7600 paired with a 4070. Personally though, I upgrade my GPU pretty much every generation and CPU platform less often, so if i was building from scratch and had to pick from either option it would be the 7800x3D combo. Right now as example, I have a 7900XTX paired with a 5800x3D.
This confirmed my purchase of the 7800x3d with the same ram speed and CL latency as you with the 4070. I would play on medium settings with most games anyway to hit that 144fps in which case the 4070 ti's lead would mostly fade away. I like competitive multiplayer games so I play Fortnite in performance mode which is entirely CPU limited... maybe even a 4060 ti would get the job done.
Very well done video and I appreciate the fact you created a video around a situation that most people do not necessarily think of. Hardware Unboxed FTW.
I am multiplayer gamer, so i would choose better CPU! Also it will stay with me for longer time then GPU Single player games are for days, multiplayer games are forever.
I blows my mind that so many people have a 7800x3d paired with crappy 4060 (all variants!)! I love these type of comparisons and thanks for all the hard work!
Average gamer: Alright I'll get the 7600 + 4070. The spare budget goes into beautiful case, RGB fans, liquid cooler and ROG tax! Oh wait I'm over budget... I'll just get the ROG 4060Ti instead. I mean... the computer is supposed to be looked at right? Who looks at their monitor anyway?
I would love to see 4070ti+7800X3d performance in order to see how much one could gain from just upgrading the Ryzen 7600. But great video! I really wanted to see this. I own a 6950XT (4070ti performance) and a Ryzen 7600 and I was wondering if getting the 7800x3d would benefit much at 3440x1440.
When Steve was giving his guesstimate on qanda it sounded plausible but it really is a matter of whether someone specifically plays 1080p competitive or just games in a general sense.
Thanks for doing the fortnite benchmark at medium settings! I just recently advised someone to do a 7800x3d and 4070 combo for high fps 1440p. Granted, where I live, there is a great bundle deal to get the motherboard, ram, and CPU for $500 at microcenter, so that made opting for the more expensive CPU a little easier.
A lot of people are gonna take away "GPU first is better" from this but I think the really interesting exploration was that if you are trying to push high frames for competitive games you want to focus on CPU throughput.
PCMR should already know this, but is a great informative video for new pc gamers dipping their toes in PC Gaming and efficiently allocating budgets to components. There is 1 strategy that destroys all this though, the Max-Max strategy someone who builds a 7800x3d with an RTX 4090 today will enjoy the best PC Gaming can offer for several years... ^_^
The better option is to buy the cheaper CPU and GPU as a combo. Money saved means you will have money to upgrade at some point, either combination here is gonna deliver similar performance longterm. So the logical thing is to buy 7600 and 4070. I believe you would get as much playtime as either combo by doing this, saving that bit of money for future upgrades is the smart way around it. Considering what i know and what is going with the industry right now, i would not invest in anything higher than a 4060 class card, due to TDP issues. Electricity bills are going up, need to fight back against that somehow, so GPUs that are like ~150W or less are the goto solution for me. That or IGPU in the future.
love these video, for future would love to see additional comparison for top cpu and gpu pairirng i.e 4070ti and 7800x3d in this case to see if perfomance is cpu or gpu limited
Thanks a lot Steve, might be good idea to skip zen5 than and get smth better than my 3080 10gb. Have 7600x atm, not very happy with it, not a single BIOS still works on my MAG B650 Tomahawk wifi board after 1 year passed, apart release bios and A10 version, rest black and blue screens on non EXPO Trident Z5 RGB 5600 CL 34. So was really tempted to change cpu and ram, but since gpu gives that much performance difference in 1440p, I have XG27AQM 270hz, might time to get 5070 or that 8800xt card.
![YoungBoy Never Broke Again - Sneaking [Official Video]](http://i.ytimg.com/vi/1ILUps2zDFU/mqdefault.jpg)
![YoungBoy Never Broke Again - Catch Me [Official Video]](http://i.ytimg.com/vi/KxwEXXFQMTY/mqdefault.jpg)
man I love the premise of this video, comparison of CPU+GPU combos (which are not the top end) are very interesting and useful for potential buyers
The cpu is the best cpu tho😅 7800x3d
@@tuhinfrags what about 7900x3d, 7950x3d, intel i7,i9 14xxx seriers
@@ivanpoberezhniuk9694 7800x3d still the best gaming cpu
@@ivanpoberezhniuk9694Those aren’t the best when it comes to gaming.
@@tuhinfragsTrue, but there are certainly more expensive CPUs.
Love to see my Q&A question turned into a whole video!
I agree with other people in this comment section in that this opens up a whole spectrum of comparisons related to "what if you spend more" or "what if you spend less" or "what's the cost per frame of these setups".
Well done!
Thanks for asking the question too! :D
I'm also one that's asked for this several times, but Steve thinks it's ridiculous. THIS is the info that's so rarely seen because all reviewers only use top end pairings to "remove bottlenecks" without realizing knowing WHERE those bottlenecks are is VERY relevant information to anyone building on a budget. Knowing what's the fastest with no budget is easy. Just buy the top part. I don't need to watch a review to tell me that.
I guess we're just supposed to _guess_ where those bottlenecks are when you're building a realistic system or wondering if you need to replace your 12600k if you're thinking of buying a 4080 super.
Skewing as much of your budget toward the GPU as you can get away with has been common wisdom in the PC building community for a long time now, but it's good to see periodic testing to prove the point anyway.
For gaming then yes but when building a pc used for productivity it really depends on the type of program you’re using.
@@BakedPotato05 and if you are doing lots of newish console emulation some emulators love fast cpus compared to gpus too
A good cpu is going to age better in your system than a good gpu. It's easier to turn down settings and still get good performance where if you're cpu bound you're pretty much screwed. And it's easier to upgrade the gpu later.
I think it's a mistake, the GPU is the easier component to upgrade. With a worse CPU, when in 3 years you'll get your new 2-gen-newer GPU you'll have some problems.
Wrong. The most expensive component should be your monitor.
This is one of the most useful videos i've seen in a long time. Congratulations! I wish you would do these kind of comparisons with the most common GPU+CPU combos.
i play graphics intensive singleplayer games at 1080p, should i opt for the 4070ti combo or the 4070combo?
@@Schluggaluggas you should opt for a better monitor first, it's crazy how much better 1440p is than 1080p
@jakacresnar5855 I originally wanted to do that, but I want to play a game called ark Ascended, which is a remake of Ark Survival Evolved. It's probably the most beautiful game out there IMO, but it's very unoptimized. I am talking 4070super 40fps native 1080p unoptimized. It's enjoyable with frame gen, but If I play at 1440p my fps will even with frame Gen constantly be
I noticed that the gap between the average and 1%lows are much narrower with the 7800X3D.
That's the main advantage of the 3D cache. Even if there is no improvement in average FPS, there is a great increase in 1% lows. This is true even at 1440p and 4K.
That was also a noticable thing when 5800x3D was released.
x3D's also helping a lot with truly bad coded games.
On jedi survivor, truly terrible coding on this game, at least with 5800x3D I got much less micro stutter than my friend with an i7 13700, while we both have a 3080 graphics card and game installed on a gen2 NVME. And this although benchmarks show similar results on 1% lows, the experience is smoother on 5800x3D.
Yeah, 1% low is way more important that average fps cuz high fps is nothing if the game stuttering like crazy
You could just save your money and cap FPS with RTSS
Already noted with the 5800x3d.
Could have been relevant to include one or two example results of the 7600+4070 and 7800x3d+4070ti for reference to see what the additional savings vs spending would also do here. Might be relevant especially in the cases where the result was veered to more gpu or more cpu limited.
A goo improvement, to an already good comparison.
Was honestly expecting this exact comparison as well.
GPU pricing may change to the worse or the better depend on supply & demand, some political move also play a big part of it.
So it's better to get as high GPU while you can. CPU pricing will always go down so you can upgrade later.
@@initialds4903100% true gpu prices change al the time but cpu usually go down
I also was hoping that they included this. Would be a great comparison point
With AM5 I'd be tempted to start with the 7600+4070Ti then drop in a '9800X3D' in a few years.
By that point a 5070ti or whatever will be out that matches a 4090. There's an argument to be made for going 7800x3d + 4070 super. The upgrade 4070 super to 5070ti or whatever after 2 yrs. With how upgradable AM5 is, the best value would be 7600 to 9600 and 4070 super to 5070ti(assuming it comes close to a 4090)
That's what I did with AM4 - ditched my i5 4690k/Z97-I Plus (Asus mobo) when Ryzen 3600 had been out about 6 months - yes I kept that older system for years as it played my games just fine & all new Intel chips were only slight improvement - So I bought Asus ROG STRIX B450-I Gaming mobo/Ryzen 3600 & moved my trusty GTX 1070 (EVGA with thermal pad mod) over to the new system, upping the GPU to a 1080TI a few months later as it was great price off of eBay - did another upgrade a year later to 5900X/RTX 3080TI - these kinds of upgrades were unheard of with Intel platforms 😮 I did completely disregard Hardware Unboxed review on this Motherboard around VRM heating issues - I'm running 12 core 24 Thread CPU & a power hungry GPU (350+w) & this board is completely fantastic 🤩 - I do have the GPU on a riser so it's not radiating heat to the mobo (modded Gamemax Brufen C1 case). I will definitely be moving to AM5 but unsure about A620 or B650 as I tend not to overclock anymore (CPU/GPU are fast enough at stock for me - only Benches I overclock).
@@RadialSeeker113 RTX 5000 series should be out by end of year?
@@warsshanonly if Nvidia follows it's previous cadence. But there was a leaked Nvidia roadmap in 2023 that showed they planned on releasing 5000 series in 2025. And rumors late 2023 said Nvidia was gearing up and would be ready to respond if AMD released new video cards in 2024
That was my strategy, except I rolled over the 3080 from my previous platform. Well and truly still good enough at 1440p for what I mostly play. Significant FPS and particularly 1% low jumps from the 11400f/DDR4 platform
I would have liked to see results for at least one or two popular strategy/city-builder games, like Anno 1800 or City Skylines, since these games are really demanding on both CPU and GPU.
Same. I actually thought they were going to, since they specifically called city builders out in the beginning. It would be super helpful to have tons of games ranked by how GPU vs CPU bound they are.
Baldur's Gate 3 is another one I've seen called out as surprisingly CPU bound. MSFS as another.
I think he compared regular 5800x and 5800x3d, in cities skylines he show about like 40% maybe ,
Yeah totally missed the chance to include games that use the 3D cache. Cities Skyline, Anno, Factoro, World of Warcraft etc..
@@Chilith yeah, cuz i mostly play cpu bound singleplayer games , , i willing to combine ryzen 7 7800x3d with rx 6600 non xt which is very stupid , unless you play mostly cpu bound games whether its singleplayer or multiplayer
Even the 7500F is so good, it is probably even more in the "sweet spot" regarding value for money. Gonna build a system around it soon :)
I ordered from Aliexpress, will make a video soon.
7500F is very similar to 7600. 7500F does have the same cache. Only a very small difference in clock speed and no iGPU.
If you can get one where you are it certainly makes sense to really follow the min max strategy. 7600 prices are too high to really consider it a min, but I don't think the price gap between 7500F and 7600 would let you buy higher up the Nvidia stack anyway. Might let you get a 7900XT though.
@@AndyViant Hey I'm just pairing it with a 6700XT, might upgrade to 6800 later.
Yes but if people still have R5 5600 the upgrade should be overthought.
7500f and 7600 are only considerably faster in 1080. if u have 4070 ti that makes no sense .
7:30 - Lovely ending.
Thanks for these results. I needed to see them.
7600 is like 5800x3d so definitely good enough for 4070 ti
New 5800X3D systems no longer make sense because the 7600 is so good.
@@wayland7150 I can still see the 5800x3D pull ahead in the future when the L3 cache starts to matter..mostly in badly optimized titles
@@Greenalex89 Yes that's true which is why the 5800X3D is good for an upgrade. However it's a bad idea for a new system due to the price and lack of upgrade path.
It has 2 less cores
@@powtmow It clocks a bit faster and the AM5 systems have higher frame rates than AM4 generally.
Great thanks for making this. I’ve been expecting it since it came up in the Q and A.
I would've appreciated it more, if you would've included the combinations of 7600 + RTX 4070 and 7800X3D + RTX 4070 Ti in this comparison too. I think is could've shown better the CPU or GPU "bottlenecks" in some titles.
This has been super illuminating, I'd love to see more comparisons like this moving forward as they directly address our real-world decision processes.
That's why I always respected you, you listen to your audience, to average Joe dilemmas. Thank you!
I have a R5 3600 with a GTX 1060. I plan to upgrade to a 4070 Super when it releases💀But I am also planning to upgrade to a R7 5800X3D not too long after. My CPU still feels fine for now. It is definitely my GPU that is holding me back from enjoying modern games
5700x3d might be an option too at that point
i had a r5 3600 with a 1660super, recently upgraded the gpu to a rtx 4070 still with the same r5 3600. works okay as the most demanding games are usually gpu intensive, but a cpu upgrade would definitely be worth it if i can get a good deal somewhere
@@pistolenbob9876you must be running 1440p . Does it stutter at 1080p due to the ryzen 5 3600
@@FATBOY_. most games it doesnt, i upgraded to 1440p about a month after i got my rtx4070 so i played on 1080p with that combo for a while and it was mostly fine. honestly its kind of hard to say with the amount of shitty optimised games this year. it'll definitely be worth it to upgrade my cpu at some point but its not a must and my experience now is great already
How much RAM do you have now?
Amazing work Steve! Just as viewers have needs, Steve delivers, as always!
I’m thinking of pairing the 7800X3D with a 4070 super, after some time upgrade the gpu to the new generation. This way because the 7800X3D is a powerful chip It probably won’t bottleneck the future generation of GPU’s and you’ll get 5+ years of usage. Hopefully 🙏🏼
Have 4070ti and 5800x before the gpu. Can't wait for 8800X3D.
Finally, thank you guys, I was thinking the same since the Q&A question and trying to find relevant testings with little success. I would really love to see these kind of comparisons in the future, maybe cross test them as well so both GPUs with both CPUs. I know that a RTX4090 will get you the best CPU results when testing but for some the 4070 and the Ti is more than we can afford and if someone is planning to spend this kind of money I think tests like these are perfect to help us get the best for the price. Cheers!
I'd like to see HUB branch out into doing more hardware reviews which I think is the wheelhouse of this channel. Maybe cases? When there are no new cpu's, gpu's and motherboards to review it seems like there is just the Q&A's with Tim and Steve.
That would be a good way to kill the channel :D Why do you think we moved monitors to its own channel?
People don't get into cases that much. They're usually GN's lower viewed videos. They matter, but they're not exactly exciting. I'd rather see more of this mid range pairing when new releases slow down. Judging by these comments I'm not alone.
Yeah, I just used cases as an example. I love this channel but it seems like the ratio is too highly tilted towards the Q&A stuff lately. I rather see reviews, benchmarks etc...@@zodwraith5745
I just think there are too many Q&A's lately. Steve, your reviews and benchmarks are the best. That is what I enjoy. I can get podcast type stuff elsewhere. I mean all this in a constructive way. Not as any kind of criticism. @@Hardwareunboxed
I would very much like a video about long term choices when picking pc parts. This always has been in the back of my head for a long while.
It's easy to pick today how to buy pc parts, you buy what's best for your money. I've been building PCs for a long while and nowadays I sometimes don't necessarily buy the best bang for buck at the time of purchase because I think about the long term cost.
I'll use a concrete example; way back I had to pick from a i7 4770k or the i5 equivalent for haswell. At the time I went with the i7 even if it didn't really made sense in gaming benchmarks. I've had this cpu for 7 years or so and I'm pretty sure it lasted one GPU upgrade longer because I had hyper threading. So the 100$ difference saved me from buying new ram, mobo and cpu for a few more years.
Same thing with the power supplies, I've had many friends upgrade their 550-650w power supply, not because they aren't working, but because a modern intel gpu and 3000 series gpu was pulling too much for the wall. A decent power supply can last a decade, might as well invest a bit more so you aren't forced to upgrade while just halfway through it's life span. I'm very happy with my overkill 850w power supply at the time, it's a part I can keep for a long time.
Another example, I still have a nhd15 that I got in 2015, so many of my friends cheaped out and had to upgrade many cheap aios or air coolers I'll still use this cooler in a decade probably.. worth the 50$ premium I paid for vs buying a cheaper cooler and having it be replaced because of the lack of forward compatibility.
Picking a cpu has quite a bit of implications, because you are also picking a platform. In this example, you've got am5 first generation, meaning that there's an upgrade path without buying a new motherboard and ram in a few years, I wouldn't mind cheaping out in the CPU department knowing I could get a much better CPU latter. If I was picking up an am4 cpu right now, I would get a higher end model knowing there's no cheap upgrade path without changing ram/mobo, might as well get something that might last me one more GPU upgrade.
It's hard to reason about it because you also have to consider the selling price, but most of the time the obvious option might not be the right pick if you look at from a longer timeframe.
Well, a CPU upgrade later on is more affordable, even if you can "only" go 7800X3D, not 8000/9000/10000...
Excellent video. This pretty much exemplifies _why_ I've always argued that reviews of mid grade parts need to be paired with realistic counterparts. Or at the very least remember to throw in 1-2 benchmarks to remind people this. Most often your reviews _DON'T_ remind people with realistic budgets that the min/max of CPU over GPU is usually the most beneficial.
Having the 4090 have such a huge performance gap to the rest of the pack just amplifies this problem further. It's easy to see where someone with mediocre knowledge of building a PC would see the massive difference in the 7600 and 7800x3d when paired to a 4090 compared to the much smaller differences between a 4070 and 4070ti paired with a 7800x3d, then incorrectly assume "Well, obviously I'll see more gains with the faster CPU at the same price."
And this is when the 4070 and 4070ti are still relatively powerful GPUs for what many people can afford these days helping the x3d look even better. The problem gets compounded even worse as you move into lower tiered GPUs which would be quite likely with current prices. Do the same comparison with a 4060 and 4060ti that many people buy and you'd rarely be CPU bound even in competitive shooters. This would show in most games that CPU is completely irrelevant when paired with the most popular GPUs and using settings above 1080 low, as long as you're at least using a modern CPU.
While I understand the _theory_ that you need to remove any bottleneck to amplify differences in parts being reviewed, that's not how real people build real systems unless they're building a $3500+ system. This is why I think it would be beneficial to show at least 1-2 "don't overspend on the wrong shit" benchmarks to show real world pairings. It's insane the idiots I've built systems for that insist on an overpriced X3D because they saw the reviews just to pair it with a 3060 or 4060.
In reality the opposite is true, you should always remove bottlenecks when testing a component, you want to test that component in isolation as best you can. Reviews aren't designed to convince you to upgrade, if that's how you view them you're doing it wrong. Reviews are there for when you've decided it's time to upgrade and they should show you exactly how competing parts really compare.
If there are two CPUs that cost $200, why would you want them compared with a low-end GPU? Those GPU bound results are only testing the GPU and tell you nothing about how the CPUs compare, one CPU could be 40% faster than the other, but you'd never know.
Any heavily GPU bound CPU testing is extremely misleading, as is any CPU bound GPU testing.
@@Hardwareunboxed Which would be exactly _why_ I said include 1 or 2 benches with real world pairings to remind people to build balanced systems, not omit bottleneck removing pairings all together.
I was agreeing with the point of the video all along, Steve. You of all people should know just by some of the BS you see in comments that people need to be reminded more often about building balanced systems. Especially if they're watching as a buying guide and not just us enthusiast fanboys.
You're absolutely correct that not removing bottlenecks is misleading when reviewing parts, but _not_ showing where those bottlenecks will occur is just as misleading when someone is trying to spec out an _entire_ system. Not just the individual part.
That's why this video was more valuable than you think. There's a lot of people out there that don't have the understanding of the basics like you and I do. And they sure as hell haven't gone through as much hardware as we have and just automatically know where those bottlenecks occur.
You're asking for the impossible, as this video has just demonstrated. Who are you to determine what 'useful bottlenecked' data people should be presented with? What does that even look like? What hardware are we using? What games are we testing? What are the quality settings and resolutions?
You are working under the assumption that all gamers play the same games, the same way, when in reality that's not the case at all.
Of course we can test a 7800X3D and a range of other CPUs with a low-end GPU and show them all rendering 60 fps, how is that useful? I assume you've watched this video: ruclips.net/video/Zy3w-VZyoiM/видео.html
@@Hardwareunboxed _You're_ working under the assumption that I'm saying all your testing needs to change, which I never did, and I haven't said anything that was incorrect. I simply stated it would be nice to get a graph once in a while to show _why_ you build balanced systems, because a lot of people don't seem to get that. Which you yourself HAVE done a video like that in the past and I praised you for it. Ironically your theoretical test at the end WOULD be useful because it tells people if you can't afford a 4 figure GPU don't waste your money on an overpriced CPU. You yourself constantly recommend the i5s and R5s.
You think this video is ridiculous but from my point of view it carries a message that is rarely heard: Don't buy more hardware than you need. You just PROVED that by showing the second you're not playing at 1080p in competitive shooters that GPU is more important. You chose 4 very relevant parts and proved WHY "it depends" is on your shirt.
You keep getting these questions _because_ it's rare anyone gets to see mid range pairings. We know you can't test every configuration out there, but don't attack the viewer for wanting relevant information for realistic builds, when you provide more than enough info that's irrelevant to anyone building within a budget.
Almost no one asks me to build them a "money doesn't matter" system anymore because it's gotten horrifically expensive, but TOO many people insist on an X3D so they can pair it with a 3060. That's a problem and at least now I can link them this video.
Would be funny that RTX 4070 Super will release later and pairing with the 7800X3D will be the perfect 3rd option/answer in this comparison. If 4070 Super is indeed close to 4070 Ti performance.
Why not add in the 7600 + RTX 4070 Ti Super while we are at it and if is indeed close to a RTX 4080 performance we are back into this video again.
The new title would have a “Super Edition” perhaps
This is the best type of review for me. paring the CPU with Right GPU has always been my problem. Great video!
missed opportunity to also test nVidia CPU overhead by getting a 7900xt + 7600x in the mix
And 7800XT + 7800X3D
@@BusaNZ1340 that's a config for competitive gaming
This is a great premise for a video series. Please keep doing this!
This was a great video. I actually just assumed almost nothing is CPU limited anymore (and functionally yeah above a certain fps it's just academic) but I was shocked that the 7800x3d4070 system won anything. I have a ryzen3600/rx6800 and I wonder sometimes what would be the best upgrade path from here, and since I do play VR maybe the additional frames would make something like a 5600x3D a worthwhile purchase.
That's sounding like a good plan getting the 5600x3d. I went from a 1800x/ Vega 64 to 5800x/RX6800 and its a really nice 1440p combo.
Of course the ultimate upgrade right now for AM4 would be the 5800x3d if you have the cash.
New CPU won't help VR. There's still something with Radeon that causes the 1% low issues. AMD admits it but haven't fixed it for years now. I can't use my AMD system for VR so it stays on one of the Nvidia systems.
As I said in the community post. It 100% depends on the games. AAA is better with the beefier GPU but there are plenty of other styles of game where the CPU will matter more. ARPGs, especially late game like Path of Exile get very CPU bound, simulation titles like ACC, iRacing, Flight Sim, Cities Skylines can be very CPU bound for both frame rates and the games tic rate. Strategy titles can also be very CPU dependent for tic rate and simulation speed. Then there are MMOs which also get CPU limited in large fights where you actually want to maintain a reasonable fps.
So while this is a pretty good video the conclusion was predictable purely based on the titles tested and to do this kind of test justice really needs a far broader suite of games that cover all the bases so people can actually make use of it to inform buying decisions.
Will ryzen 7700 be enough for PoE late game or do I need to upgrade to the x3d version?
Would be really neat to see how each combo holds up a few years from now.
This is a really good video. You should do this more often to help people understand how old cpu's work with new gpu's and whether or not they need to upgrade CPU or GPU first!
A good review would be, test which cpu is needed for example a 4070 and test with (8700, 9900, 12400, 12700, etc.)
@Eternalduoe just suggested that too. At what CPU does the frame rate stop getting better. Couple that with asking at what CPU is it better to upgrade the GPU instead. This would result in the most perfect balance between CPU and GPU.
This is exactly why I bought a 4070Ti and 7600. At the time, it was the best value and will hopefully be an upgradable platform for a long time. Not thrilled about the price of the 4070Ti, but that's a different story for a different time. I consistently get 5.3GHz on the CPU with the BeQuiet Dark Rock TFs and my system is whisper quiet. No regerts.
7600 and 7900xt for sure
I recently put together a 4070 and 7800 X3D build and I'm pretty happy with it. I had initially got the card first for my older system but decided to update my whole system and switch from Intel to AMD as it was aging. It's nice that I can keep the powerful CPU and always upgrade to a better GPU in the future.
Am planning to upgrade to the same specs soon as well, currently rocking an old i7-4790K + RTX2060.
I'm working as a video editor, and have finally save up enough for an upgrade :')
And since I'll probably do 60% work/productivity and 40% game on my pc, I hope I won't regret with this decision 🙏
I feel like the 7800x3d combo will have more utility once a single GPU upgrade comes into the equation (in the event that the entire system isn't done at the same time).
I feel like the 4070ti combo will have more utility once a single CPU upgrade comes into the equation :)
@@Grena567Exactly, you can actually afford CPU upgrades.
@@Grena567 I think typically speaking your average user is less likely to perform a CPU upgrade than a GPU upgrade, even with the longer tooth support we enjoy with current AMD platforms.
Depends on your resolution, obviously.
I like these types of scaling videos. It really helps with real world combo choices
I would take the 4070 Ti + 7600 combo every time in these 2 choices.
Yeah. That 35% advantage in Avatar and 27% in SWJS at 1440p is pretty impressive. 1440p gaming at Max settings is clearly very GPU demanding. And 4070 is not powerful enough for stable 60 fps experience.
When I moved apartments recently, I decided to save space and consolidate my two PCs into one SFF pc using the latest parts from each build. I saw basically no change in performance going from a i7-11700KF + 3080 12GB to a i3-12100F + 3080 12GB. I ended up switching to a 12600K on sale, but I'm not even sure it was worth it since that 12100F was such a little powerhouse.
Personality I would rather be in a gpu than a cpu bottlneck, i can always reduce gpu heavy settings but there a few to none cpu depending settings.
That was great, Steve. This is useful info for determining where to spend limited funds. Stir into the mix a 7700 and the new 4070 Super for those seeking a middle ground. In the same ballpark cost-wise.
If you play MMO's you are going to want to go with the 7800X3D option. The cpu power and cache make a significant difference.
Same with simracing. All the sims are very cpu heavy. Good example is ACC. And when you add a lot of AI's your fps tanks in rF2 and AMS2 too.
I watched my i7-12700K destroying the Ryzen 7 5800X3D for emulation by 51%.
It chills at very low CPU usage for anything minus any console emulators and UT2004 (singlethread bottleneck, barely more than 300fps).
More performance now: skimp on CPU (although here skimping is a big word, the 7600 is a great CPU)
More performance over the lifespan of the platform, assuming 2 or 3 years upgrade cycle: better CPU, you swap the 4070 for a 5070 or 6070 and you can keep the rest of the platform untouched.
Still I "built" 3 PCs so far and every single one of them picked the better CPU than GPU option simply because the 3 people using them (me included) use the pc to also do very CPU intensive work (e.g. python and AI classes), so the 5800X3D for me and yhe 7800X3D for thebother 2 was basically a necessity (I got a 3060, the other 2 a 4070 each)
I really appreciate this type of video! One thing i would love to see explored is where the ceiling in GPU performance for a particular CPU is.
E.g. For a 7600, which GPU generally maxes-out the performance of that chip?
I envisage it as 4070/3080 Ti/4070 Ti/7800 XT/ 7900 XT comparison. Or something similar...
You can get an idea of this by looking at the different frame rates per resolution. You can see that generally the X3D system wins at lower resolutions and the Ti system wins at higher. Also the 1% lows are better with the X3D.
Something is going to be the bottleneck. I had a chap with a GTX 1060 and a gen 4 i5 at a time when new GPUs were super expensive. He wanted a new GPU so I upgraded him to a RYZEN 5 3600 whilst still using his original GPU. He was delighted with the improved frame rate and smoothness. His old CPU was not driving his GPU hard enough.
There is not so much of a performance ceiling but a more effective way to spend your money.
I just bought a 4070 super fe and was planning to buy ryzen 7600 to go along with, then came across this video! Helped quite a bit, thank you very much..
You guys should really add benchmarks of games that are *actually* CPU bound to these videos. Grand Strategy games like Hoi4, Vic3, Civ6, etc. would be really useful to also have in benchmarks of these types.
These games are CPU heavy but you don't need 300 FPS like in CS2. 60 FPS is 100% ok. So in reality these games are very little CPU limited. 60 FPS can achieve even the Ryzen 5 5600.
@@JackJohnson-br4qr The limiting aspect of these games is tic rate and turn time which is 100% on the CPU.
Microsoft flight simulator. DCS. X4 foundations would also want the 7800x3d
@@andrewcross5918 Which means what? Even fast pace competitive shooters like Fortnite or COD have only 30Hz tick rate. Even the Ryzen 5 3600 can achieve that.
If you look on paradox forums you'll find benchmark threads for their games. I recently looked at the victoria 3 benchmark thread and late game the 7600x is about 50% slower than a 7800x3d ( about 3 mins to simulate a year in 1900 with the 7600x versus 2 mins on the 7800x3d ).
these are the videos people need. You might want to include the results of the best combo as well. In this case 7600 + 4070 ti // 7800x3d + 4070 // 7800x3d + 4070ti. That way people can see how much they are actually compromising and it gives them a path to follow in their choice between bang for buck or ultimate performance.
Now I would have liked the 5800x3d thrown in as a bonus, paired with the 4070Ti. I hope you guys will be able take a look at the 5700x3D now that it has been announced officially.
@noir-13 5700x3d at 250
7600 is 5800x3d performance for $150 less
5800X3D is a severely overpriced CPU. And even 5700X3D is pretty bad value as well.
@@Boopity7739 Depends on the region, in EU the 7600 is 240€ while the 5800x3d is 320€
@@GLDragon93 Valid point actually. Plus AM5 and DDR5 is more expensive, though the 80€ saved on a 7600 could probably cover the cost difference and even it out (unless you're already on AM4)
Expected outcome but nice to see it tested out with data. Thanks for the video 👍🏼
I'd go for the better cpu weaker gpu to start with. You change your GPU more often so would benefit next time you upgrade your GPU
imo that used to be the case - but until the gpu manufacturers stop pricing in the fps gains you get as extra cost, I doubt this rings as true as it used to
Both are AM5 platform
Honestly gpu upgrade is way easier to upgrade when it comes to installation proces than cpu
The GPU market is dogsh*t. I can't find any good price perfomance pairing for my i7-12700K.
A friend had luck and he got the 3080Ti for 700USD (the 4070Ti sucks, 800USD for 192 bit memory bus).
amazing test, was looking for exactly this and was not disappointed!
Glad to see my gut feeling comfirmed!
R5 7600 + 4070Ti for gamers.
This one for gaming.
4070 ti is a terrible choice, probably the worst GPU in this generation after 4060 ti. 12 gb will not cut it for 1440p in a few years.
@@beachslap7359 tbh every GPU of this generation suck
hopefully this guy at least waited a couple of weeks now that the super series has been announced
@@beachslap7359agree. I think this was a rare dud for not covering that angle. Both cards will age similarly while I expect the CPU will age well. Better to put that money towards the CPU now and have the tidier upgrade path.
My 7800x3d and 4070 was surprisingly amazing for 4k. I got it with only 1440 at first and decided to push 4k, of course I still plan to go for 4080 super soon because as amazing as it is, it is still struggling at times.
I use a 4070ti + 5800x3d and most performance insufficiencies come from being CPU bound i would say.
An average and max fps bar is not really an optimal way to measure this btw, because as soon as you become CPU bottlenecked you will feel it, the frametimes become uneven and it will be stuttery. This is much more noticeable than a GPU dipping a bit, especially on a vrr monitor.
I hate shader compilation stutter, that's why i have a i7-12700K, not a Ryzen 7.
where do you notice your cpu bottleneck?
World of Warcraft and MS Flight Simulator are my most played games, both are almost always limited by the CPU, MSFS massively. Also Ray-tracing games like the Spider-Man games for example.
And if not CPU limiting, the VRAM becomes an issue first, but the GPU clock speed is almost never the problem.
@@capdave5
I think you switched the CPUs for the 4070 + 7800X3D vs 4070TI + 7600 in the corner title at 0:11
Since linux are getting better and better with the time thanks to valve and the steam deck
I would love to see a comparison between linux vs windows and how it preformed on different GPUs
Linux is never faster in games
Agreed! The linux drivers for amd gpus in particular are very fast and often even faster than on windows. I recommend nobara for linux newbies and arch based distros for more advanced users. All of my games are running on linux. And hopefully valve pushes SteamOS for pc gamers soon as well. It is quite amazing how many possibilities one have to tune your rig.
@@GewelReal Wrong.
@@notjustforhackers4252 I never played a game on Linux that ran better than on Windows.
Productivity apps on the other hand stomp Windows on Linux
One thing that was IMHO maybe missed was :
-IF you plan on getting a GPU now and replacing the GPU in say .... 3 years but keeping the platform as it is, then an argument could be made for the stronger CPU IMHO
but that's also IF you're perfectly happy having a little bit less fps right now.....
And lets keep in mind that nobody has a crystal ball that would perfectly predict whether or not the 7800X3D will be obsolete in 3 years time but I find that possibility very very small IMHO
AND also it is true that in theory there should be next generation of CPU possible for upgrades on this platform
nevertheless I still think upgrading a GPU is just much easier than upgrading a CPU
In my opinion always get the best cpu you can, you won't have to upgrade it for a long time. And you can extract maximum performance from your gpu. Especially when you decide to upgrade in 2-3 years
Thought this same concept till now cuz not if you play various games especially gpu bound
This is a great comparison, need some comparison for online games tho, MMO like when we raid with 40+ ppl on screen
I'd probably get the 7800X3D either way, as I feel like a lot people would upgrade their GPU before their CPU anyway, but I guess it really does depend on what games you play.
Exactly I upgrade my cpu every 5 years. GPU is every 2-3 years
I have a 7950X3D + 7900XTX at 1440p 165Hz combo despite only playing single player and RTS. I need the cores for software compilation but opted for the premium over the 7950X "because why not", and I'm actually rather amazed in how many visually demanding games the XTX can still make the 7950X3D sweat. I'm using launch scripts to tie the games to the CCD they respond the best to (which still leaves 8 cores for background compilation, which is surprisingly stutter free) and yet still Cyberpunk with RT reflections set to highest, and everything else at highest raster settings, the GPU load is only 92-94% while the main game thread is pinning a core to 100%-flatline. Granted, RT in it's current iteration is fundamentally flawed, as having to "bother" the CPU for every single ray-hit is dumb as fuck, but still, that is some brutal CPU utilisation.
Very very good point.
yeah imo its way easier to swap out a gpu than a cpu
@@Clashy69 Another good point, bravo sir.
I personally don't care about playing at ultra settings even in single player games. High-medium looks good to me and I'll even turn on DLSS. However I do value having relatively high and consistent framerate (90 FPS 1% lows). So I would lean towards buying a better CPU
7800X3D hands down
@@Paul-qk3wr cope
Very useful, would have loved to see the numbers of how the two higher and two lower compared to these listed numbers in the video as comparison (from previous video data).
Just a couple weeks later and we already need an updated "4070 Ti Super + 7600 or 4070 Super + 7800X3D" now that they're taking over the old model's price points. 😅
I liked the sample of the games chosen, very good coverage of a wide variety of players.
This is very good to know. Thought about getting a 7800x3d but now I know I dont need a high end cpu. Might settle for 5800x3d, 5600x3d, or 7700x.
My R5 7600 just arrived today can't wait to change it, I have an R7 2700x with 3070ti now
Your videos are awesome, it helped me a lot !!!
Thnx for Testing these Combos, so there is more details to the phrase "it depends".
I have a 5800X and really thinking if I even need the 4070 Ti Super at this point I'm at 1440p, usually set my settings to high and am happy with 90-120 FPS max. Looking at these results, I won't even be close to taking advantage of what the 4070 Ti offers, so not even talking about the Super variant. And in Europe the cards will be hella expensive. Any opinions?
Interesting video would like to see more stuff with different combos
I like the premise of the video, but feel the best way to compare which combo is "best" would have the combos being compared against a baseline, which in this case would be a 7600 paired with a 4070.
Personally though, I upgrade my GPU pretty much every generation and CPU platform less often, so if i was building from scratch and had to pick from either option it would be the 7800x3D combo. Right now as example, I have a 7900XTX paired with a 5800x3D.
Legit this review was helpful. I am about to build a sff 7600 4070super build. Purely for CS2 and travel. Thanks for sharing
This confirmed my purchase of the 7800x3d with the same ram speed and CL latency as you with the 4070. I would play on medium settings with most games anyway to hit that 144fps in which case the 4070 ti's lead would mostly fade away. I like competitive multiplayer games so I play Fortnite in performance mode which is entirely CPU limited... maybe even a 4060 ti would get the job done.
Very well done video and I appreciate the fact you created a video around a situation that most people do not necessarily think of. Hardware Unboxed FTW.
I am multiplayer gamer, so i would choose better CPU!
Also it will stay with me for longer time then GPU
Single player games are for days, multiplayer games are forever.
This is great to see! I'd love to see the upgrade path adding 7800X3D with the RTX4070Ti
You guys really need to add the finals!!!
really happy to see you do this test I hoped you would when the Q&A was done
Some really interesting results here. Thanks for sharing.
Thank you for using Basic/Medium settings. ❤
I blows my mind that so many people have a 7800x3d paired with crappy 4060 (all variants!)! I love these type of comparisons and thanks for all the hard work!
Thanks for giving us actual data on this!
Great vid especially for someone who doesn’t primarily play story games at 4k
Average gamer: Alright I'll get the 7600 + 4070. The spare budget goes into beautiful case, RGB fans, liquid cooler and ROG tax! Oh wait I'm over budget... I'll just get the ROG 4060Ti instead. I mean... the computer is supposed to be looked at right? Who looks at their monitor anyway?
I would love to see 4070ti+7800X3d performance in order to see how much one could gain from just upgrading the Ryzen 7600. But great video! I really wanted to see this. I own a 6950XT (4070ti performance) and a Ryzen 7600 and I was wondering if getting the 7800x3d would benefit much at 3440x1440.
The i7-12700K and Ryzen 7 5800X3D are faster than a Ryzen 5 7600.
i would be interested in this too
You should add Escape From Tarkov to these tests as it is very CPU intensive game and x3d cache makes huge difference there
When Steve was giving his guesstimate on qanda it sounded plausible but it really is a matter of whether someone specifically plays 1080p competitive or just games in a general sense.
Congrats on the million subs man ❤
I cut my budget from the AIO and SSD, and get the 4070ti instead of 4070 for the same price of full pc build.
Please do more of this comparsion thank you !
This seems like a popular question. Nice that they actually made a video about it.
Thanks for doing the fortnite benchmark at medium settings! I just recently advised someone to do a 7800x3d and 4070 combo for high fps 1440p. Granted, where I live, there is a great bundle deal to get the motherboard, ram, and CPU for $500 at microcenter, so that made opting for the more expensive CPU a little easier.
Great video! Next video like this, can you also add the cheap-cpu/cheap-gpu and expensive-cpu/expensive-gpu test results?
A lot of people are gonna take away "GPU first is better" from this but I think the really interesting exploration was that if you are trying to push high frames for competitive games you want to focus on CPU throughput.
PCMR should already know this, but is a great informative video for new pc gamers dipping their toes in PC Gaming and efficiently allocating budgets to components. There is 1 strategy that destroys all this though, the Max-Max strategy someone who builds a 7800x3d with an RTX 4090 today will enjoy the best PC Gaming can offer for several years... ^_^
The better option is to buy the cheaper CPU and GPU as a combo.
Money saved means you will have money to upgrade at some point, either combination here is gonna deliver similar performance longterm. So the logical thing is to buy 7600 and 4070. I believe you would get as much playtime as either combo by doing this, saving that bit of money for future upgrades is the smart way around it.
Considering what i know and what is going with the industry right now, i would not invest in anything higher than a 4060 class card, due to TDP issues. Electricity bills are going up, need to fight back against that somehow, so GPUs that are like ~150W or less are the goto solution for me. That or IGPU in the future.
love these video, for future would love to see additional comparison for top cpu and gpu pairirng i.e 4070ti and 7800x3d in this case to see if perfomance is cpu or gpu limited
Thanks a lot Steve, might be good idea to skip zen5 than and get smth better than my 3080 10gb. Have 7600x atm, not very happy with it, not a single BIOS still works on my MAG B650 Tomahawk wifi board after 1 year passed, apart release bios and A10 version, rest black and blue screens on non EXPO Trident Z5 RGB 5600 CL 34.
So was really tempted to change cpu and ram, but since gpu gives that much performance difference in 1440p, I have XG27AQM 270hz, might time to get 5070 or that 8800xt card.
@Hardwareunboxed U are going a great job keep it up but can you just get a same video with the super series cards.