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

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.

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".

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 noticed that the gap between the average and 1%lows are much narrower with the 7800X3D.

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.

With AM5 I'd be tempted to start with the 7600+4070Ti then drop in a '9800X3D' in a few years.

Great thanks for making this. I’ve been expecting it since it came up in the Q and A.

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.

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.

7:30 - Lovely ending. Thanks for these results. I needed to see them.

This is a great premise for a video series. Please keep doing this!

Amazing work Steve! Just as viewers have needs, Steve delivers, as always!

Would be really neat to see how each combo holds up a few years from now.

Interesting video would like to see more stuff with different combos

really happy to see you do this test I hoped you would when the Q&A was done

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 :)

missed opportunity to also test nVidia CPU overhead by getting a 7900xt + 7600x in the mix

Great video! Next video like this, can you also add the cheap-cpu/cheap-gpu and expensive-cpu/expensive-gpu test results?

Thnx for Testing these Combos, so there is more details to the phrase "it depends".

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.

7600 is like 5800x3d so definitely good enough for 4070 ti

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 like these types of scaling videos. It really helps with real world combo choices

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.

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

Expected outcome but nice to see it tested out with data. Thanks for the video 👍🏼

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).

That's why I always respected you, you listen to your audience, to average Joe dilemmas. Thank you!

Have 4070ti and 5800x before the gpu. Can't wait for 8800X3D.

amazing test, was looking for exactly this and was not disappointed! Glad to see my gut feeling comfirmed!

Some really interesting results here. Thanks for sharing.

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 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...

Congrats on the million subs man ❤

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'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

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.

Thanks for the tests!

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.

A good review would be, test which cpu is needed for example a 4070 and test with (8700, 9900, 12400, 12700, etc.)

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).

Actual Good info with practical use. Great work 👏

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.

7600 and 7900xt for sure

This is the best type of review for me. paring the CPU with Right GPU has always been my problem. Great video!

Thank you for this video, quite informative!

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.

I would take the 4070 Ti + 7600 combo every time in these 2 choices.

Please do more of this comparsion thank you !

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.

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.

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.

YESS, i was just looking for this video yesterday!!!

Hey steve, can you do a value check for gaming laptops since there is a wide price margin between different gpus & the cpus. It would be a great help to see which cpu & gpu combo is enough for 1080p,1440p gaming for now & a few years in future.

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.

R5 7600 + 4070Ti for gamers.

Thanks for giving us actual data on this!

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

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

7800X3D hands down

really really interesting. More of these kind of videos please

love the comparison idea and as ususal awesome presentation

There is one aspect completely missing here. You can scale GPU performance really well by changing resolution, using upscaling or lowering graphics settings. So if you are GPU limited you can almost always reach your target framerate just by adjusting some settings. Combining Upscaling + Graphics settings can make a game run twice as fast or even more. On the other hand if you are CPU limited there is nothing you can do. You can try lowering some graphics settings and hope that there is one setting that gives you some CPU performance back but usually there are no settings impacting CPU or atleast they make not nearly as much of a difference as you'd hope. Long story short: GPU performance is scalable, CPU performance is not. So if you are CPU bottlenecked and framerates are too low you are f***ed. Ok, if the game supports Frame Generation you could bypass the CPU limit and get twice the framerate anyways. But if not it's always better to have the faster CPU. Of course it depends on how much faster the more expensive CPU really is in CPU limited scenarios and how much worse the cheaper GPU is. I also agree that in general you want the GPU as fast as possible. But I just want you to keep in mind that GPU usage is literally completely adjustable via graphics menu. Most graphics settings have an impact on GPU exclusively and not on the CPU. And there is one more aspect. If you are GPU limited frametimes are fine just how they should be. But the moment you hit the CPU limit frametimes get much worse (not in every game but in most games). So basically you want to avoid CPU limits at all cost because it feels bad to play and when you hit the CPU limit while framerates are too low you usually just have to accept that you will get a bad experience because there is nothing you can do despite upgrading your CPU.

This video is such a sick concept!! Cannot wait to watch it

I think you switched the CPUs for the 4070 + 7800X3D vs 4070TI + 7600 in the corner title at 0:11

Thx 4 including CS2 ! 🙏🏼

That's very interesting. Your audience does need to understand that "it depends" isn't a cop out answer, but the only honest and accurate one you could provide. Great work as always!

This is a great video and extremely interesting. Thanks!

Nice test, good info!

This is great to see! I'd love to see the upgrade path adding 7800X3D with the RTX4070Ti

Would like to see a similar test of external gpu vs Strix Point APU hopefully with a 3d cache version on the CPU part. Not to compare to a specific external GPU card but to find out which GPUs at the low end are no longer required.

THIS should be a series!

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 as usual. Thank you.

These kind of videos are awesome, thank you

Legit this review was helpful. I am about to build a sff 7600 4070super build. Purely for CS2 and travel. Thanks for sharing

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.

You guys really need to add the finals!!!

This is the good stuff, Thanks Steve.

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.

This is a great comparison, need some comparison for online games tho, MMO like when we raid with 40+ ppl on screen

This is benchmarking!!! Well done Steve, good and interesting info 👍 Really useful info for those who want it. I think the GFX card is the way to go, especially with AMD CPUs, as CPU in socket upgrades are more attainable and cost effective for the future.

Just curious, with the bench showing the CPU's with the 4090 at 1080p, what memory was paired with the 11900k? Great video as always. I know my 11900k get's slapped around by newer cpus, just curious. I've got mine paired with tuned 32gb dual rank cl14-14-14 3733 and at 4k it's obviously holding up fine.

Keep up the good work !

Thank you for using Basic/Medium settings. ❤

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.

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.

Hey Thanks for doing this video! I would say do it with intel as well, but this has been helpful!

I had a feeling you'd bench this question. For the most part the theory comes right. it's probably a little closer than I'd have thought when going up the resolutions.

Nice video. Depends on game engine and optimizations.

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.

Do you have any GPU tests with MSFS 2020?

very interesting thanks for sharing, wondering how an old cpu would fair with a 4070 like I5 8600 or a ryzen 1600

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.

thank you so much , very useful . I always had this doubt . 🤩🤩🤩🤩

From Denmark, Isn't your title on the screen wrong right at the start.

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.

@Hardwareunboxed U are going a great job keep it up but can you just get a same video with the super series cards.

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.