News AMD Shares First Official Ryzen 7 7800X3D Gaming Benchmarks vs Core i9-13900K, Up To 24% Faster

[hotaru251]

is it best in every game?
no. (duh)

however in the games/tasks that benefit it is a significant improvement...and the ones its not? its what? 5% slower? i'd take 5% slower in some for up to 25% faster in some others. (on top of less pwr usage which impacts thermals & energy bill)

 

[PlaneInTheSky]

(on top of less pwr usage which impacts thermals & energy bill)

Not that I disagree, but the people buying $500 CPU probably couldn't care less about their energy bill.

It's like installing energy efficient light bulbs in a private jet.

 

[rluker5]

Is it me or are AMD CPUs depreciating faster and faster?
Apparently, according to AMD, if this is true, the 7800X3D is 6-9% faster in gaming than the 7950X3D. That's like the difference between the 13900k and the 13600k. And the 7950X3D costs a bunch more and falls behind the 7950X in productivity. There are probably people who either still have their freshly depreciated to 2nd best CPU still in the mail or still in the box waiting for watercooling parts to get installed. I would just return that scam.

Is there going to be a significant price cut on the 7950X3D and 7900X3D now?

Just like the 7000 series? Which if tested with these games was never as fast in gaming as the 5800X3D. Basically Alder Lake class.
Just like the 5000 series other than the 5800X3D? Can the 5000 series even hit 60fps on that outdated platform anymore? Just kidding on that last one. I know the 5000 series beat the Skylake arch.

Those AMD customers who waited for the good chip are being rewarded for their patience. Must sting for the ones who didn't.

The cache does help overall responsiveness and smoothness. At least it does with those old Intel edram ones. (Which did hit the wall with CP2077. Really fast in simpler games is one thing, but being unacceptably slow in one I really wanted to play with maxed CPU settings is what convinced me to move on. 5775c couldn't hold 60 fps in full crowd, non HDD mode high traffic locations. Just not enough IPC)

But low power< low temps for me. Mainly for fan noise. My CPU generally doesn't use as much power as my GPU or monitor so I don't fret about that.

 

[dalauder]

These chips are exactly why the 7000 series didn't sell initially. Who's gonna do a $1500 AMD build when they know this is coming?

 

[dalauder]

Actually, the best resolution to test CPUs at is 720p, even, since thwn the GPU doesn't limit at all and you can best see actual CPU raw performance. True, nobody buys them for that. However, in higher resolutions, the GPU limits the CPU to the point they all become the same. That is quite the moot comparison. It is important to keep in mind that on those resolutions, especially with GPUs that aren't the latest and greatest, it almost doesn't even matter what CPU you have. A 12400 will be the same as a 7950X3D, for example, or indistinguishably close, when paired with, say, a 3070Ti playing at high resolution. That's why I don't really buy the future-proofing argument, either. But people will be people, and no matter how much you say, some will simply refuse to listen. Still, low-res CPU tests make sense is the point here.

That's the ridiculous thing--I've got a 3600X and my 2080 Ti is over 100fps, on most titles--my CPU is fine. So long as you have a Ryzen 5/i5 from the last couple generations, it's not gonna limit you. Of course, it's more about bragging rights than usage for most people.

I guess there are plenty of people gaming at 4K though, in which case the CPU REALLY doesn't matter.

 

[CelicaGT]

Purchase the CPU that gives the best performance for the games and/or applications you run. Anyone shilling (paid or otherwise) for either side? Pathetic. Go look in the mirror and question your life goals. Neither corporation deserves such slavering loyalty. It's fine to be a fan of a company/product but good grief folks don't define your identity by it.

 

[TerryLaze]

LOL, Really ?
Surely you're smart enough that a typo doesn't stop you from understanding the message, just as you should be able to understand why benchmarks are a complex thing which makes disclaimers necessary.

And I'm sure that you are smart enough to understand sarcasm, Drax.
Yes, disclaimers are always necessary, and they always mean a specific thing, and so does 'up to' when it's right below the title on every single bench.

 

[The Historical Fidelity]

Lol, did they actually not even know what game they are testing or is there really a horizon zero down?!
Also having the 'up to' disclaimer on every pic probably means that they took the max FPS and not the avg.

Oh you don’t know about Horizon Zero Down? It’s that new Bank Loan simulator where the objective is to finagle your customers accounting figures to unlock Zero Down financing. I’ve almost unlocked it but it always seems just over the horizon lol. Sorry about this…

 

[The Historical Fidelity]

I checked if there is bias in AMD's latest slide where they hand-pick games (above).

I checked how the 5800X3D historically performed in these AMD-handpicked titles in reviews.

There is a clear bias in this slide, all of the games AMD picked, are games where AMD performed much better than Intel with X3D.

There is an abnormal level of AMD bias in the games AMD puts in their slides.

It’s not bias, it’s marketing. The company’s decision not to use games that would hurt their products perceived viability is SOP. Besides that’s what tech reviewers are for.

You don’t see, for example, Valvoline advertise their motor oil as “pretty much the same performance as our competitors, but our logo looks different, buy our oil”

 

[DavidLejdar]

Not that I disagree, but the people buying $500 CPU probably couldn't care less about their energy bill.

It's like installing energy efficient light bulbs in a private jet.

Some of us can consider a $500 CPU because we keep our monthly bills low. In example, a mobile phone tariff with data usage can come to around 30 Euro here. And not using that, that's more than 300 Euro a year I don't spend on it. So, assuming that MB, RAM, and CPU will last me at least 3 years, I can pay for it just by missing out on being online all the time.

50W difference in power consumption doesn't add up as much, but still adds up. Whether one really needs such a CPU, respectively how much one is willing to spend on their hobby, that's a bit different question. But the point is that when two evenly priced CPUs are on offer, with around the same performance, then the one with lower power consumption results in a financial saving which can pay e.g. for the CPU cooling (counted over a 2-3 year period).

Not easily applicable for everyone of course. E.g. someone living in London, where rent is up to $2,000 for an one-bed flat, and with a zero-hour work contract, such person may not have any spare money to begin with. But just meant to point out that one doesn't need to be rich or starving to be able to afford something nice now and then.

 

[ottonis]

I'm sure they do. For me it is not about either company.

Promotional slides with hand-picked games need to stop and companies need to be called out for them, since they are manipulative.

And the "up to X% faster" bs from AMD has to stop too.

And reviews need to pick a wider variety of games to do the testing. Including games that don't have built-in benches, it's not hard to run an input script and test a game without a built-in bench.

While there is some argument to be made that including wildly popular games in a bench has some community benefit, there is no reason why every AMD review has to have ridiculous anomalies like F1 in them that completely skew results.

Companies are free to post benchmark results from any games they see fit.
Nobody is going to base their decion ton buying a CPU based only on the advertorial slides from the manufacturer but will rather wait until independent tests and reviews become available.

 

[KyaraM]

Is it me or are AMD CPUs depreciating faster and faster?
Apparently, according to AMD, if this is true, the 7800X3D is 6-9% faster in gaming than the 7950X3D. That's like the difference between the 13900k and the 13600k. And the 7950X3D costs a bunch more and falls behind the 7950X in productivity. There are probably people who either still have their freshly depreciated to 2nd best CPU still in the mail or still in the box waiting for watercooling parts to get installed. I would just return that scam.

Is there going to be a significant price cut on the 7950X3D and 7900X3D now?

Just like the 7000 series? Which if tested with these games was never as fast in gaming as the 5800X3D. Basically Alder Lake class.
Just like the 5000 series other than the 5800X3D? Can the 5000 series even hit 60fps on that outdated platform anymore? Just kidding on that last one. I know the 5000 series beat the Skylake arch.

Those AMD customers who waited for the good chip are being rewarded for their patience. Must sting for the ones who didn't.

The cache does help overall responsiveness and smoothness. At least it does with those old Intel edram ones. (Which did hit the wall with CP2077. Really fast in simpler games is one thing, but being unacceptably slow in one I really wanted to play with maxed CPU settings is what convinced me to move on. 5775c couldn't hold 60 fps in full crowd, non HDD mode high traffic locations. Just not enough IPC)

But low power< low temps for me. Mainly for fan noise. My CPU generally doesn't use as much power as my GPU or monitor so I don't fret about that.

I think they claimed similar numbers, but slightly higher, for the 7950X3D, so the 7800X3D would land right behind it. It's funny to me, though, how littleanyone talks about the 7900X3D. But then, iirc that one is quite underwhelming so that's probably why. Also, it's quite obvious to me at least that they waited with releasing the 7800X3D until enough people bought the other two already, because they know they wouldn't have sold the more expensive chips otherwise. It's regular company behavior, look at any GPU series launches lately, or how lower end CPUs from either company are rolled out later, after the run on the high-end models is done.

 

[baboma]

This isn't about whether AMD is better/worse than Intel. It's about whether 'X3D' make sense as "gaming" CPUs.

We'll skip the Intel-vs-AMD fanboy-bait and just use AMD CPUs here for illustration. The 7600X was branded as the "best" gaming CPU on some sites when it came out. The 7600X is $232 on Amazon right now. That's roughly half the 7800X3D's $449 price (when it comes out in April).

Here's the value proposition: Put the $217 difference into the next higher tier GPU (which will get better gaming perf in ALL games and not just some). Then, compare the 7600X + better GPU vs 7800X3D + normal GPU, and see which wins. My money is on the first combo.

For most of us who have a midrange or low-end GPU and not the 4090, the X3D's "superior" perf is a fantasy, because the bottleneck will most likely be the GPU. Sure, there may be a few exceptions, but not enough to warrant the $200+ price difference.


This is my request to Tom's HW: Test the 3XD gaming value proposition by conducting the above 7800X3D vs 7600X test, levelling out the price difference by using a better GPU for the latter.

 

[TerryLaze]

This isn't about whether AMD is better/worse than Intel. It's about whether 'X3D' make sense as "gaming" CPUs.

It makes the same amount of sense as a heavily overclockable CPU does, the games that do get a boost get a good boost but the overall benefit is questionable.

This is my request to Tom's HW: Test the 3XD gaming value proposition by conducting the above 7800X3D vs 7600X test, levelling out the price difference by using a better GPU for the latter.

What is there to test?!
Look at benchmarks of the 7600x and the 7800x3d done at 1080p and look at GPU benches done at 1080p and you can see yourself how a 7600x will do with a faster GPU compared to the 7800x3d with a weaker one.
The maximum amount of frames a CPU can produce does not change.
The x3d will still be faster in the games that can use the extra cache even with a slower GPU, you might have to turn down a few settings in the worst case but it will still be faster.

 

[zipspyder]

First few post look like complete AMD thread crapping...

Its going to match the 13900 or beat it in gaming while it sips power compared to Intel.

 

[dshodson]

While it makes sense to test these cpus at 1080 or less, it just isnt real world performance bc few except possibly professional gamers will play at that resolution. From practical standpoint, It would also be nice to test them with cards one generation back since there have to be many like myself who arent going to buy everything new and trying to figure out if its worth going to the new platform at this time or sticking with something slightly cheaper. Im building a new system to go with a 6900xt and trying to decide where the best bang is located, purchasing from microcenter. Im wondering if the release of the 7800x3d will create better deals or if actually the current combo deals will go away as people rush to pick up the 7800x3d?

 

[frogr]

It is the exact opposite.

X3D works better in older titles.

Intel works better on newer titles.

A 5800X3D can keep up with a 13900k in old titles. But it can not keep up in new titles.

That is likely why AMD always picks older games for their slides.

TechPowerUp has shown this in multiple benchmarks by comparing old and new game titles.

In 26 of 32 games listed above, the 13900k and 5800x3D are +/- 3%. The 7800x3D will likely be > 10% faster in all 26 of the 32.
There will likely be ~ 3 games where the 7800x3D isn't faster than the 13900K.

 

[-Fran-]

This isn't about whether AMD is better/worse than Intel. It's about whether 'X3D' make sense as "gaming" CPUs.

We'll skip the Intel-vs-AMD fanboy-bait and just use AMD CPUs here for illustration. The 7600X was branded as the "best" gaming CPU on some sites when it came out. The 7600X is $232 on Amazon right now. That's roughly half the 7800X3D's $449 price (when it comes out in April).

Here's the value proposition: Put the $217 difference into the next higher tier GPU (which will get better gaming perf in ALL games and not just some). Then, compare the 7600X + better GPU vs 7800X3D + normal GPU, and see which wins. My money is on the first combo.

For most of us who have a midrange or low-end GPU and not the 4090, the X3D's "superior" perf is a fantasy, because the bottleneck will most likely be the GPU. Sure, there may be a few exceptions, but not enough to warrant the $200+ price difference.


This is my request to Tom's HW: Test the 3XD gaming value proposition by conducting the above 7800X3D vs 7600X test, levelling out the price difference by using a better GPU for the latter.

On any other context, I'd agree with you, but some people don't need or want "MOAR FPSSSSS" all the time: they want consistency (A.K.A: good/high 1% lows).

I say this, because the VCache is something that helps more than hampers. Well, the only thing that you could consider a problem is the extra heat and voltage cap for it due to the TSV method used, which basically forces the cores to be clocked lower or the cache goes poof.

Back to the point: I agree on the "value" perspective you mention, but I disagree that for "gaming", the VCache siblings don't offer a valuable extra oomph.

I mean, check MS Flight Sim (and the outliers). If you love those games, the VCache CPUs are like 2 or 3 generations upgrades by themselves. That is bananas.

Also, keep this in mind: Intel's been increasing cache for a good while and most of their gaming gains come directly from those cache increases, more so than their clock increases. HardwareUnboxed investigated this very point and found it to be super linear. So even Intel does benefit from extra cache, but since they don't have it vertically, they don't have AMD's drawbacks, but it's the same situation when you compare across their segmentation.

Regards.

 

[hannibal]

The main point is that according to AMD 7950x3d is about 3% faster than 7800x3d. Wel have all 7950x3d data we need. Take away that 3% and we have 7800x3d!
Impressive for the price, but not earth shattering.
As allways the flagship 7950x3d gets the highest binned chips. And 7800x3d will use the second stage binned chips that expains why 7800x3d is tiny bit slover. 7800x3d is really good gaming CPU, but not as fast as 7950x3d, but it does not lose much!

 

[ManDaddio]

It's always nice to get the facts on the table. Luckily we have you, the resident AMD-hater, to spew nonsense and cheer us up while we wait.

Do you reject the all in one score of only 4% faster from a compilation of many benchmarks?
That is not impressive for the price.
The 7950x is the clear winner. The 13900k is also still a winner for many reasons.

 

[ManDaddio]

The main point is that according to AMD 7950x3d is about 3% faster than 7800x3d. Wel have all 7950x3d data we need. Take away that 3% and we have 7800x3d!
Impressive for the price, but not earth shattering.
As allways the flagship 7950x3d gets the highest binned chips. And 7800x3d will use the second stage binned chips that expains why 7800x3d is tiny bit slover. 7800x3d is really good gaming CPU, but not as fast as 7950x3d, but it does not lose much!

Are trying to say 7950x in some places?

 

[baboma]

I agree on the "value" perspective you mention

Good. Yes, a value proposition test is very worthwhile.

but I disagree that for "gaming", the VCache siblings don't offer a valuable extra oomph.

I never said VCache doesn't offer an "extra oomph" in gaming. Obviously it does. How "valuable" it is to justify the $200+ extra is the question, hence the value proposition test: "gaming" CPU + GPU, or normal 6-core CPU + higher-tier GPU.

CPU and GPU are the two most expensive components in a PC build, and unless you have unlimited money to buy a 4090 + "best" gaming CPU, which most people don't, then you will always have to juggle your budget and decide where your money would be best spent, more CPU and less GPU, or vice versa. This is very real-life question that most face, and a test to determine that would help many people.

some people don't need or want "MOAR FPSSSSS" all the time: they want consistency (A.K.A: good/high 1% lows).

If consistency is what you're after, then a X3D CPU isn't the answer, as it only offers a boost to some games and not all, and the variance of boost is high, making your PC gaming performance much more inconsistent.

The benchmarks we have thus far are all distorted (running 4090 @ 1080p) to remove GPU as a bottleneck. But in real-life, GPU is always a bottleneck because most of us don't have a flagship GPU, or run it at 1080p. The real-life value proposition of a "gaming" CPU is thus much smaller than what the benchmarks indicate.

The main point is that according to AMD 7950x3d is about 3% faster than 7800x3d. Wel have all 7950x3d data we need. Take away that 3% and we have 7800x3d!

This is non-news. We don't need "experts" to tell us 7950X3D has neglible improvement over 7800X3D, because gaming doesn't need 16-cores. It's why the 7600X is a better choice for gaming than the 7950X. In fact, gamers have cried for AMD to do a 7600X3D, but that didn't happen, since AMD wants fatter profits, not more "value" to gamers. Hence, we get white elephants like 7950X3D and 7900X3D.

 

[ingtar33]

Yeah, right. And in a 50 games bench it will be maybe 1%, as last time.

except thats not true. the other x3d chips are all about 10-15% faster then the best intel on the market. this one will be about the same. unlike the other x3d chips this one might fall a little behind on productivity, but gaming should be a slam dunk.

 

[KyaraM]

except thats not true. the other x3d chips are all about 10-15% faster then the best intel on the market. this one will be about the same. unlike the other x3d chips this one might fall a little behind on productivity, but gaming should be a slam dunk.

Please continue reading before replying, because only shortly after I made clear what I'm talking about, including evidence to back my prediction. And no, looking at enough tests and excluding stark outlyers says the difference is maybe 6%, not 10-15%.

 

[-Fran-]

I never said VCache doesn't offer an "extra oomph" in gaming. Obviously it does. How "valuable" it is to justify the $200+ extra is the question, hence the value proposition test: "gaming" CPU + GPU, or normal 6-core CPU + higher-tier GPU.

CPU and GPU are the two most expensive components in a PC build, and unless you have unlimited money to buy a 4090 + "best" gaming CPU, which most people don't, then you will always have to juggle your budget and decide where your money would be best spent, more CPU and less GPU, or vice versa. This is very real-life question that most face, and a test to determine that would help many people.

If consistency is what you're after, then a X3D CPU isn't the answer, as it only offers a boost to some games and not all, and the variance of boost is high, making your PC gaming performance much more inconsistent.

The benchmarks we have thus far are all distorted (running 4090 @ 1080p) to remove GPU as a bottleneck. But in real-life, GPU is always a bottleneck because most of us don't have a flagship GPU, or run it at 1080p. The real-life value proposition of a "gaming" CPU is thus much smaller than what the benchmarks indicate.

This is entirely anecdotal and probably statistically irrelevant, but I'll mention it anyway.

In VR games (and you can extrapolate to the outliers for non-VR experiences) the 5800X3D is basically unbeatable for frame consistency and, funnily enough, I have better frame pacing and consistency in VR stuff than people with better CPUs on paper. For instance, I get better frame consistency than all my friends with Alder or Raptor Lake CPUs and 3090s using my 5800X3D and 6900XT. And a fiend had to swap their 12700K when they realized some games were still unplayable for him after swapping to a 7900XT from a 2080S. They now have a 7900X3D system and they were blown away by the difference in pacing. Not because of the net frame increase, but the how consistent they now are. Even when they drop in FPS, the consistency is not broken for the VR rendering, which is really impressive; and those were their words, not mine.

Let me put it in another perspective: if you buy a 4090 and a 13600K you will definitely get a superb gaming experience, but in the specific use cases where consistency is king, you're better off going to a 7900XTX and pair it with a 5800X3D or a Ry7000X3D. You will even SAVE money doing that.

That's why I'm so keen on saying that the monstrous VCache on charts doesn't tell the whole story of how massively it improves things. That is where the real value of these CPUs comes from.

I think only full frame-time charts can explain that. Intel closes the gap only with super fast DDR5, from what I've noticed and seen everywhere, but it's still not good enough for some specific scenarios; at least in VR.

Regards.