freeagentI hope they kill X3D, and just unify the line :)

i doubt it that would be alot of wasted silicon. Not everyone that buys computers has gaming as a number #1 priority.

There are people that will choose a 9950X vs a 9950X3D because they do more work and don't really play games. And the money saved they can put towards ram or an SSD.

I just had a friend that did this because he mostly games on his console but needs the PC to render.

AusWolfYes, but realistic. People who build high-end systems to play games at 720p low are a myth. Their existence isn't entirely unimaginable, but I haven't met them.

As explained already... but this is a short sighted approach to a CPU review. Its the wrong place to be gauging the performance of your gaming rig, which relies primarily on GPU performance. Every once in a while we get new people that bring this Youtube stance to this forum, so I'm surprised you're saying this.

For CPUs what really matters is two things:
- is it fast enough to run your content
- is it fast enough to carry your GPU

You can't review that properly against the other CPUs in the playing field, too many variables. So the best way to approach CPU performance is by relative performance. That, combined with the knowledge of what minimum CPU you need to get by for your personal use case, allows you to make an informed choice. And to get the best look at relative performance, you make sure the CPU is the bottleneck.

Vayra86For CPUs what really matters is two things:
- is it fast enough to run your content
- is it fast enough to carry your GPU

Any AM5 CPU is going to do day to day tasks and be enough to drive a GPU, but if someone's buying a 5090 they're probably going to buy a 9800X3D anyway.
I can understand the complaints of unrealistic benchmarks, and I definitely appreciate seeing real world gaming tests over GPU limited 720P tests or a bajillion FPS in a competitive game that no one but a handful of actually competitive e-sports players care about.

Hecate91Any AM5 CPU is going to do day to day tasks and be enough to drive a GPU, but if someone's buying a 5090 they're probably going to buy a 9800X3D anyway.
I can understand the complaints of unrealistic benchmarks, and I definitely appreciate seeing real world gaming tests over GPU limited 720P tests or a bajillion FPS in a competitive game that no one but a handful of actually competitive e-sports players care about.

If you understand the complaints you've failed to interpret CPU reviews all your life.
The tests are not unrealistic, they isolate the CPU, and if you have a fast GPU, or plans to buy one going forward, those tests are extremely accurate to gauge what's the best CPU to buy. If you have a slower GPU, the CPU results show you a realistic scenario too, because the GPU doesn't weigh in on the results. It makes the CPU testing pretty much 'content agnostic' and just shows raw performance. It doesn't get a lot more 'what you see is what you get' than that. Your request is an exercise in futility.

If all you want is 'enough to play game X' go ahead and read or watch a performance review on that specific game - which also do exist.

Beyond that, there are also various game scenarios where you want as much CPU as you can throw at it, even completely discounting what GPU you have - this applies to almost every sim building game, grand strategy, 4X, because the game logic will eventually just choke your framerate - and every 100 mhz helps to stay away from that as long as possible. Lots of cache helps too, and if the game is properly threaded, even core count can help. For that scenario, I really do want to see those 720p results and isolated CPU performance relative to the rest of the CPUs I could buy. Even on AM5, it doesn't matter, you can choke every CPU, no problem.

MakaveliThere are people that will choose a 9950X vs a 9950X3D because they do more work and don't really play games.

This is the part about X3D that annoys me.

Say you are just a Joe, not knowing what is what these days. You buy the 9950 because you work and play games, maybe do some science stuff on the side.

But then 2 months later you see there is a 9950X3D. Does everything you wanted your 9950 to do, but it does it better for games and science.

So now you are pissed at AMD. Just make everything on the top end that would have an X to an X3D. Problem solved.

Edit:

I almost bought a 9800X3D this weekend, glad that I did not.. because I will get the faster one :)

The double X3D version would mean no more core parking, right? And by extension, and this is the real kicker, no more CrapBox GameBar and no more 'Game Mode On' and all those unnecessary shenanigans, riiiiight? :)

freeagentSo now you are pissed at AMD. Just make everything on the top end that would have an X to an X3D. Problem solved.

The problem with that is everything at or near the high would be expensive with an X3D, clocks lower, and they'd only lose sales to Intel on price.

Or you could just wait 5-6 months for a sale or new products, I'd be pissed having not waited for at least a sale on the non-X3D, but that's just me not being a "true enthusiast" being able to resist buying the new shiny right away.

Hecate91The problem with that is everything at or near the high would be expensive with an X3D, clocks lower, and they'd only lose sales to Intel on price.

Or you could just wait 5-6 months for a sale or new products, I'd be pissed having not waited for at least a sale on the non-X3D, but that's just me not being a "true enthusiast" being able to resist buying the new shiny right away.

I have been good with this socket. On AM4 I started with 3600XT, then 5600X, 5900X, 5800X3D.. so far I am still using just the 9900X lol..

Its been pretty good man, cant complain. Does everything I want except boost to 6GHz :)

Hecate91Have a source to confirm this?
What AMD is doing is no different than what Intel does with a refresh or "KS" version of a CPU.
Newer manufacturing runs likely means AMD could get higher clocks, but why am I not surprised by reactions like this.

Glass half full vs. glass half empty. I'll take the latter approach. KS processors have a whole marketing fanfare and a binned silicon shtick, either way you're paying for that privilege and its existence makes a lesser bin (aka cheaper SKU) less likely to exist in the future, it may come after Zen 5 has been superseded by something else

Either way it goes: you lose

freeagentThis is the part about X3D that annoys me.

Say you are just a Joe, not knowing what is what these days. You buy the 9950 because you work and play games, maybe do some science stuff on the side.

But then 2 months later you see there is a 9950X3D. Does everything you wanted your 9950 to do, but it does it better for games and science.

So now you are pissed at AMD. Just make everything on the top end that would have an X to an X3D. Problem solved.

Edit:

I almost bought a 9800X3D this weekend, glad that I did not.. because I will get the faster one :)

I see your point but this to me is on the end users.

If both parts are available on the market at the same time you have to do your due diligence before opening your wallet.

If you didn't do proper research and have buyers remorse after that's really on the user.

MakaveliI see your point but this to me is on the end users.

If both parts are available on the market at the same time you have to do your due diligence before opening your wallet.

If you didn't to proper research and have buyers remorse after that's really on the user.

Correct, and honestly this isn't really a deal breaker as nothing is gonna change, anyone who's had their 9800X3D for what, a year now, has owned what's essentially the exact same processor anyways. No one who bought a i9-13900K was bothered by the i9-14900K existing, after all. It's OK to release a higher end SKU, the problem I have is that by introducing a 9850 instead of a 9700X3D, they're effectively relegating its position to the 9800X3D - without a price reduction. It'll likely just be phased out or both of them will be sold at their MSRPs (or above), effectively raising ASP.

If anything, I'm more miffed at AMD not releasing the dual 3D processor already. The 9950X3D is a compromise, I don't do compromises.

RogueSixThe double X3D version would mean no more core parking, right? And by extension, and this is the real kicker, no more CrapBox GameBar and no more 'Game Mode On' and all those unnecessary shenanigans, riiiiight? :)

No, because even with X3D on both CCDs, there's still the inter CCD latency penalty. Double X3D doesn't make sense until the packaging is improved.

dgianstefaniNo, because even with X3D on both CCDs, there's still the inter CCD latency penalty.

But it'll be just like the regular Ryzen 9's, by "core parking" I think they're probably asking about the whole affinity issue the current processors have. It won't be 2x 9800X3D just like the 9950X isn't 2 9700X processors, but it'll definitely be better than what we have now and won't require you to isolate or outright disable cores to ensure the chip behaves. No need for drivers, no need for special optimization, no need for core affinity - just boot and use, as a CPU should be

MakaveliIf both parts are available on the market at the same time you have to do your due diligence before opening your wallet.

Which is why I said 2 months later you see there is an X3D on the market..

There are people out there who just look at what the store has, and buy the top stuff without batting an eye.

If the top product changes within a month or two, I would be pissed too.

But how often does that happen? I know people love X3D for obvious reasons, but to me the segmentation is annoying :)

Dr. DroBut it'll be just like the regular Ryzen 9's, by "core parking" I think they're probably asking about the whole affinity issue the current processors have. It won't be 2x 9800X3D just like the 9950X isn't 2 9700X processors, but it'll definitely be better than what we have now and won't require you to isolate or outright disable cores to ensure the chip behaves. No need for drivers, no need for special optimization, no need for core affinity - just boot and use, as a CPU should be

You'll still want all game threads on a single CCD. It's better for multithreaded work that expects all cores to behave the same instead of half the CPU having 3-400 MHz difference from the other half, sure, but I highly doubt the dual X3D chip is better for gaming, if it also has higher clocks than the X3D CCD of the 9950X3D then it will perform better than the 9950X3D, but the same logic of sticking to one CCD at a time applies. Willing to bet a stock 9850X3D will perform better in games than a hypothetical stock 9950X3D2, just like a stock 9800X3D beats a stock 9950X3D.

Also, the regular Ryzen 9s still have one CCD being faster than the other, the 9950X has one fast CCD and one slower one, this causes issues in applications that expect all cores to be the same.

Dr. DroBut it'll be just like the regular Ryzen 9's, by "core parking" I think they're probably asking about the whole affinity issue the current processors have. It won't be 2x 9800X3D just like the 9950X isn't 2 9700X processors, but it'll definitely be better than what we have now and won't require you to isolate or outright disable cores to ensure the chip behaves. No need for drivers, no need for special optimization, no need for core affinity - just boot and use, as a CPU should be

Agreed its fixes one problem but not the other.

I'm hoping this will be fixed in Zen 6.

freeagentWhich is why I said 2 months later you see there is an X3D on the market..

There are people out there who just look at what the store has, and buy the top stuff without batting an eye.

If the top product changes within a month or two, I would be pissed too.

But how often does that happen? I know people love X3D for obvious reasons, but to me the segmentation is annoying :)

I mean, it's very common in the tech world in a steady cadence, at least once a year you can expect your shiny investment to have a "scuff" in the form of something cooler being brought to market. We only really have a problem with this scenario: a "refresh" of an existing product is introduced at a higher price level without reducing the original's price point, setting the stage for a generational price hike. The next one will cost the same as the higher-priced refresh, as the market will then bear said price.

dgianstefaniYou'll still want all game threads on a single CCD. It's better for multithreaded work that expects all cores to behave the same, sure, but I highly doubt the dual X3D chip is better for gaming, unless it also has higher clocks, the same logic of sticking to one CCD at a time applies.

Most of the time, yes, but this also applies to the standard Ryzen 9 chips. Ever since the 5950X it's been that way, if more resources are needed, they will be available with a minor to moderate latency hit but will be available nonetheless, something that plain isn't possible with the current CPUs that have a 3D and a standard die. It's always been more of a concern with the 7900X3D and 9900X3D, the performance on those falls off a cliff if more than 6 cores are requested.

9800X3D is already the worst fps/dollar gaming CPU right now, 9850X3D pricing concerns me, but this is good news to those who don't care about value and just want the best product

I wish we could get a 9600X3D already. X3D is great and all but its gate-kept to the high end. Even a last gen 7800X3D is more expensive than a 265K/9700x

Praying for Nova Lake bllc to challenge the X3D monopoly so there can be better prices on astronomically large L3 CPUs

dgianstefaniAlso, the regular Ryzen 9s still have one CCD being faster than the other, the 9950X has one fast CCD and one slower one, this causes issues in applications that expect all cores to be the same.

For my 9900X when I let PBO run the show it is a bit annoying. If I take control and do per CCD static clocks, best I can do is 5600MHz on CCD0 and 5500MHz on CCD1.

And I cannot get more than 15 passes in Linpack, I have to drop clocks down to 5400 to do it. I can already do that with PBO so not much point.

I hope the dual X3D gives you the best that Zen 5 can offer in both clocks, power tolerance, and of course raw performance.

One last real hurrah before Zen 6 comes in for the slaughter lol..

AusWolfI'm not talking about 20 W. I'm talking about the 9800X3D eating twice as much as the 7800X3D does. That's 155 vs 74 W according to the TPU review. All that for a 15% performance gain in applications (translate that to 1.5x the power consumption in gaming for 3% performance gain). And now we have this thing pushing the limits even further. Please allow me to call it wasteful.

I suppose that isn't exactly wrong, but it's more the fault of power profiles rather than fundamental CPU design flaws.
If you're really anal about power usage, you can undervolt and underclock a 9800X3D and it will still be 5-10% faster than a 7800X3D while using the same amount of power.
To be fair, that still isn't amazing in comparison, and we can blame AMD for pushing the 9800X3D out of its optimal efficiency range, but the 9800X3D is an improvement over the 7800X3D even if it's only a fairly modest one.

And you should also consider that in most applications with a 9800X3D, you won't be CPU-limited, so the CPU normally won't be running at its maximum clocks and using 155W. There are some situations where it can, but only when you would actually benefit from the extra CPU performance.

MakaveliThis review is using this for a cooler which is similar in size to your Noctua NH-U9S

CPU Cooler - be quiet! DARK ROCK 5

Wait, you're taking about gaming power consumption? I'm taking about peak power. W1zz could barely contain the 9800X3D with an NH-D15 at peak, while the 7800X3D only needs about -10 W with my U9S to achieve similar temperatures as W1zz did with the 9800 and D15.

freeagentThis is the part about X3D that annoys me.

Say you are just a Joe, not knowing what is what these days. You buy the 9950 because you work and play games, maybe do some science stuff on the side.

But then 2 months later you see there is a 9950X3D. Does everything you wanted your 9950 to do, but it does it better for games and science.

So now you are pissed at AMD. Just make everything on the top end that would have an X to an X3D. Problem solved.

Edit:

I almost bought a 9800X3D this weekend, glad that I did not.. because I will get the faster one :)

This makes no sense at all. There is always a better CPU coming out if you wait. It's not AMD's fault if you bought at the wrong time.

Vayra86As explained already... but this is a short sighted approach to a CPU review. Its the wrong place to be gauging the performance of your gaming rig, which relies primarily on GPU performance. Every once in a while we get new people that bring this Youtube stance to this forum, so I'm surprised you're saying this.

For CPUs what really matters is two things:
- is it fast enough to run your content
- is it fast enough to carry your GPU

You can't review that properly against the other CPUs in the playing field, too many variables. So the best way to approach CPU performance is by relative performance. That, combined with the knowledge of what minimum CPU you need to get by for your personal use case, allows you to make an informed choice. And to get the best look at relative performance, you make sure the CPU is the bottleneck.

There is a way to test CPUs against each other - by limiting them as much as possible (720p low).

And there is a way to test if a CPU is good enough for one's needs (at real resolutions and settings that people use).

I'm not questioning the testing methodology, I'm only questioning the usefulness of the results. As someone who has gone through a 7700X, 7800X3D, 7950X and 7950X3D, and saw absolutely no difference in gaming, please allow me to be skeptical.

AusWolfI'm not talking about 20 W. I'm talking about the 9800X3D eating twice as much as the 7800X3D does. That's 155 vs 74 W according to the TPU review. All that for a 15% performance gain in applications (translate that to 1.5x the power consumption in gaming for 3% performance gain). And now we have this thing pushing the limits even further. Please allow me to call it wasteful.

You're allowed to call it whatever you want. I'm allowed to tell you you're making a mountain out of a molehill. Gaming is a purely wasteful hobby in the first place.

AusWolfWait, you're taking about gaming power consumption? I'm taking about peak power. W1zz could barely contain the 9800X3D with an NH-D15 at peak, while the 7800X3D only needs about -10 W with my U9S to achieve similar temperatures as W1zz did with the 9800 and D15.

And how often are you running at peak power? 0.1ns a day?

AusWolfThere is a way to test CPUs against each other - by limiting them as much as possible (720p low).

And there is a way to test if a CPU is good enough for one's needs (at real resolutions and settings that people use).

I'm not questioning the testing methodology, I'm only questioning the usefulness of the results. As someone who has gone through a 7700X, 7800X3D, 7950X and 7950X3D, and saw absolutely no difference in gaming, please allow me to be skeptical.

Again, no one says you cant be skeptical. But we can call you out for not understanding the purpose of testing, which if you did you wouldnt have to question the usefulness of the results. Unless you played a game that was CPU limited you are not going to see the results. Such games are extremely rare now, but in 8+ years? CPUs have long lifespans.

People had these exact same complaints when reviewing ivy bridge VS piledriver back in the day. Guess what? Those CPUs that were ridiculously fast in those 720p tests that produced useless data? They were still capable of gaming a decade later, while their Piledriver counterparts were left on the struggle bus.

Hecate91Any AM5 CPU is going to do day to day tasks and be enough to drive a GPU, but if someone's buying a 5090 they're probably going to buy a 9800X3D anyway.
I can understand the complaints of unrealistic benchmarks, and I definitely appreciate seeing real world gaming tests over GPU limited 720P tests or a bajillion FPS in a competitive game that no one but a handful of actually competitive e-sports players care about.

The purpose of 720p testing is to find bottlenecks in CPU performance, not to show a realistic resolution to game on. Yet we always hear certain members of the peanut gallery complaining about how nobody games at 720p.

People really, REALLY struggle with this basic concept. Reminds me of the automotive industry where people whined and gnashed their teeth for years that German car makers were testing the 150-0 MPH break distance when nobody ever realistically drove at 150 MPH. Even when shown that the British cars, which were never designs to go over 70, handled significantly worse and slowed down far worse then their German counterparts, they still cried about how nobody drove 150.

*shrug* some people will just never be able to comprehend such advanced ideas.

TheinsanegamerNYou're allowed to call it whatever you want. I'm allowed to tell you you're making a mountain out of a molehill. Gaming is a purely wasteful hobby in the first place.

That doesn't mean we have to be crazy about it.

TheinsanegamerNAnd how often are you running at peak power? 0.1ns a day?

Pretty much all the time when I'm not sitting in front of it. I run BOINC/WCG. Why would I have a 7950X3D running at base clock if I only cared about gaming?

Besides, I believe that one's temperatures should be under control in all conditions.

TheinsanegamerNAgain, no one says you cant be skeptical. But we can call you out for not understanding the purpose of testing, which if you did you wouldnt have to question the usefulness of the results. Unless you played a game that was CPU limited you are not going to see the results. Such games are extremely rare now, but in 8+ years? CPUs have long lifespans.

People had these exact same complaints when reviewing ivy bridge VS piledriver back in the day. Guess what? Those CPUs that were ridiculously fast in those 720p tests that produced useless data? They were still capable of gaming a decade later, while their Piledriver counterparts were left on the struggle bus.

The purpose of 720p testing is to find bottlenecks in CPU performance, not to show a realistic resolution to game on. Yet we always hear certain members of the peanut gallery complaining about how nobody games at 720p.

People really, REALLY struggle with this basic concept. Reminds me of the automotive industry where people whined and gnashed their teeth for years that German car makers were testing the 150-0 MPH break distance when nobody ever realistically drove at 150 MPH. Even when shown that the British cars, which were never designs to go over 70, handled significantly worse and slowed down far worse then their German counterparts, they still cried about how nobody drove 150.

*shrug* some people will just never be able to comprehend such advanced ideas.

I understand the purpose of 720p tests very well. But please, have a look around TPU, and tell me how many people you see running a 5080 or 5090 with a 9900K.