• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

Zen5 only 16 core.

Let's be real they want to save 24 cores for Hedt or at least what's left of it and Epyc and charge $$$$ for it. If intel had a 16P16E configuration I bet it wouldn't take AMD long to figure it out.
They're not going 24c without quad channel (DDR5) or DDR6 ~ this should be absolutely clear to anyone who understands how cores/memory/data work.


Anyone remember this?
 
Seems zen 5 might be only 16 core. It's all you need for gaming though so I guess it's good enough. Intel arrow lake is definitely going to be more than that. So is it going to be AMD for gaming and Intel for Gaming/productivity I wonder.
I'm still going to wait for reviews on both before I decide where to go next.
https://www.pcgamer.com/hardware/pr...on-we-couldnt-do-more-than-16-cores-says-amd/
When current PS5 & Xbox series have only 8 cores - that's what game devs are aiming for, then 12 threads is the most ever needed for game titles. That's why I stick to my budget & "humble" 7600X & tweak the PBO to its max.
But I'm looking forward to what Zen 5 brings to the table when officially released into retail channels with upping of IPC.
 
They're not going 24c without quad channel (DDR5) or DDR6 ~ this should be absolutely clear to anyone who understands how cores/memory/data work.


Anyone remember this?

I meant 24 or more cores and yeah, we will likely not see it till am6 at the same time if Intel was destroying them in non hedt MT I bet it would come sooner... They will likely need to increase the cores per CCD to do it I doubt we will ever see a 4 CCD chip on a standard socket.

I think most, me included would prefer more cores per CCD not more CCDs if they were to go beyond 16 on mainstream desktop.

What's most likely to happen is we are going to get some Hybrid BS with 16 ZenC cores on the second CCD smh.
 
Last edited:
Sure but the cores will be starved, that's the point & probably why AMD's not gone for that many on regular desktop till now. Although Strix Halo is an interesting product in that regard with 256bit memory support, I wonder if they'd ever launch it on desktops with quad channel?
 
Sure but the cores will be starved, that's the point & probably why AMD's not gone for that many on regular desktop till now. Although Strix Halo is an interesting product in that regard with 256bit memory support, I wonder if they'd ever launch it on desktops with quad channel?

I doubt we will actually see it at all except in a laptop with soldered on super fast memory..... Maybe in an AMD version of an intel NUC type device....
 
Try altavista or something.

My point is that it has more cores, yet sells worse.

The word is that 16 cores isn't enough on AM5.

Still, no one thinks the 14900K should have even more cores in order to sell better than the one with less cores (9950X).

Hypocrisy.

You don't understand a thing, and you mix the things.
If and only if the Core i9-14900K doesn't sell, is not because it has more cores, or enough cores, but because it is extremely power hungry, and very expensive (inferior process technology, and not good enough micro-architecture).
 
Sure but the cores will be starved, that's the point & probably why AMD's not gone for that many on regular desktop till now. Although Strix Halo is an interesting product in that regard with 256bit memory support, I wonder if they'd ever launch it on desktops with quad channel?
I doubt we will actually see it at all except in a laptop with soldered on super fast memory..... Maybe in an AMD version of an intel NUC type device....
They probably will, if only because non-HEDT performance desktops with inferior performance to laptops, and for that matter tablets, is not a desirable or stable state of affairs, until it's time to abandon that segment altogether.

Next generation desktop processors and motherboards may very well get 256-bit quad-channel (actually octa-channel DDR5) memory controller at long last, if the AI bubble/boom holds up, as that is something most often memory bandwidth bound. That would also imply some significant changes to how AMD builds their desktop processors, when their current offerings are idling at 20W+ with 1CCD and 30W+ with 2CCD, with an inter-chiplet bandwidth 1/3 less than what 4*32bits of DDR5 theoretically offers right now.
 
They should because Mac mini/pro/studio is going to be a major problem down the road.

Next generation desktop processors and motherboards may very well get 256-bit quad-channel (actually octa-channel DDR5) memory controller at long last, if the AI bubble/boom holds up, as that is something most often memory bandwidth bound.
While they can't match their efficiency, nor is it needed, but getting it more performance/bandwidth is the next logical step IMO. They can also release high(er) end APU with wider memory on desktop ~ that's killing two birds with one stone as they'd get more volume+margins(?) when they're using same chips on desktop and notebooks.

I tried the last Mac Mini & probably the best "PC" in that form factor. The only thing that I hate about Apple's current lineup is the insane markup on memory/storage ~ if they didn’t rob you for a simple upgrade to 16GB RAM or allow more storage I'd definitely have more Apple products right now!

This is where AMD can do better than probably even they'd expect ~ get a 32/64GB Strix halo option in a NUC or a similar BGA board with expandable storage. I'm 100% sure it will be a massive hit!
 
They're not going 24c without quad channel (DDR5) or DDR6 ~ this should be absolutely clear to anyone who understands how cores/memory/data work.


Anyone remember this?
Exactly, one of those reasons I had in my mind. Its not just commercial of nature, but technical too.
And its also the reason I say 'I'll wait until Intel releases a sane CPU'. I mean yay, tons of E cores, but its really only there to counteract their horrible stagnation across the entire package they produce. Monolithic, archaic, too power hungry and way past EOL and yet they still manage to keep it afloat. Its an achievement for sure, but not one I'm rewarding with a purchase. Make something great again pls instead of these bandaid CPUs with excessive burst power usage and TDPs.
 
While that CPU is not shown, it does illustrate that the CPU is very nearly maxed while the GPU has headroom. This is an indication of a CPU bottleneck situation. Throw in an 8 core CPU and that dynamic would change. This very clearly shows that the game in question will utilize lots of cores. And this is more common in game from the last 6 or so years.
Sorry, yeah, the cpu is 12900k with 8e cores and 2p cores off.

Exacty, one of those reasons I had in my mind. Its not just commercial of nature, but technical too.
And its also the reason I say 'I'll wait until Intel releases a sane CPU'. I mean yay, tons of E cores, but its really only there to counteract their horrible stagnation.
Maybe you should try it first. Like grab a cheap one and see for yourself. It might not be ecores themselves that are amazing, but the way they work with the thread director is kinda nuts. You can do stuff that no other cpu can unless you manually set affinities.
 
Maybe you should try it first. Like grab a cheap one and see for yourself. It might not be ecores themselves that are amazing, but the way they work with the thread director is kinda nuts. You can do stuff that no other cpu can unless you manually set affinities.
I'm a simple guy. I buy a CPU once every 5-7 years because the games don't run fine on it anymore. I tinker with it a little bit, and then usually default to some nice 24/7 setting that uses little power but does everything. Thread director... ? Who cares? Drop CPU in socket, play. All I care about is top-end ST performance and sufficient core count at a low power budget, and nice price tag.

As pointed out earlier, the magic is irrelevant to me as long as games run well, especially the performance constrained scenarios, and they are never multithreaded of nature.
 
I'm a simple guy. I buy a CPU once every 5-7 years because the games don't run fine on it anymore. I tinker with it a little bit, and then usually default to some nice 24/7 setting that uses little power but does everything. Thread director... ? Who cares? Drop CPU in socket, play. All I care about is top-end ST performance and sufficient core count at a low power budget, and nice price tag.

As pointed out earlier, the magic is irrelevant to me as long as games run well, especially the performance constrained scenarios, and they are never multithreaded of nature.
I'm just like you frankly, but sometimes I want to do something heavy on the background while playing my game. I just had a huge file unpacking on the background while playing cyberpunk, 0 performance impact. That can't be done on eg a 5950x unless you use Process Lasso.
 
I'm just like you frankly, but sometimes I want to do something heavy on the background while playing my game. I just had a huge file unpacking on the background while playing cyberpunk, 0 performance impact. That can't be done on eg a 5950x unless you use Process Lasso.
You should really stop trying to sell your 14900K. Its too obvious bud. You enjoy your heavy workloads while gaming.
 
I'm a simple guy. I buy a CPU once every 5-7 years because the games don't run fine on it anymore. I tinker with it a little bit, and then usually default to some nice 24/7 setting that uses little power but does everything. Thread director... ? Who cares? Drop CPU in socket, play. All I care about is top-end ST performance and sufficient core count at a low power budget, and nice price tag.

As pointed out earlier, the magic is irrelevant to me as long as games run well, especially the performance constrained scenarios, and they are never multithreaded of nature.

All this talk about 16 cores not being enough 'with a solid IPC to begin with' while I was happily gaming on my 12100F for more than 2 years.:oops:
Single thread performance is still what I care about the most cause most of the games I play prefer that vs high core count and its not like the core count heavy games were unplayable or anything. 'ran into more GPU bound scenarios with my 3060 Ti than CPU bound ones, for example I did finish Cyberpunk with RT on Ultra with that setup and all I had to do is drop crowd density to medium'

Tbh the main reason why I even bothered to upgrade to a 12600KF is cause I wanted a bit more headroom and to speed up the shader compilation in the more recent games.:laugh: 'that and cause I've got a good deal on it since the seller also bought my i3'

Most likely I aint upgrading for years to come, not like I have plans on pairing this with a high end GPU+high refresh rate monitor. '4070 Super/future 5060-Ti is as high as I might go'
 
Tbh the main reason why I even bothered to upgrade to a 12600KF is cause I wanted a bit more headroom and to speed up the shader compilation in the more recent games.:laugh: 'that and cause I've got a good deal on it since the seller also bought my i3'

Having now apparently jumped into 8-core territory with both feet (even in a MiniPC), but still gaming on occasion with a 4790 (and soon a 6700K), I really see those 8-core benefits vs. quad in frame pacing smoothness and the shader compilation step. I still have my i5-8400 but it's been ignored for a while, I should dig it out and see if it plays with the big boys or seems more like a slightly faster quad core.
 
AMD zen5 stuck as Skylake Refresh.... :D
Zen 5 is a grounds up architecture. It's the basis for the upcoming gens. It might still have room for lots of optimizations. It's not even close being a "Skylake refresh". I expected more like >20% IPC improvement. So, actually I am a little bit disappointed. But 16% average isn't that bad at all. Especially because AMD has to think economically. Their 4/3nm production capacities are limited. They have to care about area efficiency. And I am glad that we see 65W TDP on the lower SKUs again. I am looking forward to the 9700X. That could be a really good allrounder and a worthy successor to my 5700X.
 
Zen 5 is a grounds up architecture. It's the basis for the upcoming gens. It might still have room for lots of optimizations. It's not even close being a "Skylake refresh". I expected more like >20% IPC improvement. So, actually I am a little bit disappointed. But 16% average isn't that bad at all. Especially because AMD has to think economically. Their 4/3nm production capacities are limited. They have to care about area efficiency. And I am glad that we see 65W TDP on the lower SKUs again. I am looking forward to the 9700X. That could be a really good allrounder and a worthy successor to my 5700X.

I think what he meant was it's been stuck at 16 cores maximum for half a decade now which is valid intel stuck to 4 cores for too long but that was becuase there was almost 0 competition, and forcing people to a very expensive threadripper isn't much different than what intel was doing back then.

My issue is the entry level is still 6 cores they should have shifted everything down a tier by now imho. So R3 6 cores, R5 8 cores and so on.

Maybe AM6 will see this change but it's over due imo
 
Zen 5 is a grounds up architecture. It's the basis for the upcoming gens.

If so, why didn't they invent a new name for it? Something like Next Zen 1?
Zen 5 is just an evolution with small updates of the original Zen micro-architecture.
 
If so, why didn't they invent a new name for it? Something like Next Zen 1?
Zen 5 is just an evolution with small updates of the original Zen micro-architecture.

I don't even understand why people care about this so much if performance is better what else matters. People were ok with single digit IPC gains from intel for a decade but now we need 20-30% or bust lol. Meteorlake was brand new and guess what it wasn't even good enough to grace us on desktop and until we have the fixed version with Arrowlake we have to wait.

You hear the same things about gpus which also cracks me up....
 
I don't even understand why people care about this so much if performance is better what else matters. People were ok with single digit IPC gains from intel for a decade but now we need 20-30% or bust lol. Meteorlake was brand new and guess what it wasn't even good enough to grace us on desktop and until we have the fixed version with Arrowlake we have to wait.

You hear the same things about gpus which also cracks me up....
Intel gave much more than single digits ST performance gain for a decade. Yes I know you said IPC - but at the end of the day it's ST that matters, not the IPC. Off the top of my head I can tell you that 3770k ---> 6700k is 28% ST gain. 2600k ---> 6700k is 35%. 4770k ---> 6700k is 17% That's within 3 4 and 2 years respectively. AMD aren't doing any better than what Intel did back then, sad but true.
 
I don't even understand why people care about this so much if performance is better what else matters. People were ok with single digit IPC gains from intel for a decade but now we need 20-30% or bust lol. Meteorlake was brand new and guess what it wasn't even good enough to grace us on desktop and until we have the fixed version with Arrowlake we have to wait.

You hear the same things about gpus which also cracks me up....

The difference between Ryzen 9 7950X, Ryzen 9 9950X and Core i7-4770K, Core i7-5775C lies in the TDP. 170 watts vs. 84/65 watts.
There is a very big difference.

Just correcting the misleading claim about "grounds-up" or from scratch.

Let's start with what hasn't changed. Zen 5 is fundamentally the same as Zen 4, with the CCDs (Core Complex Dies) comprising up to eight cores, sharing 32MB of L3 cache. The IOD (Input/Output Die) is practically the same, too, although few details about its feature set have been released so far.

So, you're not getting any more cores, threads, or cache in Zen 5 compared to Zen 4. That may disappoint some folks but given that there was little to complain about the previous architecture, there wasn't much call for any sweeping changes to be implemented.

What is new, though, lies deep within the core structures themselves and it mostly revolves around getting more data to where it needs to be, while improving the overall efficiency of a CCD's number-crunching ability. The Zen 5 announcement is quite light on specifics, but here's what we've been told.

Each core's branch prediction unit has been tweaked to make it more accurate and spit out results quicker. This circuit is responsible for determining what instructions are most likely to be next in line for a core to process, and the rest of the core then fetches the necessary data and instructions from caches based on what the predictor calculates. If it gets it wrong, then precious cycles are wasted in getting the right information.

Thus a branch predictor that's more accurate and responds faster means the overall efficiency of the processing side of the chip is better, so nothing to complain about or criticise there.

Zen 5 also sports "wider pipelines and vectors" but without knowing exactly what AMD is referring to here, it's hard to identify what's precisely changed compared to Zen 4. Traditionally, a wider pipeline means the sequence of logic units that process instructions can handle larger data formats, but in this case, AMD probably means that Zen 5 has more pipelines (or ports, to use the correct term), more instruction schedulers, and the ability to dispatch more instruction per cycle in each core.
 
The difference between Ryzen 9 7950X, Ryzen 9 9950X and Core i7-4770K, Core i7-5775C lies in the TDP. 170 watts vs. 84/65 watts.
There is a very big difference.

Just correcting the misleading claim about "grounds-up" or from scratch.





My only point is marketing from any company like to spew BS all over the place what matters is that it works and performs better than what we had before. Zen5 isn't much more dense than Zan4 so whatever changes they've made regardless of how small/big sound like a decent upgrade. How new or different the actual cores are is irrelevant in my book as long as performance is there.

Who knows maybe these come out and stink up the joint but I doubt it.

I am less optimistic about gaming performance increases at least vs 3D 7000 parts though.

At this point we know almost nothing with how these will perfom untill Wiz gets ahold of them and at the very least I'm excited for that.

Intel gave much more than single digits ST performance gain for a decade. Yes I know you said IPC - but at the end of the day it's ST that matters, not the IPC. Off the top of my head I can tell you that 3770k ---> 6700k is 28% ST gain. 2600k ---> 6700k is 35%. 4770k ---> 6700k is 17% That's within 3 4 and 2 years respectively. AMD aren't doing any better than what Intel did back then, sad but true.

I owned almost every generation from sandybridge to skylake it was a yawn fest to the point I jumped to hedt and they got stuck after that again only increasing core counts and cache... The biggest jump came from DDR4 based systems and even then it was rough during the first generation. Intel also left less and less overclocking headroom generation to generation so only if you looked at reviews the ST was decent, if you overclocked every chip they were much closer.
 
Back
Top