Discussion Speculation: Zen 4 (EPYC 4 "Genoa", Ryzen 7000, etc.)

[Vattila]

Except for the details about the improvements in the microarchitecture, we now know pretty well what to expect with Zen 3.

The leaked presentation by AMD Senior Manager Martin Hilgeman shows that EPYC 3 "Milan" will, as promised and expected, reuse the current platform (SP3), and the system architecture and packaging looks to be the same, with the same 9-die chiplet design and the same maximum core and thread-count (no SMT-4, contrary to rumour). The biggest change revealed so far is the enlargement of the compute complex from 4 cores to 8 cores, all sharing a larger L3 cache ("32+ MB", likely to double to 64 MB, I think).

Hilgeman's slides did also show that EPYC 4 "Genoa" is in the definition phase (or was at the time of the presentation in September, at least), and will come with a new platform (SP5), with new memory support (likely DDR5).

What else do you think we will see with Zen 4? PCI-Express 5 support? Increased core-count? 4-way SMT? New packaging (interposer, 2.5D, 3D)? Integrated memory on package (HBM)?

Vote in the poll and share your thoughts!

 

[eek2121]

At the level of M1 Ultra, they kind of are. Those users are at least aware of what "the other side" has and are actively interested in what "the other side" can and can't do faster or better. Those users still wind up with Macs for reasons other than performance much of the time, but they're aware.

I am just getting caught up in this thread and have to go offline, but I had to respond to your comment.

In the workloads that matter, Apple is still not able to compete. Just this last week I found, and had to track down multiple folks regarding an M1/ultra/max bug that exists (it involves character encoding and compiler “assumptions” so it may be a bug in xcode) while running on an M1 ultra vs. an x86 Mac with the same setup. Apple is looking into it apparently. Unfortunately our team had to do a rollback in preparation for a hotfix. We were able to reproduce the issue via an automated test, however, open source compilers on a raspberry pi running raspbian or ubuntu do not exhibit the issue.

Apple needs to do better if they want to convince folks to move to their platform. They need an enterprise team and products ASAP.

 

[Doug S]

I am just getting caught up in this thread and have to go offline, but I had to respond to your comment.

In the workloads that matter, Apple is still not able to compete. Just this last week I found, and had to track down multiple folks regarding an M1/ultra/max bug that exists (it involves character encoding and compiler “assumptions” so it may be a bug in xcode) while running on an M1 ultra vs. an x86 Mac with the same setup. Apple is looking into it apparently. Unfortunately our team had to do a rollback in preparation for a hotfix. We were able to reproduce the issue via an automated test, however, open source compilers on a raspberry pi running raspbian or ubuntu do not exhibit the issue.

Apple needs to do better if they want to convince folks to move to their platform. They need an enterprise team and products ASAP.

Apple sees themselves as a consumer focused company. They will sell to business, but that's not their primary focus like it is for Intel and Microsoft. They don't feel the need to convince businesses to move to Mac. If they want to do so, great, but they will never devote the necessary resources to put their enterprise support at the level of Intel's and Microsoft's.

They are concerned with getting consumers to switch - or at least consider switching which is over half the battle in a world where "PC" and "Windows" have become synonymous for most. As with iPhone they only care about the upper end of the PC market. You can look at iPhone's market share in various countries and that probably provides the absolute ceiling for where Mac could reach in their wildest dreams. They will happily let Windows 10S and Chrome fight it out at the low end.

 

[DrMrLordX]

And we are totally sure Intel will have no problem launching Intel 4 on time, with the desired characteristics and volume?

Intel 4 is already horribly late, unless you count Loihi 2.

All that aside

Is Mi300 going to exist only in APU format, or will they still have add-in accelerators via PCIe/CXL? I guess this may be the wrong thread for that but if Mi300 is only going to be paired on board/package with Genoa, it certainly changes the face of AMD's enterprise dGPU lineup (and CPU lineup).

 

[Exist50]

Intel 4 is already horribly late, unless you count Loihi 2.

All that aside

Is Mi300 going to exist only in APU format, or will they still have add-in accelerators via PCIe/CXL? I guess this may be the wrong thread for that but if Mi300 is only going to be paired on board/package with Genoa, it certainly changes the face of AMD's enterprise dGPU lineup (and CPU lineup).

It has to support at least some PCIe/CXL lanes for networking. Presumably will do the same thing they do for EPYC, with lanes functioning as either infinity fabric or PCIe.

 

[Vattila]

Is Mi300 going to exist only in APU format, or will they still have add-in accelerators via PCIe/CXL? I guess this may be the wrong thread for that but if Mi300 is only going to be paired on board/package with Genoa, it certainly changes the face of AMD's enterprise dGPU lineup (and CPU lineup).

CDNA 3 will most likely arrive in other configurations and form factors. Whether it will be marketed in the MI300 line-up, I don't know. AMD may invent separate product lines.

In any case, MI300 has been announced as "the world's first data centre APU" based on "Zen 4", so this is certainly the right thread to speculate about it.

 

[Hans Gruber]

You seem to forget 4 things about Alder Lake:
1) They are only leading in Single thread
2) They are only leading in Desktop
3) They suck power like no tomorrow.
4) Zen 4 is just around the corner and will most likely take over single thread and maintain power/perf.

I have a 12700F, and 6 5950x's. The 12700F sucks the power of 2 5950x's and get creamed in total performance 4x by 2 5950x's, or 2x one on one.
When your electric bill is $800 a month, electricity use really makes a difference. I don't even have the wiring to support 6 12900k's. And in the server world, its not just me, but MANY companies are switching to Milan due to performance and power/perf.

YOU are the delusional one.

So Intel made leaps and bounds with Alder Lake and few acknowledge or realize that Alder Lake was the first Intel processor not on the 14nm process in more than 6 years. The improved efficiencies came from the 10nm process. AMD has benefitted hugely from TSMC 7nm process efficiency wise. Obviously, 5nm will be good for AMD.

The problem you raise is a server issue. You (Markfw) and a few other guys have Threadripper systems. Most do not have a 5950x here. I have said in past posts that Alder Lake wiped the floor with every Zen 3 CPU other than the 5950x. If AMD had fire sale pricing on Zen 3 CPU's, that would change how the market weighs the value in processors. In computer terms they do reviews based on overall performance and factor in power usage as a secondary consideration.

Kids today have gotten over their fear of going into the bios. But they are afraid to tweak anything. Manual memory timings, too risky. Jumpers to restore a hanging system back to stock bios, too complicated.

I have said that Zen 4 should have been on the market yesterday. The way it looks, Zen 4 will be on the open market 1-2 months prior to Raptor Lake. Nobody should be worried about energy efficiency on Zen 4. Most are not worried that Raptor Lake may destroy Zen 4. But those who understand the CPU industry are worried about the added efficiencies that Intel's 7nm (Intel 4) will bring to Meteor Lake. Zen 4 will be on 5nm but it's expected to be pretty close in performance to Raptor Lake which is not good considering the Fab change from 7nm to 5nm for AMD. Many thought the 5nm process would put AMD way ahead of Raptor Lake not in just energy efficiency but also performance wise.

From personal experience I can tell you that Zen 3 is a significant jump in performance vs. Zen 2. People expect big performance from Zen 4 on the 5nm process.

 

[uzzi38]

Intel 4 is already horribly late, unless you count Loihi 2.

All that aside

Is Mi300 going to exist only in APU format, or will they still have add-in accelerators via PCIe/CXL? I guess this may be the wrong thread for that but if Mi300 is only going to be paired on board/package with Genoa, it certainly changes the face of AMD's enterprise dGPU lineup (and CPU lineup).

If you have both CPU and GPU cores on the same socket, do you really need to ship it with discrete CPUs as well?

Mind you I'm just guessing here, but wouldn't it make more sense to create a package that pairs, say, 16 Zen 4 cores and a GCD of a certain size together alongside some HBM2E stacks? Rather than have an entire Genoa package, you ship multiple smaller GCD + CCD combos, effectively?

Maybe in the future if AMD's software stack is more comparable to Nvidia's, but for now I don't think having discrete parts is really all that necessary.

 

[DrMrLordX]

If you have both CPU and GPU cores on the same socket, do you really need to ship it with discrete CPUs as well?

Mind you I'm just guessing here, but wouldn't it make more sense to create a package that pairs, say, 16 Zen 4 cores and a GCD of a certain size together alongside some HBM2E stacks? Rather than have an entire Genoa package, you ship multiple smaller GCD + CCD combos, effectively?

APUs introduce inflexibility, but they can reduce costs and improve performance, assuming you have all the compute resources you want. Which for once, it seems like AMD can deliver on a single APU module. There's tradeoffs. AMD can probably rely on their semi-custom experience to get customers the right ratio of CPU/GPU per APU package. But for off-the-shelf buyers this may not suffice.

Maybe in the future if AMD's software stack is more comparable to Nvidia's, but for now I don't think having discrete parts is really all that necessary.

If AMD's software stack is going to be fully CXL-compliant, it needs to be able to do all the things NV can do with CUDA already. Even on an APU. Fortunately I think they're close, if not there already, assuming you use the right hardware.

 

[naad]

Hopefully this isn't off topic.

So around the zen2 launch, some folk managed to get zen2 cores clocking significantly higher than their boost, it was close to 5.0 IIRC and using crazy amounts of power, yet the game performance was unchanged from stock, very different behavior compared to heavily overclocked Intel chips.
Does anyone know why this was? Making an uarch scale with clocks is a challenge on its own, so I appreciate what amd is trying with zen4, but what's the actual issue in this case? slower memory subsystem couldn't keep up? monitor applications read higher clocks but internally chip was downclocking?

 

[biostud]

So Intel made leaps and bounds with Alder Lake and few acknowledge or realize that Alder Lake was the first Intel processor not on the 14nm process in more than 6 years. The improved efficiencies came from the 10nm process. AMD has benefitted hugely from TSMC 7nm process efficiency wise. Obviously, 5nm will be good for AMD.

The problem you raise is a server issue. You (Markfw) and a few other guys have Threadripper systems. Most do not have a 5950x here. I have said in past posts that Alder Lake wiped the floor with every Zen 3 CPU other than the 5950x. If AMD had fire sale pricing on Zen 3 CPU's, that would change how the market weighs the value in processors. In computer terms they do reviews based on overall performance and factor in power usage as a secondary consideration.

Kids today have gotten over their fear of going into the bios. But they are afraid to tweak anything. Manual memory timings, too risky. Jumpers to restore a hanging system back to stock bios, too complicated.

I have said that Zen 4 should have been on the market yesterday. The way it looks, Zen 4 will be on the open market 1-2 months prior to Raptor Lake. Nobody should be worried about energy efficiency on Zen 4. Most are not worried that Raptor Lake may destroy Zen 4. But those who understand the CPU industry are worried about the added efficiencies that Intel's 7nm (Intel 4) will bring to Meteor Lake. Zen 4 will be on 5nm but it's expected to be pretty close in performance to Raptor Lake which is not good considering the Fab change from 7nm to 5nm for AMD. Many thought the 5nm process would put AMD way ahead of Raptor Lake not in just energy efficiency but also performance wise.

From personal experience I can tell you that Zen 3 is a significant jump in performance vs. Zen 2. People expect big performance from Zen 4 on the 5nm process.

I think a lot of enthusiasts uses very big words, to describe differences in processor speeds that on benchmarks might show differences that could support these words, but in real life usage really show very little difference. If you do have your computer working full load or have some very specific software you need your computer optimized for, sure choosing the right processor can make a huge difference. As a tech enthusiast I like to spend a lot of time on it as well, but basically whether I choose Raptor lake or Zen4 in real life would make zero difference, when I game and do some photo editing, and my guess is that would be the case for most users. Probably more than 90% of users could go either way, and maybe 5% has a workflow that strongly favors Intel and 5% has a workflow that favors AMD.

I like to discuss the benefits and downsides of both brands, so I can choose what I prefer.

 

[itsmydamnation]

So Intel made leaps and bounds with Alder Lake and few acknowledge or realize that Alder Lake was the first Intel processor not on the 14nm process in more than 6 years. The improved efficiencies came from the 10nm process. AMD has benefitted hugely from TSMC 7nm process efficiency wise. Obviously, 5nm will be good for AMD.

Why do you keep telling this lie , we had both Cannonlake and Icelake. Just because intel was unable to bring them into 1 of many form factors doesn't change this fact. its been all mediocre uarch upgrades for 6 years.

I have said that Zen 4 should have been on the market yesterday. The way it looks, Zen 4 will be on the open market 1-2 months prior to Raptor Lake. Nobody should be worried about energy efficiency on Zen 4. Most are not worried that Raptor Lake may destroy Zen 4. But those who understand the CPU industry are worried about the added efficiencies that Intel's 7nm (Intel 4) will bring to Meteor Lake. Zen 4 will be on 5nm but it's expected to be pretty close in performance to Raptor Lake which is not good considering the Fab change from 7nm to 5nm for AMD. Many thought the 5nm process would put AMD way ahead of Raptor Lake not in just energy efficiency but also performance wise.

Absolute load of crap. Intel P cores are 1.5 - 2.0 times the size in terms of decode , L1D, L2 , ROB , load store etc and the best they can manage is what a tie in SPEC2017-int and a 14% lead in SPEC2017-fp , thats with DDR5 to boot. AMD are about to stretch that efficiency even further with Zen4. that means anything thats power limited per core ( Servers, Laptops, embeded ) The performance gap is going to grow and by a lot. So if we keep with SPEC2017-int Zen4 is probably going to score around 9.3 vs 8.14 for GoldenCove at the same power as Zen3 or the same score as GoldenCove at what like 1/4th the power.

Now lets flash forward into the future, what are intel going to do to scale core performance with all these transistors ( targeting highest density of any fab worked so well for 10nm) . Anything that is increasing width is going to cost a lot because scaling units/width of a core is not linear. So intel will have to do something that they haven't done since like haswell ( skylake might just make it) which is be more efficient at ISO process then the previous Core or they "waste" the node jump. Now we look at AMD, they have a "small" core in terms of structures and width compared to GC and from Zen 1 to 2 to 3 and probably to 4 demonstrated increased efficiency at ISO process. So come Zen5 they will be able to grow core width easier and thus scale performance.

So then you will run back to the E cores, they might save intel on throughput in the desktop space but thats about it. In Servers they will be nothing but the value option vs Zen4d. We can already see what Zen3 on 6nm does relative to intel in the laptop space, more e-cores wont save them

So you might be "worried" for AMD with 4nm, i would be far more worried about intel being able to design cores anywhere near at the same level of ARM/Apple/AMD/soon to be QUALCOMM. This isnt the days of AMD SOI 90nm vs intel 65nm , you don't cut your die in 1/2 , cut your power in half and get a big clock jump all at the same time, now you have to choose which one of the 3 you want.

 

[MadRat]

Has anyone been considering the physical constraints of cramming all this memory and bandwidth into a single mainboard? Surely they are reaching the point where they will have to be more creative than ever for meeting these new power requirements. Progress on the CPU moves in lockstep with the mainboard.

 

[JoeRambo]

Absolute load of crap. Intel P cores are 1.5 - 2.0 times the size in terms of decode , L1D, L2 , ROB , load store etc and the best they can manage is what a tie in SPEC2017-int and a 14% lead in SPEC2017-fp , thats with DDR5 to boot.

They will be able to scale the this new core for several generations - THEY ALREADY HAVE IT, while AMD is at the end of the road for Zen architecture. AMD is not guaranteed to succeed with Zen5 architecture, they might get delayed, might take a page too many from Bulldozer or K10 "success" stories. They are sure executing real good lately, but local circlejerkers were already overjoyed by "massive IPC increase of Zen4", so while unlikely it might happen.
The reality is that Zen3 is currently loosing in PPC to ADL very substantially and mediocre PPC gains ensure that Raptor Lake will also be in the lead versus Zen4. I'd be damned if AMD was not more efficient with full node advantage, but their days of performance leadership over 2013 year designs in 2020 are over.

Intel's future will be decided by their process, if they can come up with Intel 4 and move to Intel 3 on time, they will be just fine.

what a tie in SPEC2017-int and a 14% lead in SPEC2017-fp , thats with DDR5 to boot.

Common, even Anandtech with their JEDEC loving, but not for real world testing found:

The increases over Rocket Lake come in at +18-20%, and Intel’s advantage over AMD is now at 6.4% and 16.1% depending on the suite, maybe closer than what Intel would have liked given V-cache variants of Zen3 are just a few months away.
Again, the E-core performance of ADL is impressive, while not extraordinary ahead in the FP suite, they can match the performance of some middle-stack Zen2 CPUs from only a couple of years ago in the integer suite.

Hard to imagine Intel not continuing to iterate on these things, they sure plan to use 32 E-Cores to make AMDs day miserable in throughput tests. The power "efficiency" of core pushed to 4Ghz by marketing on 10nm might not apply to chip on Intel 3 at correct voltage sipping power.

 

[inf64]

The reality is that Zen3 is currently loosing in PPC to ADL very substantially and mediocre PPC gains ensure that Raptor Lake will also be in the lead versus Zen4. I'd be damned if AMD was not more efficient with full node advantage, but their days of performance leadership over 2013 year designs in 2020 are over.

We must have different definitions of "substantially". If ~10% on average means substantially, then Zen4 should erase the lead and it's highly unlikely Raptor lake will be more than a few percent faster at the same clock. Perf./watt should be massive in AMD's favor everywhere, especially in servers where they will have 2x the core count with the same IPC and higher clocks. It will be bad for intel, real bad.

 

[DrMrLordX]

They will be able to scale the this new core for several generations - THEY ALREADY HAVE IT, while AMD is at the end of the road for Zen architecture.

Let's be realistic: execution-wise, AMD has done a better job than Intel over the last 5 years. If you don't like Zen4 then so be it. Don't bet too much on Intel fixing their manufacturing problems overnight, much less turning Golden Cove into a performance monster iteratively. From what I can tell, they'll update to Raptor Cove and then it's back to the drawing board, more-or-less, since Meteor Lake/Redwood Cove won't be prominent outside of mobile and may also be very low-volume. Arrow Lake may not entirely be clean-sheet, but I think Intel is going to have to make some big changes to keep up.

And of course that doesn't even bring enterprise into the picture. Zen4/Genoa will have a field day.

 

[JoeRambo]

We must have different definitions of "substantially". If ~10% on average means substantially, then Zen4 should erase the lead and it's highly unlikely Raptor lake will be more than a few percent faster at the same clock. Perf./watt should be massive in AMD's favor everywhere, especially in servers where they will have 2x the core count with the same IPC and higher clocks. It will be bad for intel, real bad.

Raptor Lake will be just several percent behind in IPC gain and just as good clocking. Efficiency of course is different thing due to process disadvantage that might last for some time. And in my post I was responding to ridiculous claims of gloom, that where in response in equally ridiculous pro-Intel post. It's in between obviously.

 

[Abwx]

Raptor Lake will be just several percent behind in IPC gain and just as good clocking. Efficiency of course is different thing due to process disadvantage that might last for some time. And in my post I was responding to ridiculous claims of gloom, that where in response in equally ridiculous pro-Intel post. It's in between obviously.

By Intel s own admission AMD will retain the perf/watt crown till 2024 at least, in the meantime they ll have to inflate TDP and inefficient E cores to keep up with their competitor s more risk averse approach.

Also nowhere in Intel s slide it is said that RPL will have better MT IPC, let alone in ST, but what seems sure is that they ll somewhat increase TDP to better feed the augmented E cores silicon.

Intel-Core-Roadmap: Vier neue „Lakes“ in drei bis vier Jahren

Auf je ein Hauptprodukt folgt ein Refresh – das ist der Intel-Core-Plan bis zum Jahr 2025 mit vielen „Lakes“.

www.computerbase.de

That being said Zen 5 will surely be largely inspired by Zen uarch, they just increased width valuably, that is, not marginaly.

 

[inf64]

Raptor Lake will be just several percent behind in IPC gain and just as good clocking. Efficiency of course is different thing due to process disadvantage that might last for some time. And in my post I was responding to ridiculous claims of gloom, that where in response in equally ridiculous pro-Intel post. It's in between obviously.

Do you have any slides/official info on IPC gains in Raptor Cove? I have seen none, maybe I missed it.

Edit:


Found this on the link Abwx posted:
"Raptor Lake - up to double digit performance boost
Based on performance projections from Intel on 02/17/2022 on SPECint_rate_2017 n Copy ICC 19u4comparing Raptor Lake 125W (8P+ 16 E )versus Core i9 12900K.Results may vary. "

So it's double digit improvement in MT (rate) benchmark. No surprise there. Zero information on IPC and ST improvements. If I had to guess, they might push the core to 5.8Ghz ST to milk one last few percent of performance no matter how much the core draws in such scenario. Add a few percent from more L2 cache and you get ~5-7% improvement in ST versus 12900KS. 12900KS scores 121% versus base 100% for 5950X according to Computer base, so 13900K could be in the 127-130 range.
AMD showed >5.5Ghz for clocks and if we assume they will hit similar clock speeds with ~10% IPC gain in desktop workloads, Zen4 could hit ~123-130 range, so basically neck even with 13900K. There goes the ST performance crown, the last advantage intel had.

 

[JoeRambo]

Do you have any slides/official info on IPC gains in Raptor Cove? I have seen none, maybe I missed it.

Was there any "official" information when I estimated 5-10% gain for Zen4 that turned out right on target? Intel is fixing the right things, speeding up DDR5 speeds and that will be ~5-7% PPC?
What they are doing in scaling performance s already doable somewhat with tweaking memory subsystem, except that 2MB of L2 cache thing that reduces reliance on it even further. And Alder Lake responds real nice to memory:

AMD showed >5.5Ghz for clocks and if we assume they will hit similar clock speeds with ~10% IPC gain in desktop workloads, Zen4 could hit ~123-130 range, so basically neck even with 13900K. There goes the ST performance crown, the last advantage intel had.

Easy to find outlier cases where one or the other will be ahead, but on average it will still be ~120 vs 127, with Intel still holding the ST lead. Throughput will probably be very similar too on average.

 

[Abwx]

Throughput will probably be very similar too on average.

That part is likely, just that one will use 250W while the other will be around 130W for the same throughput, so basically nothing will change from the current situation...

 

[inf64]

Easy to find outlier cases where one or the other will be ahead, but on average it will still be ~120 vs 127, with Intel still holding the ST lead. Throughput will probably be very similar too on average.

~5% higher ST performance is basically non-existent, whichever way it goes. MT should be a clear win for AMD IMO, but lets wait and see. Gaming, Zen4 should be faster but that's debatable as well. Finally, power and perf/watt will be massively in AMD's favor. So you have a chance to pick 16C high(est) performing cores with no scheduler issues on a superior node with superior performance/watt versus a chimera product with half of the chip being "E cores" that are way slower (think 0.5x) and having scheduler issues. I would always pick the first option, no matter if it was intel or AMD that made it.

edit: add on top AVX512 support

 

[JoeRambo]

That part is likely, just that one will use 250W while the other will be around 130W for the same throughput, so basically nothing will change from the current situation...

News at eleven? Marketing SKU using 250W again, when pushed. AMD is pushing their own CPU too, i doubt V/F curve is that generous and they need to match 24 cores now, not 16, so might be above 142W this time.
What I expect will change is current stalemate at normal SKUs like 12600K vs 5600x, where Alder Lake is somewhat faster and ~as power efficient. To Intel's credit they will simply use E-Cores to make low core count AMD chips non starters. Can they even afford a price war with 5nm stuff? They really needed Zen3 over Skylake lead in ST to command the premiums they need.

 

[JoeRambo]

~5% higher ST performance is basically non-existent, whichever way it goes. MT should be a clear win for AMD IMO, but lets wait and see. Gaming, Zen4 should be faster but that's debatable as well. Finally, power and perf/watt will be massively in AMD's favor. So you have a chance to pick 16C high(est) performing cores with no scheduler issues on a superior node with superior performance/watt versus a chimera product with half of the chip being "E cores" that are way slower (think 0.5x) and having scheduler issues.

I am picking 7950x, just like i bought 5950x and 3950x and 9980XE and dozens if not hundreds of AMD/Intel CPUs before, while typing this on 12900K with disabled E cores and undervolted/underclocked to static 5Ghz setup. Will i use 7950x as personal system? Probably not enough advance for me, still being chiplet based versus monolithic design from Intel. Will it fit wonderfully underclocked/undervolted in what it's siblings do? That's for sure.

 

[Timmah!]

Do you think AM5 socket will support other generation beyond 7000 series? I mean, if AMD hypothetically releases 24 core Zen4+ or Zen5 in a year or 2, do you think they will fit these 670 boards?
If yes, thats one more thing to consider when choosing between Raptor Lake and Zen4.

 

[jpiniero]

Do you think AM5 socket will support other generation beyond 7000 series? I mean, if AMD hypothetically releases 24 core Zen4+ or Zen5 in a year or 2, do you think they will fit these 670 boards?
If yes, thats one more thing to consider when choosing between Raptor Lake and Zen4.

Zen 4+/Zen 5 yes but what core counts it would have is still TBD.