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!

 

[Carfax83]

It's not just spec, it's Geekbench as well - Dr Cutress mentioned that Geekbench was a good IPC measuring tool (for general purpose IPC). Zen 4 and Raptor Cove are basically neck and neck in GB5(even with cryptography excluded; intel part runs at 100Mhz higher clock in ST test): https://www.anandtech.com/show/17601/intel-core-i9-13900k-and-i5-13600k-review/12

If there's one thing I've learned over the years of being into PC hardware, is that IPC is really difficult to pin down and quantify properly. It's not clear cut in the least and can differ from app to app. Case in point, I remember when Alder Lake launched last year and there were several people saying how it was only 6-7% or so higher IPC than Zen 3 based on Spec........but when you looked at the actual benchmarks, I was like

In some types of applications like code compilation and encoding, Alder Lake had a high double digit lead over Zen 3. Raptor Lake seems to continue that trend, because when you look at the Phoronix review, the 13900K beats the 7950x in ALL of the code compilation benchmarks. Now to be fair, it does this at significantly higher power draw than the 7950x, but the fact that it can even do that in a hybrid configuration vs a full fledged 16 core CPU with a process node advantage is mightily impressive from a raw performance standpoint.

Seeing stuff like that makes me doubt Spec and Geekbench. The encoding benchmarks are similar, but not as pronounced as with the code compilation. I'm just going by what I see in real world applications, and they tend to show that Raptor Cove has more low end grunt than Zen 4 core for core in heavy CPU workloads. Rendering is probably a bit of an exception as rendering scales so much better with core count than code compilation and encoding so Zen 4 can really stretch its legs.

 

[Carfax83]

Right, so when the game favors Intel it's because the game really taxes the CPU, but when it favors AMD it has to be a glitch.

Didn't we just spend the last few pages debating this very thing from the other side's perspective? See how easy it flips when the shoe is on the other foot and now CapFrameX is being accused of bias for just pointing out a discrepancy with the performance of Zen 4 compared to Raptor Lake in one of the most CPU bound areas in the game.

But at least this one has an explanation of sort, unlike with HZD.

 

[Joe NYC]

Probably not happening.

Intel could increase the L3 size on Meteor Lake, since CPU chiplet is separated.

It seems (from latest MLID video) that Intel is rethinking E-Core spam.

If that's the case, it is possible that Intel figured out that helping P-Cores with more L3 cache leads to better results than more E-Core spam.

 

[In2Photos]

Didn't we just spend the last few pages debating this very thing from the other side's perspective? See how easy it flips when the shoe is on the other foot and now CapFrameX is being accused of bias for just pointing out a discrepancy with the performance of Zen 4 compared to Raptor Lake in one of the most CPU bound areas in the game.

But at least this one has an explanation of sort, unlike with HZD.

I never responded to anything CapFrameX said (and never have). He's not here to defend himself. I only responded to what you said.

 

[MarkPost]

If there's one thing I've learned over the years of being into PC hardware, is that IPC is really difficult to pin down and quantify properly. It's not clear cut in the least and can differ from app to app. Case in point, I remember when Alder Lake launched last year and there were several people saying how it was only 6-7% or so higher IPC than Zen 3 based on Spec........but when you looked at the actual benchmarks, I was like

In some types of applications like code compilation and encoding, Alder Lake had a high double digit lead over Zen 3. Raptor Lake seems to continue that trend, because when you look at the Phoronix review, the 13900K beats the 7950x in ALL of the code compilation benchmarks. Now to be fair, it does this at significantly higher power draw than the 7950x, but the fact that it can even do that in a hybrid configuration vs a full fledged 16 core CPU with a process node advantage is mightily impressive from a raw performance standpoint.

Seeing stuff like that makes me doubt Spec and Geekbench. The encoding benchmarks are similar, but not as pronounced as with the code compilation. I'm just going by what I see in real world applications, and they tend to show that Raptor Cove has more low end grunt than Zen 4 core for core in heavy CPU workloads. Rendering is probably a bit of an exception as rendering scales so much better with core count than code compilation and encoding so Zen 4 can really stretch its legs.

I dont think so. Windows 11 (5950X @4.5; 12900KF @5.0/4.0):

Unreal Engine (VS2019-MSBuild), 5950X 28% faster


Chromium (Ninja), 5950X 14% faster


LibreOffice (Make), 5950X 35% faster


LuxCoreRender (CMake), 5950X 6% faster


Firefox (Clang++), 5950X 6% faster


Blender (CMake), 5950X 14% faster

https://imgur.com/guWN379

https://imgur.com/NqxCt16

 

[MadRat]

These CPU benchmarks basically show that most games are not wriiten with future hardware in mind, but rather push GPU loads to their maximum. CPUs should scale regardless of GPU in some way. It would be interesting to see them play head to head with AIs at resolutions that never really max out the GPU. Let the champion CPU prove it with brilliant micromanagement.

 

[inf64]

If there's one thing I've learned over the years of being into PC hardware, is that IPC is really difficult to pin down and quantify properly. It's not clear cut in the least and can differ from app to app. Case in point, I remember when Alder Lake launched last year and there were several people saying how it was only 6-7% or so higher IPC than Zen 3 based on Spec........but when you looked at the actual benchmarks, I was like

In some types of applications like code compilation and encoding, Alder Lake had a high double digit lead over Zen 3. Raptor Lake seems to continue that trend, because when you look at the Phoronix review, the 13900K beats the 7950x in ALL of the code compilation benchmarks. Now to be fair, it does this at significantly higher power draw than the 7950x, but the fact that it can even do that in a hybrid configuration vs a full fledged 16 core CPU with a process node advantage is mightily impressive from a raw performance standpoint.

Seeing stuff like that makes me doubt Spec and Geekbench. The encoding benchmarks are similar, but not as pronounced as with the code compilation. I'm just going by what I see in real world applications, and they tend to show that Raptor Cove has more low end grunt than Zen 4 core for core in heavy CPU workloads. Rendering is probably a bit of an exception as rendering scales so much better with core count than code compilation and encoding so Zen 4 can really stretch its legs.

Alright here is the Expreview's composite ST index :https://topic.expreview.com/CPU/# (they used a wide range of desktop workloads)
13900K @ 5.8Ghz is 4.7% faster than 7950X @ 5.7Ghz. (which equates to ~3% IPC difference and confirms what the SPEC shows : neck and neck integer performance and slightly higher fp performance for Raptor Cove). This confirms what Geekbench and SPEC show, almost to the percentage point. Sure, you will find outliers for both chips, when the difference can be double digit for specific workloads, but it all evens out in the end.

We cannot ignore the elephant in the room : from a perf/mm^2, even when adjusting for process node disadvantage, Zen 4 is way more efficiently built chip with less imbalances overall. AMD will have a HUGE headroom once they expand the frontend with Zen 5, and potentially double the L/S subsystem, along with more execution ports. That's why the 40-50% higher IPC target versus Zen 3 is most likely spot on (just like SR->Zen 1 and Zen 1-> Zen 3).

The only true advantage intel has is the lower latency due to monolithic design of its current products. This will become a non issue in time as AMD will utilize better/faster memory and incorporate new caches to mask the latency hit that chiplets bring.

 

[Markfw]

Alright here is the Expreview's composite ST index :https://topic.expreview.com/CPU/# (they used a wide range of desktop workloads)
13900K @ 5.8Ghz is 4.7% faster than 7950X @ 5.7Ghz. (which equates to ~3% IPC difference and confirms what the SPEC shows : neck and neck integer performance and slightly higher fp performance for Raptor Cove). This confirms what Geekbench and SPEC show, almost to the percentage point. Sure, you will find outliers for both chips, when the difference can be double digit for specific workloads, but it all evens out in the end.

We cannot ignore the elephant in the room : from a perf/mm^2, even when adjusting for process node disadvantage, Zen 4 is way more efficiently built chip with less imbalances overall. AMD will have a HUGE headroom once they expand the frontend with Zen 5, and potentially double the L/S subsystem, along with more execution ports. That's why the 40-50% higher IPC target versus Zen 3 is most likely spot on (just like SR->Zen 1 and Zen 1-> Zen 3).

The only true advantage intel has is the lower latency due to monolithic design of its current products. This will become a non issue in time as AMD will utilize better/faster memory and incorporate new caches to mask the latency hit that chiplets bring.

I agree to all of that... EXCEPT when the workloads are avx-512 Zen4 obviously wins easily. Also, for true multitasking workloads, especially those that last for hours or days, (DC type work) the 7950x will also win. For gaming, it would seem the small memory speed and latency issues VS Zen4 make the 13900k win by a few %

So, for average desktop use, it is slightly faster, and takes quite a bit more juice, and produces more heat.

 

[Carfax83]

I dont think so. Windows 11 (5950X @4.5; 12900KF @5.0/4.0):

That's Alder Lake with DDR4, so it's not the same. I appreciate the reference though. This is from TPU:

And you can check out Phoronix's results HERE for another point of reference which is much more exhaustive.

 

[Carfax83]

Alright here is the Expreview's composite ST index :https://topic.expreview.com/CPU/# (they used a wide range of desktop workloads)

That looks like mostly canned benchmarks, and it's hard for me to interpret even after using the translation function in the Edge browser. At any rate, this is a tough subject and there's going to be disagreements.

I tend to prefer Phoronix as they have one of the most exhaustive test batteries on the net for workloads that involve lots of compute power.

13900K @ 5.8Ghz is 4.7% faster than 7950X @ 5.7Ghz. (which equates to ~3% IPC difference and confirms what the SPEC shows : neck and neck integer performance and slightly higher fp performance for Raptor Cove). This confirms what Geekbench and SPEC show, almost to the percentage point. Sure, you will find outliers for both chips, when the difference can be double digit for specific workloads, but it all evens out in the end.

But the thing is that Raptor Lake uses a hybrid configuration. which has to be taken into account That automatically sets it at a disadvantage against a full bore CPU like Zen 4 in any heavy multithreaded workloads like rendering, code compilation etcetera, but Raptor Lake is still either highly competitive in those workloads or slightly beating the 7950x.

Let that sink in. A hybrid configuration with half the big cores of the 7950x is beating it in many of those types of workloads. I find that to be highly impressive, even after the higher power consumption because Raptor Lake to me is heavily disadvantaged and not really geared towards those workloads. The HEDT for Raptor Cove would be a much better comparison point when it launches.

We cannot ignore the elephant in the room : from a perf/mm^2, even when adjusting for process node disadvantage, Zen 4 is way more efficiently built chip with less imbalances overall. AMD will have a HUGE headroom once they expand the frontend with Zen 5, and potentially double the L/S subsystem, along with more execution ports. That's why the 40-50% higher IPC target versus Zen 3 is most likely spot on (just like SR->Zen 1 and Zen 1-> Zen 3).

Remember how disappointed this forum was when the grand proclamations about Zen 4's IPC was found to be false? I hope Zen 5 is a badass CPU, but 40-50% higher IPC seems outlandish.

The only true advantage intel has is the lower latency due to monolithic design of its current products. This will become a non issue in time as AMD will utilize better/faster memory and incorporate new caches to mask the latency hit that chiplets bring.

A strong AMD that can produce worthy competing products is necessary for the health of the tech industry and the betterment of consumers like us so I'm all for AMD bringing it to Intel. They need to step up in the GPU department though and nail Nvidia's ass to the wall.

 

[DrMrLordX]

Intel could increase the L3 size on Meteor Lake, since CPU chiplet is separated.

The only "hard" data we have at this point are die shots taken from a demo wafer of Meteor Lake. You can possibly extrapolate L3 size from those shots. In fact, some have already done so, and concluded that Redwood Cove looks an awful lot like an optical shrink of Golden Cove/Raptor Cove.

 

[alexruiz]

Dude, it's not rocket science. Reviewers do this stuff for a living, and AMD told them exactly how to configure it to get the best performance.

Yes, how very convenient. And I'm not saying that the issue is necessarily Tech Spot's testing methodology. The game itself may be bugged for all we know. How else do you explain how a 5950x beat a 13900K with DDR5?
This is all just basic stuff. I think you are severely underestimating the competence of most YouTube reviewers.

HWU is one of my preferred go to channels for reviews, along with Computerbase.de and a few others. I always thought they did a bang up job and I still do. But I think they do themself a disservice when they publish benchmarks that are nonsensical without additional information or context, because it makes people wonder if they made a mistake or worse, that they are biased because they choose to upload those graphs knowing they are misrepresentative.

Zen 4 is a 4 issue CPU, while Raptor Lake is a 6 issue CPU. Raptor Lake also has a bigger OoO window than Zen 4, 512 vs 320. Raptor Lake can also do 3x 256 bit loads per cycle vs 2 with Zen 4. These are just a few of the differences. I'm not saying that Zen 3 or Zen 4 is weak by any means, but Golden and Raptor Cove are just significantly wider with higher throughput.

Case in point, Zen 4 never trounced Alder Lake when it launched. Alder Lake got plenty of wins in despite having a lower clock speed than Zen 4.

Again, did you even analyze the numbers?
Did you even understand what I wrote ?
"There is little variance among reviewers for RPL numbers. There is huge variance for Zen 4"

If the tune up of the systems is "not rocket science", then why your preferred outlets have much lower numbers for Zen 4?
No, I am not underestimating the competence of most youtubers. If anything, I am giving them too much credit by calling them clueless as opposed to downright incompetent.
It is all about the clicks, and getting a review up faster than the competition gives them more clicks.
Again, why are they getting much lower numbers for Zen 4? It is just basic "not rocket science stuff", right?

Go and read the RPL review by techspot, check the Zen4 numbers
Now, go check one by the sites you shill for, compare the Zen 4 numbers
The numbers by techspot are much higher, as these guys did the basic "not rocket science stuff"

You were totally unable to answer me how the reviewers you shill for configured their Zen4 systems.
Here is answer for you:
"out of the box bios with drivers installed by windows from a master intel image"

ps. It is not one game that might be bugged, techspot showed Zen 4 ahead across the board. If their RPL numbers were lower, you could claim that was the reason. Their RPL numbers are the same as everyone else, but their Zen 4 are higher, so this is the key. However, hey are not the only ones getting higher Zen 4 numbers.

 

[Thunder 57]

That's Alder Lake with DDR4, so it's not the same. I appreciate the reference though. This is from TPU:

And you can check out Phoronix's results HERE for another point of reference which is much more exhaustive.

FWIW, I wouldn't trust TPU. They have shown their bias. It's not as obvious as userbenchmark, but it sure seems to be there. The 5800X3D goes from being ahead of the 12900K to 8% slower than it? Yea, OK.

 

[Yosar]

I'm not saying the 7600x can't be faster than the 13600K in gaming. Don't get me wrong. What I'm saying is that HWU's benchmarks for HZD and BF5 are clearly anomalous and need further looking into.

It's just those two benchmarks that look to be off.

Yep. Only those two benchmarks are off. And when buggy as hell CP2077 cripples AMD processors (famous SMT OFF) and GPUs it's all fine and dandy.

BTW People are more interested in BV than B2042. B2042 is almost totally non existent. It failed miserably and people returned to BV and BI. BV numbers are much more worthy for them now.
And anytime AMD processors perform subpar it's all AMD fault, but anytime intel processors perform subpar it's all developers fault, not intel.

Seeing stuff like that makes me doubt Spec and Geekbench. The encoding benchmarks are similar, but not as pronounced as with the code compilation. I'm just going by what I see in real world applications, and they tend to show that Raptor Cove has more low end grunt than Zen 4 core for core in heavy CPU workloads.

Oh yeah, all we needed in this thread is 'real world applications' reference and superior performance of current intel architecture in them.
Not like 'unreal world applications' in the form of HZD or BV.
I'm getting a suspicion who designed some of intel launch slides not so long ago.

 

[Kaluan]

So you're mad because Alder Lake was significantly faster than Zen 3 in Riftbreaker? Newsflash, Alder Lake was newer than Zen 3 (which had already been out for a year) and possessed the advantages of DDR5, higher clock speeds and higher IPC. You would have to be firmly in zealot territory to see anything wrong with that.

It's clear that Riftbreaker loves cache, clock speed and bandwidth and Alder Lake provided all three in healthy amounts which is why it performed so well on 12th gen. The 5800x 3D clawed back a lot of performance due to its humongous cache and is a whopping 35% faster than the 5800x in Riftbreaker for instance.



Optimizing in this case likely meant making sure the E cores weren't being used by the game, or using them for background tasks. Hitman 3 did the same thing. It had nothing to do with optimizing specifically for Intel architectures. As I said before, PC developers don't target particular CPUs because the PC is an open platform with tens of thousands of permutations, or millions if you count non gaming PCs.

And this is shown by how much better Zen 4 did compared to Zen 3 without any specific optimization on the developer's part for Zen 4, because Zen 4 has higher IPC, clock speeds, more cache and more bandwidth; all the things that the game loves.



I've been PC gaming for over 20 years now and I know a decent amount about how 3D engines work and how they leverage hardware. When I first saw Riftbreaker benchmarks, I did some research to see why it performed so well on Alder Lake in comparison to not just Zen 3, but other CPUs.

Seeing the game in action told me most of what I needed to know. The game is actually very technically sophisticated and supports RT shadows and ambient occlusion, as well as a sophisticated physics engine for destruction, particle effects and vegetation deformation. The game also features thousands of entities on screen at once.

All these things together makes the game very CPU demanding, so CPUs that can process and dispatch more instructions have the advantage. Alder Lake possessed these qualities in greater measure comparative to Zen 3, as it's a much wider architecture with more instruction throughput and bandwidth. These qualities were sustained for Raptor Lake, as Raptor Lake is also faster than Zen 4 in Riftbreaker for the same reasons.

In fact, the 13900K is still being GPU bottlenecked by an RTX 4090 at 1080p in this game.



Yes, and I always found that strange. I expect things like that to happen with graphics cards but not with CPUs because developers target x86-64 for CPUs and the CPUs using their internal OoO logic and microarchitecture determine how fast those instructions are processed. On paper, Alder Lake has a distinct advantage over Zen 3 in all the performance categories ie clock speed, IPC, instruction throughput, bandwidth etcetera.

It should theoretically blow the doors off of Zen 3 in HZD. It's also weird that in their 12900KS review, Alder Lake is in front of Zen 3, but in the 13900K review, it's behind Zen 3.

Try explaining that one:

What you call nitpicking I call asking questions. Can you explain the above graph, where the 12900KS is in front of the 5950x, but in the 13900K review (a faster CPU than the 12900KS), it's behind the 5950x?

Perhaps there are sections in the game that perform better on Intel or on AMD.

Shockingly, I'm starting off again about how you don't really seem to read or understand what I say...

Literally explained in my reply how running a faster GPU can obviously exacerbate performance deltas between CPUs. And guess what? 4090 is quite a bit faster than 3090Ti.

Bonus: Game still received at least 1 update this year, funnily enough it addressed a VRAM overflow issue that affected performance when changing resolutions (guess what CPU/GPU reviewers do a lot of...).
Second, the plethora of GPU driver changes in the last year, notable ones being AMD's May performance boosting driver and nVidia's 4090 launch driver and post-4090 review day performance boosting v521 and 522. Which the 3090Ti in the 12900KS review clearly wasn't using.
Third, also running DDR4-3600 DR CL14 in the 13900K review vs 3200 DR CL14 in the 12900KS review.

And no, don't think "sections" of the game are a thing, they use (like everyone else) the 3 minute long in-game benchmark tool.

But eh, who cares right? None of them are a silver bullet explanation of how "muh strong Raptor Cove can't possibly be slower than puny Zen3!". So let's just pretend that ranting over 4 pages, full tinfoil hat mode on, over literally nothing, is sane, cool and good.

Oh and the bit about The Riftbreaker is a doozy, it's like straw-manning and purposefully misunderstanding into one.

From Exor Studio's PR piece on Intel's gaming blog:

"The game's pretty well optimized for Intel CPUs," said Pawel, "and I guess it's a good fit for Intel's audience, because it's a strategy game for people who think."

Blue team arrogance in a nutshell.

Not that you'll actually read what I said this time either. lol

FWIW, I wouldn't trust TPU. They have shown their bias. It's not as obvious as userbenchmark, but it sure seems to be there. The 5800X3D goes from being ahead of the 12900K to 8% slower than it? Yea, OK.

They actually debunked themselves in their recent 12900K v 5800X and 5800X3D v 5800X roundups, if you cross-examine their data. Why not just cut the middle man (5800X)?

Their main review gaming performance charts often make no sense, both in retrospect (like when they showed Skylake still being faster than Zen3 on launch vs more modern data) as well as vs their other articles. But IMO this is present with many outlets, too lazy to put in the effort in more exhaustive testing or too unethical to actually properly inform the viewer/reader that their data may not be perfect/entirely representative or that they changed the methodology/suite in drastic ways. Glad they've started doing these big gaming round ups at least. The bigger the suite, the lesser faulty methodology or questionable ethics are a factor.

Either way, I'm leaning more towards incompetence than malice.
But the fact that they've yet to have a AMD CPU top a gaming chart, even tho they very well could have, a bunch of times, over the last 2 years, and that they're yet to do a versus were Intel loses (12900K v 5800X3D, hello?) is very strange.

Let me guess, their next one will be 12600K vs 13600K, or 7600X vs 13900K or any other random thing, as long as they don't show Ryzen lead of Intel's gens.

 

[Exist50]

The only "hard" data we have at this point are die shots taken from a demo wafer of Meteor Lake. You can possibly extrapolate L3 size from those shots. In fact, some have already done so, and concluded that Redwood Cove looks an awful lot like an optical shrink of Golden Cove/Raptor Cove.

This is extreme pedantry on my part, but that's not what "optical shrink" refers to. That would be something like N6 or N4.

 

[inf64]

But the thing is that Raptor Lake uses a hybrid configuration. which has to be taken into account That automatically sets it at a disadvantage against a full bore CPU like Zen 4 in any heavy multithreaded workloads like rendering, code compilation etcetera, but Raptor Lake is still either highly competitive in those workloads or slightly beating the 7950x.

Let that sink in. A hybrid configuration with half the big cores of the 7950x is beating it in many of those types of workloads. I find that to be highly impressive, even after the higher power consumption because Raptor Lake to me is heavily disadvantaged and not really geared towards those workloads. The HEDT for Raptor Cove would be a much better comparison point when it launches.

I was referring to their ST performance index, there is no hybrid "deficit" there... I even wrote the ST boost clocks which should have made it clear what workloads I was referring to.

Remember how disappointed this forum was when the grand proclamations about Zen 4's IPC was found to be false? I hope Zen 5 is a badass CPU, but 40-50% higher IPC seems outlandish.

The IPC jump I mentioned is projected versus the Zen 3 core (not Zen 4). So versus Zen 4, it should be 24-32% higher if history repeats.

 

[eek2121]

That looks like mostly canned benchmarks, and it's hard for me to interpret even after using the translation function in the Edge browser. At any rate, this is a tough subject and there's going to be disagreements.

I tend to prefer Phoronix as they have one of the most exhaustive test batteries on the net for workloads that involve lots of compute power.



But the thing is that Raptor Lake uses a hybrid configuration. which has to be taken into account That automatically sets it at a disadvantage against a full bore CPU like Zen 4 in any heavy multithreaded workloads like rendering, code compilation etcetera, but Raptor Lake is still either highly competitive in those workloads or slightly beating the 7950x.

Let that sink in. A hybrid configuration with half the big cores of the 7950x is beating it in many of those types of workloads. I find that to be highly impressive, even after the higher power consumption because Raptor Lake to me is heavily disadvantaged and not really geared towards those workloads. The HEDT for Raptor Cove would be a much better comparison point when it launches.



Remember how disappointed this forum was when the grand proclamations about Zen 4's IPC was found to be false? I hope Zen 5 is a badass CPU, but 40-50% higher IPC seems outlandish.



A strong AMD that can produce worthy competing products is necessary for the health of the tech industry and the betterment of consumers like us so I'm all for AMD bringing it to Intel. They need to step up in the GPU department though and nail Nvidia's ass to the wall.

Man, maybe someone was disappointed by Zen 4, but not me. Setting power limits to 142W, running benchmarks and productivity workloads, and seeing that massive performance uplift in a chip design that is largely unchanged since Zen 1 or Zen 2 is unreal!

 

[therealmongo]

Can we keep the Intel chat out of this thread named "Speculation: Zen 4 (EPYC 4 "Genoa", Ryzen 7000, etc.)" ?

Surprised a moderator has not stepped in to kick out the Intel brigade from here.

Just create a thread titled "Zen4 vs Alderlake" and post to your hearts content....

 

[nicalandia]

AMD EPYC 9374F Linux Benchmarks - Genoa's 32-Core High Frequency CPU

AMD EPYC 9374F Linux Benchmarks - Genoa's 32-Core High Frequency CPU Review - Phoronix

www.phoronix.com

That is quite impressive for 32 Cores

 

[nicalandia]

You would think that the 7950X and 13900X being above 5 Ghz CPUs with 32T would match a slower 32C/64T Genoa CPU.


But this monster just crushes them



32 Core Genoa vs 13900K and Xeon Platinum

"The EPYC 9374F had an average power consumption of 195 Watts and a recorded peak of 275 Watts "

 

[inf64]

Interesting comparison in AVX-512 workloads (crypto) between 11900K, 7700X and 12900K by Phoronix

11900K and 7700X numbers: https://www.phoronix.com/review/zen4-avx512-7700x/7
12900K P cores + AVX512 numbers: https://www.phoronix.com/review/alder-lake-avx512/2

Goes to show that Zen 4 has a superior AVX-512 implementation, even if it's "double pumped" as AMD calls it.

 

[nicalandia]

Goes to show that Zen 4 has a superior AVX-512 implementation, even if it's "double pumped" as AMD calls it.

Intel AVX-512 Design is Superior, but AMD AVX-512 design is better at executing the code giving the core resources. They were better at implementing that design with the strength and weakness of their core.

 

[DrMrLordX]

This is extreme pedantry on my part, but that's not what "optical shrink" refers to. That would be something like N6 or N4.

It really does. What the analysts were saying is that the core layout looked about the same. It was just on a smaller/more-dense process. Optical shrink. Point being, it doesn't appear as though any major layout changes were implemented moving to Redwood Cove. People expecting major IPC shifts from Redwood Cove are possibly not going to like what they get. Phoenix should be decent competition for it.

 

[jamescox]

Don’t know if this has already been discussed. I just noticed that the model listing on Wikipedia has 2, 3, and 6 chiplet devices listed. Milan was all 4 or 8. Rome had some 2 CCD parts.

I didn’t think we would go back to asymmetric configurations, so I only expected 4, 8, or 12 CCD devices. Perhaps they are optimizing for CCD count or just adding some market segmentation. A 3 CCD 24 core (3x8) seems rather odd compared to a 4 CCD 24 core (4x6). Not sure how the memory will be populated in such systems. Are they expecting many systems to not have the full 12 channels? Actually not sure what a 6 CCD device will look like. Is it 2 CCD in 2 quadrants plus 1 CCD in the other two? Would be really weird to have 2 quadrants with 3 CCD and two with 0 CCD.