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!

 

[biostud]

If they now say the 3D versions will come before the end of the year, is it not possible they may actually launch alongside the non v-cache versions? Or at very least be announced/revealed at the same time, with availability say month or 2 later than the vanilla versions?

If it can disrupt sales of Raptor Lake, they will mention it.

 

[Hans Gruber]

The good run of AMD is almost over. Enjoy Zen 4. It's not that AMD is failing. It's that Intel is improving generationally at a much faster rate. I am not saying AMD is returning to the decade of darkness and nothingness in CPU's. When Intel brings out Intel 4 aka. 7nm. That will bring Intel inline with AMD for power use. AMD can't keep up with Intel's IPC from generation to generation.

Send the hate. Intel was on 14nm for more than 6 years. Now they are on 10nm and after Raptor Lake they will be on 7nm. Ryzen had the fabrication advantage since Zen 2 over Intel. I should note that the power use for Intel to get the performance has not been practical. I think that will continue to some degree with Raptor Lake.

I will have a Zen 4 build of some kind within the next year.

 

[igor_kavinski]

Another fine pro-Intel post with intention to buy AMD again

 

[pakotlar]

Another fine pro-Intel post with intention to buy AMD again

I think his point is that people are over-estimating how behind Intel is. Known paychological bias: we tend to think that good situations (stocks, success, etc) will never get much worse, and that bad ones will never get much better. People suck at estimating non-linear situations.

 

[Mopetar]

Including the combined total of L1 and L2 caches it would be over 100 MB

That's a really stupid way to measure cache.

I will have a Zen 4 build of some kind within the next year.

Somehow you've just become less believable than Amber Heard.

 

[Hans Gruber]

That's a really stupid way to measure cache.



Somehow you've just become less believable than Amber Heard.

I have a Zen 2 build and a Zen 3 build. How can you compare me to ****?

Profanity is not acceptable in the tech forums, please don't do this again, not to mention the insults to a person, even if a celebrity. -AT Moderator Shmee

 

[Tuna-Fish]

That's a really stupid way to measure cache.

The L1 needs to be left out because it's inclusive, but there is an argument to be made that the correct way to measure Ryzen caches are L3+L2. Because that represents the maximum amount of data that can actually be cached. For an 8c Zen 4 X3D, that would be 104 MB (96+8*1).

 

[igor_kavinski]

I have a Zen 2 build and a Zen 3 build. How can you compare me to *AMBER HEARD*?

Man, edit your post ASAP unless you want mod hammer knocking sense into you.

 

[Schmide]

The good run of AMD is almost over. Enjoy Zen 4. It's not that AMD is failing. It's that Intel is improving generationally at a much faster rate. I am not saying AMD is returning to the decade of darkness and nothingness in CPU's. When Intel brings out Intel 4 aka. 7nm. That will bring Intel inline with AMD for power use. AMD can't keep up with Intel's IPC from generation to generation.

A trend of 1 is not a trend and there are so many caveats it's almost flat.

Have you ever traded penny stocks? Asking for a friend.

 

[RnR_au]

A tentative AM5 motherboard page is up at ASRock...

https://www.asrock.com/mb/AMD/X670E Taichi/index.asp

 

[CakeMonster]

A tentative AM5 motherboard page is up at ASRock...

https://www.asrock.com/mb/AMD/X670E Taichi/index.asp

I don't get the M2 setups. I'd rather have 2x M2 PCIE5 than 4x M2 out of which only 1 is PCIE5. Maybe my instincts are wrong but I'd be surprised if most prospective users need or prefer their setup.

 

[carrotmania]

The good run of AMD is almost over. Enjoy Zen 4. It's not that AMD is failing. It's that Intel is improving generationally at a much faster rate. I am not saying AMD is returning to the decade of darkness and nothingness in CPU's. When Intel brings out Intel 4 aka. 7nm. That will bring Intel inline with AMD for power use. AMD can't keep up with Intel's IPC from generation to generation.

Send the hate. Intel was on 14nm for more than 6 years. Now they are on 10nm and after Raptor Lake they will be on 7nm. Ryzen had the fabrication advantage since Zen 2 over Intel. I should note that the power use for Intel to get the performance has not been practical. I think that will continue to some degree with Raptor Lake.

I will have a Zen 4 build of some kind within the next year.

Come on man, I've already asked you nicely to leave this thread. You sound like a paid Intel mouthpiece to nearly everyone here, and you add nothing of value to this thread. Go spout your clap in the dedicated Intel threads, and let the civilised of us read and discuss in peace...

 

[John Carmack]

The good run of AMD is almost over. Enjoy Zen 4. It's not that AMD is failing. It's that Intel is improving generationally at a much faster rate. I am not saying AMD is returning to the decade of darkness and nothingness in CPU's. When Intel brings out Intel 4 aka. 7nm. That will bring Intel inline with AMD for power use. AMD can't keep up with Intel's IPC from generation to generation.

Send the hate. Intel was on 14nm for more than 6 years. Now they are on 10nm and after Raptor Lake they will be on 7nm. Ryzen had the fabrication advantage since Zen 2 over Intel. I should note that the power use for Intel to get the performance has not been practical. I think that will continue to some degree with Raptor Lake.

I will have a Zen 4 build of some kind within the next year.

I would like to remind you you're not on Wccftech.

 

[itsmydamnation]

The good run of AMD is almost over. Enjoy Zen 4. It's not that AMD is failing. It's that Intel is improving generationally at a much faster rate. I am not saying AMD is returning to the decade of darkness and nothingness in CPU's. When Intel brings out Intel 4 aka. 7nm. That will bring Intel inline with AMD for power use. AMD can't keep up with Intel's IPC from generation to generation.

Send the hate. Intel was on 14nm for more than 6 years. Now they are on 10nm and after Raptor Lake they will be on 7nm. Ryzen had the fabrication advantage since Zen 2 over Intel. I should note that the power use for Intel to get the performance has not been practical. I think that will continue to some degree with Raptor Lake.

I will have a Zen 4 build of some kind within the next year.

I find stuff like this dumb and funny, Intel have a massively bigger core in every area as well as bigger L1 and L2 cache and comparable L3. Yet they barely win and burn power like nothing else. This gets just more and more ridiculous when factoring in Server and Laptops. You sound like Pat , All tip and no iceberg.

A wide AMD core is coming, about the only thing to be guaranteed come from intel is another server delay.

 

[Schmide]

That's a really stupid way to measure cache.

Well back in the day AMD had an exclusive cache while intel had a victim. Memories

 

[biostud]

The good run of AMD is almost over. Enjoy Zen 4. It's not that AMD is failing. It's that Intel is improving generationally at a much faster rate. I am not saying AMD is returning to the decade of darkness and nothingness in CPU's. When Intel brings out Intel 4 aka. 7nm. That will bring Intel inline with AMD for power use. AMD can't keep up with Intel's IPC from generation to generation.

Send the hate. Intel was on 14nm for more than 6 years. Now they are on 10nm and after Raptor Lake they will be on 7nm. Ryzen had the fabrication advantage since Zen 2 over Intel. I should note that the power use for Intel to get the performance has not been practical. I think that will continue to some degree with Raptor Lake.

I will have a Zen 4 build of some kind within the next year.

And as long as both companies put competitive products on the market, we all win. The main reason for me not to like Intel is they used their market lead as a sleeping pillow sending out lackluster CPUs at high prices, and the need for new sockets/chipsets alll the time. But all markets need healthy competition, so it is good to see Intel getting their act together. But for the gaming crowd any of the modern CPUs will do fine, as the GPU will be the bottleneck most of the time anyway.

 

[jamescox]

AMD Ryzen 7000X3D Gaming CPUs with 100MB of Cache in Mass Production, Launching this Year [Rumor] | Hardware Times

Even better if they do a simultaneous AM4/AM5 launch for this new baby.

An AM4 version seems very unlikely. I don’t think they would do that unless the supply of DDR5 is

The L1 needs to be left out because it's inclusive, but there is an argument to be made that the correct way to measure Ryzen caches are L3+L2. Because that represents the maximum amount of data that can actually be cached. For an 8c Zen 4 X3D, that would be 104 MB (96+8*1).

If it is actually exclusive, then it makes some sense to add them together, but it obviously still causes confusion.

 

[itsmydamnation]

you can include it all, its cache , not total amount of unique data stored, how do you account for shadow tags etc with just l2 +l3.

 

[jamescox]

I find stuff like this dumb and funny, Intel have a massively bigger core in every area as well as bigger L1 and L2 cache and comparable L3. Yet they barely win and burn power like nothing else. This gets just more and more ridiculous when factoring in Server and Laptops. You sound like Pat , All tip and no iceberg.

A wide AMD core is coming, about the only thing to be guaranteed come from intel is another server delay.

Intel is strictly in a “I will believe it when I see it” category, so talk of the empire striking back is likely premature at best and obvious trolling in this thread. Also, Intel will likely not return to process tech dominance, so their ability to surpass other companies is limited. I suspect a lot of machines will actually go ARM instead anyway.

A lot of companies are using AMD CPUs and Nvidia cards since they have too much cuda code to switch. I am actually wondering if nvidia may be behind going forward, which I would not have expected. Intel isn’t even on the radar for most HPC machines. Some companies may port to ARM and go with Grace - Hopper, but I suspect that AMD’s MI300 might be available first and might perform better. The unified memory and bandwidth of an Integrated solution will mean that they have an advantage, although not all applications need that level of bandwidth. Nvidia’s ARM cores will likely be rather low performance per core compared to Zen 4, possibly with infinity cache and other special sauce.

AMD seems to be in the best position really. AMD has all of the components (CPUs, GPUs, FPGAs, etc) with good performance and will likely have a very modular way of combining them to make a wide array of products. It is unclear whether Intel CPUs will be competive in the server / HPC space any time soon. Perhaps, if they reduce power significantly they can compete with lower core count models, although Intel’s solution will likely be much more expensive to make. I don’t think they have anything in the pipeline to compete with the full 96 core Genoa. Their GPUs may be okay for compute even if they are terrible on games since they are so driver dependent. Intel also purchased Altera, so they have in house FPGAs. AMD acquired Xilinx, but they may have been working on integration even before that. We will see how Intel GPUs compare to AMD and Nvidia GPUs for compute. I am not expecting them to be that great for essentially the first generation. Nvidia has the GPU, but not really much else it seems? They have the cuda vendor lock-in, which might make the choice for many companies.

 

[Tuna-Fish]

I don't get the M2 setups. I'd rather have 2x M2 PCIE5 than 4x M2 out of which only 1 is PCIE5. Maybe my instincts are wrong but I'd be surprised if most prospective users need or prefer their setup.

The reason they are not offering the second PCIe 5.0 M.2 is that they have a Thunderbolt 4 controller. TB4 requires PCIe passthrough with a PCIe 5.0 link, meaning that the only place you can mount a TB4 controller in the system is right off the CPU. If you want one of those, you're only getting one 5.0 M.2.

 

[DrMrLordX]

I think his point is that people are over-estimating how behind Intel is.

So about Sapphire Rapids . . .

 

[Timmah!]

A tentative AM5 motherboard page is up at ASRock...

https://www.asrock.com/mb/AMD/X670E Taichi/index.asp

I dont understand the logic behind placing the second PCI-E slot only 3 slots apart, in a world where pretty much every GPU is going to be 3 slots high - there will be literally no space between the 2 GPUs, if someone will decide to go with 2. The upper slot should have been higher.

 

[igor_kavinski]

I dont understand the logic behind placing the second PCI-E slot only 3 slots apart

Not many people use a 2nd GPU. Those that will, they will get something more expensive and probably E-ATX. Then there's also PCIe extenders or riser cables. Most people are likely to leave the 2nd slot alone or at most, use it with an M.2 NVMe adapter.

 

[pakotlar]

So about Sapphire Rapids . . .

🤷‍♀️ If you want to believe that Intel is hopelessly behind and will never catch up, ok. The psychological bias in seeing current realities as immutable exists, it affects us all, I feel it too, but probably worth defending against with reason. What makes you certain that this is the time that AMD becomes immune to competition, and Intel unable to adapt?

 

[DrMrLordX]

🤷‍♀️ If you want to believe that Intel is hopelessly behind and will never catch up, ok. The psychological bias exists, it affects us all.

Belief doesn't come into play. What I know is that Intel is currently losing ground in the server room. The only thing saving Intel from Zen4 is that currently no one can replace Intel's 14nm volume. Not even Intel!