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!

 

[JoeRambo]

I think Linux advantages are all about scheduler and fast syscalls and there are many in a task like compilation ( I/O, process, fork and what not). Windows has it stupid to put it mildly.

 

[gdansk]

https://twitter.com/x/status/1528924916363542529

Great post considering the obsession around IPC and benchmarks.

The point of benchmarks is to try to be representative of some real world workloads but in a consistent manner. They all fail to match a real world workload exactly because we each have our own. And SPEC tries to update it over time to avoid overspecialization of processors for their benchmarks.

If you know what you'll be running then test that instead. But this IPC discussion is an attempt to measure how much ILP those processors can exploit in non-ideal scenarios. You can argue if it should be normalized or if it should be conducted at iso frequency but the general trends do not change much. FireStorm > GC > Zen 3 > Skylake.

Zen 4 we have no idea because we only have the Blender times. And that's not enough to be estimating IPC for that single test. We need clock rates over time.

 

[Doug S]

Also 15% is fine. Really it would be very impressive if they can keep giving 15% each generation.

They definitely will not get 10% bump in frequency every generation. There's a reason why the ceiling has been in the 5 GHz range for long time now.

More likely you will see them do like Intel and a future generation that increases IPC will see a max clock below 5 GHz.

 

[coercitiv]

3DCenter identified the demand and delivered:

Some Blender benchmarks for Ryzen 9 5950X vs Core i9-12900K

pro AMD:
+2% Tweakers
+10% LesNum
+11% Igor
+13% CB
+24% PCGH

pro Intel:
+9% Comptoir
+5% TPU
+4% SweClockers
+4% Golem
+2% WCCF

Tom's Hardware:
+8% AMD (Barcelona)
+6% AMD (Koro)
+1% Intel (BMW27)
+8% AMD (Classroom)

 

[ondma]

Intel is disparaged because for a decade or more they held technology back, simply because they could and it made them more money. One big leap in performance, at the expense of heat, power and price does not cut it for most people. I know you can't understand that, and that leads you deride everyone as "fanatics", but take a look in the mirror. Intel are going back to the usual 5% increase with RPL.

We don't fully know where Zen4 lies, but its obvious that AMD, with a massively smaller RnD budget, are at least attempting new solutions and new technologies. ADL is still beaten in situations by regular Zen3, and X3D stomps the KS in lots of scenarios, so no, ADL is not "on another level". It came out a year later and is pushed to the maximum to win, sometimes.

Intel has a lot of animosity to undo among those of us who like technology and not just the company. All I see so far is Intels same old PR tactics, shout loudest and longest.

Yes, they did not advance performance as rapidly as they should have, but they still offered the best performance available. Anyone who though Intel was unethical was certainly free to purchase AMD.

 

[xilli_fiberbit]

Video from MSI which show Ryzen ES 16C (OPN : 100-000000665 ) in his AM5 socket :

the video has been made private on the MSI channel but downloaded by someone and re-uploaded

 

[gdansk]

Video from MSI which show Ryzen ES 16C in his AM5 socket :

the video has been made private on the MSI channel but downloaded by someone and re-uploaded

With that unusual cutout design I'm probably going back to a single dot in the center. Looks like a mess either way

 

[Markfw]

With that unusual cutout design I'm probably going back to a single dot in the center. Looks like a mess either way

Looks just like the way Intel does it, except a MUCH thicker IHS to keep even pressure on all the pins. I like it !

Edit: It would also distribute heat from the chip much better if it was the right alloy.

 

[gdansk]

Looks just like the way Intel does it, except a MUCH thicker IHS to keep even pressure on all the pins. I like it !

Sorry, I meant cleaning the paste out of the notches.

 

[igor_kavinski]

Yes same memory and all timmings are set to Auto. I've not tested disabling E-cores yet

Hope you test with E-cores disabled.

In the interest of research, you could tighten the timings on both systems and run the tests again. Should be interesting to see how far you can tune each system with identical RAM sticks without compromising stability.

 

[Markfw]

Sorry, I meant cleaning the paste out of the notches.

I make sure to use non-conductive paste, then who cares ?

 

[Hitman928]

yeah I set x264 threads manually (150% cpu threads) with little effect in general, specially if its at 1080p or lower res

Doesn't matter what you set the threads to if the software can't actually create a multi-threaded workload to utilize them. Also, yes, the higher the resolution the more it will be able to utilize additional cores/threads, but this is especially true for x265.

 

[igor_kavinski]

Video from MSI which show Ryzen ES 16C in his AM5 socket :

Awesome! Looks a lot more painless than ADL. Gotta love that IHS.

 

[igor_kavinski]

Dont know about Linux, never used it, just Windows.

Well, there's WSL2, if you feel inclined to experiment.

 

[Shivansps]

This was already posted here?

 

[jpiniero]

This was already posted here?

Yes

 

[nicalandia]

It would also distribute heat from the chip much better if it was the right alloy.

Aren't all IHS made out of Nickel Plated Copper/Copper alloy?

 

[Det0x]

Dont click if you dont want to watch video from Adored

Spoiler: AdoredTV: Zen 4 - DOA or Sandbagging?

 

[CakeMonster]

Thanks for the warning, certainly not letting him past my block.

Would we prefer to discuss the upcoming motherboards in this thread, or its own thread in that subforum? I can see arguments for either TBH. There is info coming out already and I'm particularly interested in the kind of PCIE5 support people want and for what price.

 

[Tuna-Fish]

Thanks for the warning, certainly not letting him past my block.

Would we prefer to discuss the upcoming motherboards in this thread, or its own thread in that subforum? I can see arguments for either TBH. There is info coming out already and I'm particularly interested in the kind of PCIE5 support people want and for what price.

Definitely a new thread. There's already too much noise in this thread as is. Do you want to make the thread? If yes, please post a link to this thread once done. It should also maybe be edited into the first post.

 

[deasd]

sounds like it's clocking higher than I thought... 5.2-5.5 on all thread is insane, speed demon design comes to my mind,,,,,

https://twitter.com/x/status/1529160979011051526

 

[CakeMonster]

I tend to agree. I'm terrible at maintaining stuff though so I'd prefer someone else make the thread so that it can be updated if needed.

 

[Schmide]

Dont click if you dont want to watch video from Adored

Spoiler: AdoredTV: Zen 4 - DOA or Sandbagging?

Quoting the silicon gang. The shark has been jumped.

 

[MarkPost]

3DCenter identified the demand and delivered:

this is with the new Blender Benchmark (it uses Blender v3.1.0) Blender - Open Data

5950X @4.5 vs 12900KF @5.0/4.0


monster scene, 5950X, 13% faster
junkshop scene, 5950X, 26% faster
classroom scene, 5950X 14% faster

 

[Markfw]

Aren't all IHS made out of Nickel Plated Copper/Copper alloy?

To be honest, I am not sure what the old "pressed" metal is, it is very thin. This is very thick, and could be made of a different alloy to help spread the heat better.