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!

 

[Kocicak]

I am not sure why AMD does not offer 7800X3D NOW, when it consumes just one chiplet instead of two. It must be easier for them to make in larger numbers.

The milking strategy of forcing people to buy more expensive stuff even if they do not really need it is not nice at all, hope it is not their main motivation.

 

[Thibsie]

I am not sure why AMD does not offer 7800X3D NOW, when it consumes just one chiplet instead of two. It must be easier for them to make in larger numbers.

The milking strategy of forcing people to buy more expensive stuff even if they do not really need it is not nice at all, hope it is not their main motivation.

Forcing? Lol
Just wait.

 

[DAPUNISHER]

Wendell does a good job of talking about what it's like to use it. Instead of spitting out data that is only part of the equation, and does not necessarily reflect the user experience.

He used 7200 with the Intel, so it wasn't kneecapped either.

He talks about the 3D being able to do 4-5 days per second in late stage Stellaris too.

 

[DAPUNISHER]

Forcing? Lol
Just wait.

Nah bruh, we can't! AMD be over there like -

Spoiler: Seizure warning due to flashing lights

 

[Mopetar]

I am not sure why AMD does not offer 7800X3D NOW, when it consumes just one chiplet instead of two. It must be easier for them to make in larger numbers.

The milking strategy of forcing people to buy more expensive stuff even if they do not really need it is not nice at all, hope it is not their main motivation.

No one is forcing anyone to buy a 7950X3D any more than people are forced to buy a 7900 XTX or NVidia customers were forced to buy a 4090. It's been the typical state of any product launch to start with top-end halo products for the enthusiasts first and to move down the stack from there. This should not surprise anyone.

I don't know precisely how their production process works, but with a 6-core chiplet 7900X3D, it may well have been easier for them to launch the dual-chiplet products first given improved bin flexibility and being able to launch with less supply as there's less total demand for a $700 product.

 

[In2Photos]

Is it safe to say that all the heartburn from earlier this month about how AMD was going to handle v-cache on only one chip was a bit premature?

 

[A///]

I think it's coz Zen 4 chips boost quite aggressively, always banging their heads against the thermal limit.

Zen4 won't hit 95c at desktop browsing folders and opening up images as the claimants state.

 

[In2Photos]

Zen4 won't hit 95c at desktop browsing folders and opening up images as the claimants state.

Mine never gets anywhere near that. Granted I'm running PBO+CO with -30, but I also increased TDP. 72-75C max depending on ambient. Browsing probably hits mid 50s. But it idles around 45C.

 

[A///]

Mine never gets anywhere near that. Granted I'm running PBO+CO with -30, but I also increased TDP. 72-75C max depending on ambient. Browsing probably hits mid 50s. But it idles around 45C.

Can you disable the igpu and see what it performs like then? I'm not well versed on browsers but hardware acceleration may be using hte igpu instead of the dgpu for rendering. Which is the intended purpose because the igpu is a hardware based gpu.

 

[In2Photos]

Can you disable the igpu and see what it performs like then? I'm not well versed on browsers but hardware acceleration may be using hte igpu instead of the dgpu for rendering. Which is the intended purpose because the igpu is a hardware based gpu.

It's disabled already.

 

[biostud]

I am not sure why AMD does not offer 7800X3D NOW, when it consumes just one chiplet instead of two. It must be easier for them to make in larger numbers.

The milking strategy of forcing people to buy more expensive stuff even if they do not really need it is not nice at all, hope it is not their main motivation.

Several reasons:
1. If they can get people to buy this instead of the 7800X3D they will earn more money.
2. If people are waiting for the 7800X3D, they are not going to buy an Intel system in the meantime.
3. They can build up stock of 7800X3D processors, while no. 2.
4. They can improve drivers etc. for the 7800X3D launch.

As I see it the 7950X3D is not badly priced if you are a power user that can use all the cores and often use then and game as well. It is not as good as the 13900K, but with the lesser cost of running, lesser cost of cooling and a futureproof platform, I can see why some will choose it over the 13900K. Also, I think that the 7950X3D is a testplatform for future heterogenous AMD processors, where they can learn how to optimize drivers etc.

 

[A///]

Mine never gets anywhere near that. Granted I'm running PBO+CO with -30, but I also increased TDP. 72-75C max depending on ambient. Browsing probably hits mid 50s. But it idles around 45C.

You'll want to take ambient room temp, case temp and fan behavior into consideration. You're providing an anecdotal figure because no one knows what your house temp is like, how your fan setup is incl. curves and case temp.

I've heard of idles on aios around 5-6 deg. lower than you for example.

 

[leoneazzurro]

Several reasons:
1. If they can get people to buy this instead of the 7800X3D they will earn more money.
2. If people are waiting for the 7800X3D, they are not going to buy an Intel system in the meantime.
3. They can build up stock of 7800X3D processors, while no. 2.
4. They can improve drivers etc. for the 7800X3D launch.

As I see it the 7950X3D is not badly priced if you are a power user that can use all the cores and often use then and game as well. It is not as good as the 13900K, but with the lesser cost of running, lesser cost of cooling and a futureproof platform, I can see why some will choose it over the 13900K. Also, I think that the 7950X3D is a testplatform for future heterogenous AMD processors, where they can learn how to optimize drivers etc.

4. should be regarding only the boost behaviour because there is only one CCD so no problems in allocating threads on different CCDs.

 

[In2Photos]

You'll want to take ambient room temp, case temp and fan behavior into consideration. You're providing an anecdotal figure because no one knows what your house temp is like, how your fan setup is incl. curves and case temp.

I've heard of idles on aios around 5-6 deg. lower than you for example.

What were the ambient room temp, case temp, and fan behavior of those other setups?

 

[Heartbreaker]

Several reasons:
1. If they can get people to buy this instead of the 7800X3D they will earn more money.

That is THE reason.

This is a common product release tactic. Release the most expensive, highest margin product first. Then the (gotta have it now) early adopters have no choice but to buy the most expensive model.

If released together with the much more reasonably priced 7800X3d, MANY of those sales, would be the more reasonably priced (less profitable) version.

 

[burninatortech4]

Wendell does a good job of talking about what it's like to use it. Instead of spitting out data that is only part of the equation, and does not necessarily reflect the user experience.

He used 7200 with the Intel, so it wasn't kneecapped either.

He talks about the 3D being able to do 4-5 days per second in late stage Stellaris too.

Late game Stellaris on a 1000 star galaxy is an intense CPU workout. My 5400U can barely handle it. My 5950X is better but not as much as I'd expect.

 

[A///]

What were the ambient room temp, case temp, and fan behavior of those other setups?

65*, case about 80-85* and fan behavior was tied to the CPU utilization with a 8 second delay on the fans to avoid unncessary spinups. Case problems and a non-compute related temp probe with exterior facet.

 

[nicalandia]

In Linux(since most apps are well threaded) leaving things stock is recommended.



Performance of 7950X3D in Linux

I compiled that from the data that Phoronix just released, but they did not do show them combined like I did.

For example on the OpenVINO test, it favors AVX-512/Speed over Cache$

 

[eek2121]

You'll want to take ambient room temp, case temp and fan behavior into consideration. You're providing an anecdotal figure because no one knows what your house temp is like, how your fan setup is incl. curves and case temp.

I've heard of idles on aios around 5-6 deg. lower than you for example.

I have a small Mini ITX system (FormD T1) with an EVGA 3090 FTW 3 Ultra and a 7950X. I never see 95C in browsing or desktop workloads. Gaming takes it to 75-85c depending on the game. Only workloads involving compiling, rendering, and encoding cause it to hit 95C.

I am running stock with no PBO or CO, ambient is typically 72, idle temps in the 50s.

 

[Hitman928]

In Linux(since most apps are well threaded) leaving things stock is recommended.

View attachment 77398

Performance of 7950X3D in Linux

I compiled that from the data that Phoronix just released, but they did not do show them combined like I did.

For example on the OpenVINO test, it favors AVX-512/Speed over Cache$

View attachment 77399

How do you see the second result you posted as favoring Prefer Freq?

My understanding is that the Prefer freq/cache option does nothing in Linux as there is no Linux driver (yet) to make it work. So you can set it in the BIOS, but it doesn't actually do anything when running Linux, so you get whatever the automatic algorithm decides to give you.

 

[nicalandia]

How do you see the second result you posted as favoring Prefer Freq?

Because the 7950X non-X boost higher than the 7950X3D and the chart shows it(Is one of those charts where lower means better)

 

[Hitman928]

Because the 7950X non-X boost higher than the 7950X3D and the chart shows it(Is one of those charts where lower means better)

Ok. . .

I guess I'm just confused trying to connect the first half of your post with the second half as "leaving things stock" (in terms of CCD preference) is the only option on Linux.

 

[In2Photos]

65*, case about 80-85* and fan behavior was tied to the CPU utilization with a 8 second delay on the fans to avoid unncessary spinups. Case problems and a non-compute related temp probe with exterior facet.

So just 1 other setup? If I was saying that my system was doing something outlandish I could see why you'd want my testing methodology. I don't think that my findings were extraordinary. But here's my setup.


Lian Li 011Dynamic EVO with 9 Lian Li 120mm SL120. Bottom and side fans are intake, top is exhaust. Fans are configured in L-connect with the "quiet" curve, modified to stay around 800 RPM until the CPU reaches 70C at which point they start to climb to 1100RPM. This is when they start to become audible. I don't have any probes for thermal testing, but the ambient in my room is typically warmer than most as I have 3-4 PCs that can be on at any time, one of them running Folding at Home 24/7. So ambient typically is 21-24C (70-75F). As I stated before temps typically run in the low to mid 70C on the CPU during benchmarking. Gaming is typically upper 60s. Idle is near 45C. The reason for this is that the CPU package power at idle is near 38W with XMP enabled. With it disabled idle power is about 28W.

My daughter's 7600X system has similar results although she is using an air cooler and a Lian Li 216 case with only the stock fans (2-160 front intake, 1-140 rear exhaust). Her system also has PBO+CO (-25 all core) and XMP enabled. She idles around 23W, 40C with the CPU fan at 800 RPM and case fans around 700 RPM. She maxes out at ~80C running benchmarks. Games are low 70s. Ambient room temp for her is slightly lower than mine.

 

[leoneazzurro]

Allegedly first results for Dragon Ridge

AMD strikes back: Ryzen 9 7945HX beats Intel Core i9-13980HX despite much lower power consumption - NotebookCheck.net News

Did AMD just win the mobile CPU Game? - AMD Ryzen 9 7945HX + Zephyrus Duo 16 Hands-On - YouTube

Edit: added the video from the same source

 

[Hitman928]

Allegedly first results for Dragon Ridge

AMD strikes back: Ryzen 9 7945HX beats Intel Core i9-13980HX despite much lower power consumption - NotebookCheck.net News

Did AMD just win the mobile CPU Game? - AMD Ryzen 9 7945HX + Zephyrus Duo 16 Hands-On - YouTube

Edit: added the video from the same source

I think you can remove the “allegedly”.