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!

 

[fkoehler]

I must be old, because its makes me both mad and ROTFLMAO when I hear people complain that its 'only 35%, and costs more"...

I guess Supply and Demand is no longer taught in schools.
Lets go back to the good old days where Intel precisely dribbled out 5-7% IPC improvements every year or two, and packed it good requiring a new mobo....

 

[pakotlar]

I must be old, because its makes me both mad and ROTFLMAO when I hear people complain that its 'only 35%, and costs more"...

I guess Supply and Demand is no longer taught in schools.
Lets go back to the good old days where Intel precisely dribbled out 5-7% IPC improvements every year or two, and packed it good requiring a new mobo....

You want to be powerless as a consumer because there is a single CPU supplier that dribbles out 5-7% annual upgrades?

 

[maddie]

I must be old, because its makes me both mad and ROTFLMAO when I hear people complain that its 'only 35%, and costs more"...

I guess Supply and Demand is no longer taught in schools.
Lets go back to the good old days where Intel precisely dribbled out 5-7% IPC improvements every year or two, and packed it good requiring a new mobo....

The good old days?

The real good old days were at worst, like today, and often times much better regarding annual IPC and clock advancement. If you think 5-7 % annual IPC were the good old days, then don't worry, you're still young.

 

[pakotlar]

The good old days?

The real good old days were at worst, like today, and often times much better regarding annual IPC and clock advancement. If you think 5-7 % annual IPC were the good old days, then don't worry, you're still young.

Agreed. My Celeron 300 that hit 413mhz and annual 2x upgrades on GPU were the reality until Hector Ruiz defrauded AMD

 

[biostud]

I must be old, because its makes me both mad and ROTFLMAO when I hear people complain that its 'only 35%, and costs more"...

I guess Supply and Demand is no longer taught in schools.
Lets go back to the good old days where Intel precisely dribbled out 5-7% IPC improvements every year or two, and packed it good requiring a new mobo....

That was the desert walk of the bulldozer years, where we got a glimpse into a world where only Intel existed.

In the good old days we hadn't hit the frequency ceiling, so not only did IPC increase, but so did frequency. Going from a Pentium 75Mhz to a Pentium II 300Mhz was something that mattered.

 

[Timorous]

You want to be powerless as a consumer because there is a single CPU supplier that dribbles out 5-7% annual upgrades?

The good old days?

The real good old days were at worst, like today, and often times much better regarding annual IPC and clock advancement. If you think 5-7 % annual IPC were the good old days, then don't worry, you're still young.

Did ya not read the sarcasm in that post? That post was dripping with it.

 

[Asterox]

Hm, compared to R9 5950X in some GPU test.By logic, they are probably based on the performance of only a few CPU cores. More or less, this shows Zen4 single core CPU performanse.

AMD Ryzen 7000 6-core CPU with 4.4 GHz clock and Gigabyte X670E motherboard spotted together - VideoCardz.com

AMD Ryzen 5 7600X? An engineering sample of Ryzen 7000 CPU makes its first appearance in a BaseMark database. The result featuring AMD processor with an OPN code of “100-000000593-20_Y” has been discovered by hardware leaker @TUM_APISAK. This product code has not yet been seen before, so it is...

videocardz.com

 

[eek2121]

The consensus is that the higher frequency, is the main reason for expected higher power draw, but could it mainly be limited to situations where something like AVX512 is in use?

The higher power limit helps multicore boost. As an example, a worst-case workload of prime95 small FFTs at stock drags my 5950x down below the base clock. 1.9 ghz I believe it ended up at. By enabling PBO, I was able to get up to 3.8 ghz. The CPU used around 230W.

Nobody knows what kind of power consumption we can expect from AVX-512 from AMD. There is a chance that it doesn't consume any additional power at all, and doesn't hurt clocks.

 

[eek2121]

In basemark, the 7600x is claimed to be faster than the 5950x: AMD Ryzen 7000 6-core CPU with 4.4 GHz clock and Gigabyte X670E motherboard spotted together - VideoCardz.com

Anyone know how good of a benchmark basemark is?

Also, I would ignore the clocks, chances are basemark is just detecting the base clock *cough*

 

[Det0x]

In basemark, the 7600x is claimed to be faster than the 5950x: AMD Ryzen 7000 6-core CPU with 4.4 GHz clock and Gigabyte X670E motherboard spotted together - VideoCardz.com

Anyone know how good of a benchmark basemark is?

Also, I would ignore the clocks, chances are basemark is just detecting the base clock *cough*

Think that benchmark is all GPU.. Here are my old basemark highscore from January: (3090 @ 563watts and in DX12 mode)


*edit*
Here are a new 5950x run with daily 24/7 settings in openGL mode

 

[nicalandia]

In basemark, the 7600x is claimed to be faster than the 5950x: AMD Ryzen 7000 6-core CPU with 4.4 GHz clock and Gigabyte X670E motherboard spotted together - VideoCardz.com

Anyone know how good of a benchmark basemark is?

Also, I would ignore the clocks, chances are basemark is just detecting the base clock *cough*

If it ain't Cinebench R23 is Worthless

 

[Det0x]

Think that benchmark is all GPU.. Here are my old basemark highscore from January: (3090 @ 563watts and in DX12 mode)
View attachment 64501

*edit*
Here are a new 5950x run with daily 24/7 settings in openGL mode
View attachment 64503

But we can learn 1 thing from this, the reported "CPU Speed" is the baseclock --> This Zen4 6core sample had a baseclock of 4.4 ghz

 

[gruffi]

In the good old days we hadn't hit the frequency ceiling, so not only did IPC increase, but so did frequency. Going from a Pentium 75Mhz to a Pentium II 300Mhz was something that mattered.

In the "good old days" you didn't even need a new CPU to increase performance by 50%. You just needed to press the turbo button on my 286 system, going from 8 to 12 MHz. 😂

 

[Abwx]

But we can learn 1 thing from this, the reported "CPU Speed" is the baseclock --> This Zen4 6core sample had a baseclock of 4.4 ghz
View attachment 64505View attachment 64506

4.4GHz base despite AVX512 ?

Unless those instructions are not enabled for some SKUs that would be boldly technically speaking.

 

[nicalandia]

4.4GHz base despite AVX512 ?

Unless those instructions are not enabled for some SKUs that would be boldly technically speaking.

Official baseline never counts AVX512 Workloads, Intel never does that.

 

[Timmah!]

Official baseline never counts AVX512 Workloads, Intel never does that.

This is AMD though.

 

[biostud]

In the "good old days" you didn't even need a new CPU to increase performance by 50%. You just needed to press the turbo button on my 286 system, going from 8 to 12 MHz. 😂

We had a 386sx 16/25Mhz with turbo button

 

[deasd]

Are there any technical reason having high base clock? 4.4Ghz is the highest base clock i ever seen. And it seems this is not AVX512 clocks though. If this base clock is all due to process enhancement, all core turbo might be easily surpass 5Ghz even for 6C12T.......

 

[inf64]

Are there any technical reason having high base clock? 4.4Ghz is the highest base clock i ever seen. And it seems this is not AVX512 clocks though. If this base clock is all due to process enhancement, all core turbo might be easily surpass 5Ghz even for 6C12T.......

Well they showed 16C ES running a game where some threads hit 5.5Ghz. So all core turbo of ~5-5.2Ghz for 6C part is very possible.

 

[nicalandia]

Are there any technical reason having high base clock? 4.4Ghz is the highest base clock i ever seen. And it seems this is not AVX512 clocks though. If this base clock is all due to process enhancement, all core turbo might be easily surpass 5Ghz even for 6C12T.......

All core turbo boost for the 16C/32T is expected to be about 5.2 Ghz that is how they can achieve +35% MT Performance boost

 

[fkoehler]

You want to be powerless as a consumer because there is a single CPU supplier that dribbles out 5-7% annual upgrades?

I guess I need to spell it out and say 'Whoosh"....

We're looking at an 7xxx maxed out marchitecture with:
PCIe 5
DDR 5
AVX512 (fwiw)
+/- 5.5Ghz
Built-in RDNA 2 Gfx
>15% gain in single-threaded work
>35% overall performance gain (multi-threaded workloads)
>25% performance-per-watt gains

And this will likely be at a price Intel was only 5 years ago charging for its upper i7 systems....

IIRC I was doing the pencil trick on my Celeron 300a in an Abit BP6.
And yes, this moaning and groaning that AMD is somehow slacking off and acting like Intel of yore vis-a-vis milking it, is both maddening and laughable.

At least Timorous got it.

 

[fkoehler]

Deleted

 

[Timmah!]

I guess I need to spell it out and say 'Whoosh"....

We're looking at an 7xxx maxed out marchitecture with:
PCIe 5
DDR 5
AVX512 (fwiw)
+/- 5.5Ghz
Built-in RDNA 2 Gfx
>15% gain in single-threaded work
>35% overall performance gain (multi-threaded workloads)
>25% performance-per-watt gains

And this will likely be at a price Intel was only 5 years ago charging for its upper i7 systems....

IIRC I was doing the pencil trick on my Celeron 300a in an Abit BP6.
And yes, this moaning and groaning that AMD is somehow slacking off and acting like Intel of yore vis-a-vis milking it, is both maddening and laughable.

At least Timorous got it.

Well AMD has still competition from Intel in the client segment, since Raptor Lake, performance-wise, will no doubt match Zen4. Thus they still have to try and arent indeed nowhere near sa bad as Intel was in 2009-2016 era. That said, they are for sure offering nothing “extra”, like they did with first gen 8-core Ryzen for 300 bucks - nowadays its all about matching Intel offering. If Raptor Lake was 32 E-cores instead of 16, you can bet we would see 24 core Zen4. Since they dont, tough luck - AMD is in no need to force the progress anymore, as they have the upper hand now.

I think this is what people point at and i dont see why that should be laughable nor maddening.

Additionally, they clearly showed they can be every-bit as bad as Intel, in areas where they have currently no competition - look no further than the whole Threadripper situation. God forbid Intel somehow fell-off in comsumer market as well (not likely, as they can just keep adding E-cores to keep up for forseeable future) - we would be stuck at 16 cores max for next 10 years.

 

[Ajay]

Nobody knows what kind of power consumption we can expect from AVX-512 from AMD. There is a chance that it doesn't consume any additional power at all, and doesn't hurt clocks.

Well, there is not way to tell. There is no comparative since this will be AMD's first AVX-512 implementation.

 

[Saylick]

Not sure if this is a guess or he knows when Raphael-X goes into production but...

https://twitter.com/x/status/1548319110563434498