AMD Zen 2 Architecture: Socket AM4, 2019, Code-named "Matisse"

[TheoneandonlyMrK]

Perhaps you missed that, while it could hit 5GHz, it also consumed well north of 200 watts while doing so, giving it worse performance/watt then stock, while still getting its behind whooped by a stock i7?

As the previous poster pointed out, sure GloFo could do higher clocks, but it would eliminate the power advantage ryzen has, not to mention make the mobile market a hopeless endeavor AGAIN. I would ratherAMD continue improving IPC rather then getting into another disastrous clockspeed war.

I didn't miss your pady wtf. I have a 5ghz chip right here ,i know what it uses and i understand nodes.

The contradiction on forums is getting ridiculous, who mentioned mobile, not me im talking Hpc and how do you improve Ipc without arch changes on mass without frequency going up.
It's ok for intel to pr blurb 4.8ghz ocs with 5ghz via delid for its new coffee lake but Amd better not try for higher clocks too because you won't like it.
Do you not read,,,, that 8800k at 5ghz is not going to be power efficient.
Shit check the recent I9 world record , 1000watts plus through an Intel cpu.
Physics are what they are and intel play with the same rulebook.
. and all decrying 14nmhpe irrelevant because its used for mainframes, its not just for mainframes its a foundry node you could easily make x86 on it, they aren't and i started this by saying they could but are not.
But they could be using some of its pdk improvement recipe points on its 12nm lpp as to me that's whats implied by separate news items rumour and a bit of smoke.

 

[R0H1T]

That's why we can't agree , that statement is not entirely relevant. Like I said GF manufactured the FX chips on 32nm and those could hit 5 Ghz as well.

The fact that GF can make 5Ghz chip on their newest nodes doesn't tell us anything with regards to Zen+ or Zen 2. Can those do 5 Ghz on that node ? No way to tell , it's all 100% guesses at this point.

Those were FX 8350 OCed to the wall, besides the base clock wasn't 5GHz, it was turbo. Now anyone comparing that to the OCed Intel chips is also forgetting none of the Intel chips are sold as 5GHz monstrosities, not even with single core turbo.

We've also seen Ryzen being OCed to 5GHz or beyond with extreme cooling. So at this point in time we don't know whether Zen is suited to 5GHz, on water, or that this is purely GF's process limitation. The z14 btw is on a different HP process.

 

[Vya Domus]

My point is , you can't predict anything just by the manufacturing process.

 

[EarthDog]

Those IBM CPUs must cost a fortune.

not sure on the per unit cost. We had oracle systems in my data center and those machnes , 8 of them, cost 2.5M. Bht they were fully loaded. Engineered systems cost a ton.

My point is , you can't predict anything just by the manufacturing process.

or an unrelated architecture.

 

[R0H1T]

My point is , you can't predict anything just by the manufacturing process.

But then why are we talking about OCed Intel chips anyway? Can every chip clock upto 5GHz, perhaps 50% or is this pure silicon lottery?

 

[laszlo]

nice to see AM4 socket will remain in use.

actually we can predict to a point that new chips will be at least so fast as now clockwise and add 10-15% ipc improvement; tdp wise they shouldn't reach current one's but depend how good is this 12 nm process ...

just my 2c...

 

[Vya Domus]

Can every chip clock upto 5GHz, perhaps 50% or is this pure silicon lottery?

You are talking as if 1 in a million chips clocks up to infinity. It's not as random as you think , these chips are in fact engineered to reach those clocks and then binned. If a chip does reach 5 Ghz overclocked it's not that amazing , it was meant to reach those speeds just not optimized to do it consistently or in a cost effective manner. These chips don't reach those clocks out of "pure luck" they were designed to do it.

 

[ShurikN]

If that refreshed Zen+ could reach 4.3 (or make single core turbo higher), i'd call that a massive win for AMD. Would make a perfect gap filler to Zen2. Not to mention AM4, one socket, 3 generations (at least).

 

[R0H1T]

You are talking as if 1 in a million chips clocks up to infinity. It's not as random as you think , these chips are in fact engineered to reach those clocks and then binned. If a chip does reach 5 Ghz overclocked it's not that amazing , it was meant to reach those speeds just not optimized to do it consistently or in a cost effective manner. These chips don't reach those clock out of "pure luck".

So are Ryzen, they just need extreme cooling to get to that mark. A better comparison would be the mainstream top dogs, where 7700k enjoys at least half a GHz clock speed advantage over 1800x.

Once again OCing results are not guaranteed & when you say that it's highly disingenuous because KBL or SKLx aren't engineered to reach those clocks.

 

[Vya Domus]

So are Ryzen, they just need extreme cooling to get to that mark.

They clearly aren't , hitting 5 Ghz on LN2 doesn't count. Or rather if it does then we might just as well call even the Pentium 4 a 5 Ghz chip because, hey , that one reached 5 Ghz on LN2 too. I'm talking about the usual circumstances.

But what are we even talking about anymore ?

 

[R0H1T]

They clearly aren't , hitting 5 Ghz on LN2 doesn't count. Or rather if it does then we might just as well call even the Pentium 4 a 5 Ghz chip because, hey , that one reached 5 Ghz on LN2 too.

But what are we even talking about anymore ?

So why do you insist that KBL is engineered to reach 5GHz, I showed you 2 chips which were meant to do so even at stock!
If it's OCed & yet isn't guaranteed to reach 5GHz, what do you call that?

 

[EarthDog]

 

[Vya Domus]

So why do you insist that KBL is engineered to reach 5GHz, I showed you 2 chips which were meant to do so even at stock!
If it's OCed yet not guaranteed to reach 5GHz, what do you call that?

What ? When did I say that ?

I simply gave you an example of a series of CPUs built on a manufacturing process from 2010 , that can reach very high clock speeds , to make you understand that there are way more factors involved in this other than the node itself. Something you seem to not want to agree with , which is fine but this went on for far too long and now we're talking about things that have little to do with the original matter.

 

[R0H1T]

What ? When did I say that ?

I simply gave you an example of a series of CPUs built on a manufacturing process from 2010 , that can reach very high clock speeds , to make you understand that there are way more factors involved in this other than the node itself. Something you seem to not want to agree with , which is fine but this went on for far too long and now we're talking about things that have little to do with the original matter.

This was in response to ~

But then why are we talking about OCed Intel chips anyway? Can every chip clock upto 5GHz, perhaps 50% or is this pure silicon lottery?

You are talking as if 1 in a million chips clocks up to infinity. It's not as random as you think , these chips are in fact engineered to reach those clocks and then binned. If a chip does reach 5 Ghz overclocked it's not that amazing , it was meant to reach those speeds just not optimized to do it consistently or in a cost effective manner. These chips don't reach those clocks out of "pure luck" they were designed to do it.

I agreed with the second point on last page, the 32 nm was also SOI, probably developed in collaboration with IBM as well.

But then you insisted that Intel chips can be OCed to 5GHz, because they were meant to? This is not the case, unless we have different definitions of easy.

 

[Vya Domus]

Kaby Lake reaches 5 Ghz under pretty normal circumstance and consistently so , you're telling me that's just pure luck ?

Ryzen does not and there is no guarantee it will do on an improved node, though somehow you are convinced it will.

 

[R0H1T]

Kaby Lake reaches 5 Ghz under pretty normal circumstance and consistently so , you're telling me that's just pure luck ?

Ryzen does not and there is no guarantee it will do on an improved node, though somehow you are convinced it will.

No because ambient temps & cooling matter a lot, what's pure luck btw in this context?

 

[Vya Domus]

No because ambient temps & cooling matter a lot, what's pure luck btw in this context?

You do realize how much of a oxymoron you're claims are ? You think Intel's chips reach 5 Ghz out of pure luck but somehow you are convinced Zen+ will hit 5 Ghz guaranteed based on no real evidence.

OK , let's wrap this up according to you :

- being a engineer at Intel must be a really easy job , you get insane speeds out the chips you designed in a consistent manner just by luck !
- all that matters is the node , but not just any node , only the ones developed with IBM
- Zen+ and 5 Ghz basically confirmed

Got it. Anything else ?

 

[R0H1T]

You do realize how much of a oxymoron you're claims are ? You think Intel's chips reach 5 Ghz out of pure luck but somehow you are convinced Zen+ will hit 5 Ghz guaranteed based on no real evidence.

OK , let's wrap this up according to you :

- being a engineer at Intel must be a really easy job , you get insane speeds out the chips you designed in a consistent manner just by luck !
- all that matters is the node , but not just any node , only the ones developed with IBM
- Zen+ and 5 Ghz basically confirmed

Got it. Anything else ?

No because according to you Skylake was also engineered to reach 5GHz

So let's see what changed between SKL & KBL maybe that'll answer your own theory?

 

[Vya Domus]

No , I give up. You figured out this entire multi-billion dollar industry within seconds just by yourself. What can I say we need to step up our game.

 

[[XC] Oj101]

Ho. Lee. Crap. Why is everyone so hung up on frequency? It's only one factor of performance and MEANINGLESS without IPC.

 

[Vya Domus]

Ho. Lee. Crap. Why is everyone so hung up on frequency? It's only one factor of performance and MEANINGLESS without IPC.

Fancy big numbers look pretty.

 

[eidairaman1]

If AMD still wants to capture all those potential buyers of an 8700k, they had better push out some leaks regarding high clock speeds, fast.

That or Ryzen 1+(if they make it at all) which resolves a lot of teething issues from the first batch.

Vya Domus, you can write r01ht till you are blue in the face and fingers sore and it will still argue for arrogance's sake. Easier to walk away from situations like this.

 

[RejZoR]

Making chip run at 5+ GHz is not really a problem. Any chip can be designed to run at that speed and foundry just makes it. The thing is, Ryzen for example simply wasn't designed to run at clocks far beyond 4.1 GHz. It's not really a thing with the foundry. Partially yes, but to achieve high clocks, you need to design a pipeline in a certain way. Usually you need more pipeline stages. More stages can induce unwanted high latencies. So, vendors always try to find balance between clocks and number of pipeline stages. Going to far into one or another direction and you might end up with another Bulldozer.

They'll probably tinker with Ryzen a bit to achieve slightly higher clocks, probably at slight expense of consumption which they'll compensate with a newer node. So, in general it should clock a bit higher, but not consume any more power. Which is exacty what you want for an upgrade.

 

[R-T-B]

You do realize how much of a oxymoron you're claims are ? You think Intel's chips reach 5 Ghz out of pure luck but somehow you are convinced Zen+ will hit 5 Ghz guaranteed based on no real evidence.

I think his only real point is that GloFo has a process capable of doing higher clocks than the lpp Zen is using today.

I too believe that the LPP is holding Zen back.

 

[EarthDog]

I think his only real point is that GloFo has a process capable of doing higher clocks than the lpp Zen is using today.

Right.. which we are saying isn't relevant. Just as Huffy makes a bike that goes 50 MPH, that doesn't mean ALL bikes made by Huffy go 50 MPH... which is, seemingly, the point he keeps harping on. There are too many differences in uarch and other things which make that point, well, utterly useless.