Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)

[DisEnchantment]

Speculate at will

 

[A///]

Same as Zen4, so 32MB L3.

oops no I mean the x3d of zen 5.

 

[A///]

They have no Plan B compute dies on TSMC process? How is this possible? They were supposed to take their 10nm lessons to heart. Actually, I think they were REQUIRED to do that by their investors.

Of course they have plan b. because intel's design is likely a terrible outcome in the eyes of zen 5. fyi plan b is a day after contraception so women cannot get pregnant.

 

[Joe NYC]

I am not saying i would prefer slower 24C over faster 16C.
Just that i expect the 30 percent perf increase without the core count increase to be the same price as 7950x. If they want more money, they need to provide more on top of that - more cores would be one of the possibilities.
If AMD could sell 7950x for 699 (and then slash its price like 2 months after the launch), when 5950x was 799, i dont see how they could possibly picture getting away with 999 price for Zen5, if the rumored perf increase is more or less equal to perf increase between Zen 3 and 4.

I don't fully buy a substantial price increase.

The way I think it will play out, if Zen 5 is substantially faster than Zen 4 V-Cache and Intel ARL is disappointing is that AMD will price Zen 5 about the same as Zen 4 V-Cache (for introductory MSRP).

BTW, with Zen 4 pricing, I think AMD was compensating for the problems with the platform. With MSRP and various bundles and rebates. By the time Zen 5 is released, the problem with AM5 platform and cost of DDR5 will be the thing of the past. So, comparing prices with Zen 4 may not be the most appropriate.

Also, I don't know how much RPL-R will be, as far as Intel discounting. Intel will have a problem discounting ARL very much. Because, it will probably more costly to manufacture than Zen 5.

 

[A///]

amd dropped the price and compensated because the platform was neck in neck with intel. it wasn't until their price drops about 60 days later that they began seeing some movement on zen 4. I expect the parts to cost a lot because amd isn't stupid and they know people will pay that price if it looks like arrow lake is another intel dud .

 

[soresu]

and cost of DDR5 will be the thing of the past

DRAM costs are variable - please don't go jinxing it for us 😂

I got enough DDR4 to move to my AM4 Zen1 setup but had to wait 2 years to afford more than 32GB with the huge price hikes DRAM got just after that.

 

[yuri69]

Reading all those posts about the Intel dead, everything Intel DOA, AMD no competition, utter dominance reminds me good old times.

AMD Barcelona - the read quad core!
AMD Bulldozer - the revolutionary future-proofed multi-threaded CPU!
AMD Fusion - the seamlessly integrated GPU for uber-parallel performance!

 

[Timorous]

Reading all those posts about the Intel dead, everything Intel DOA, AMD no competition, utter dominance reminds me good old times.

AMD Barcelona - the read quad core!
AMD Bulldozer - the revolutionary future-proofed multi-threaded CPU!
AMD Fusion - the seamlessly integrated GPU for uber-parallel performance!

I think the difference is that AMD have an excellent record of execution with Zen. Aside from the K7 parts AMDs execution back the was very hit and miss

 

[Timmah!]

It's a bigger jump in ST than Zen4 and there's no competition.

How much bigger?

no competition is irrelevant, nobody is going to say hey, its 30/40 percent faster and costs 300 more than before, but intel has nothing, so i have to have this.

Not where it matters.

Who even said that.

By where it matters you mean ST i guess? So it was 20~25 percent in ST with Zen4, not 30, still completely miniscule difference.

Red Gaming Tech said that. They said high 2000 to be precise, so there is indeed some wiggle room, could be around 40.

 

[DisEnchantment]

This thread devolved in to AMD drum beating circle jerk in the last months.
No technical discussion needed, no debate, no sources, no citations, no disclaimers, because trust me bro intel dead. Sad. What's even more sad is long time members gaslighted to engage in such behaviour.





Looks At the title... AMD thread which is suposed to Talk about Zen...
Looks at thread... mostly AMD supporters in said thread...
Reads content, yep AMD Hype...
Reads your Comment....

You know your more then free to challenge or call out proof in anything you find false I suggest you do it instead of venting out on the entire thread and participants. The first is asking for support and proof, the second is near member call out which can lead to a infraction

Moderator Aigo

 

[adroc_thurston]

Reading all those posts about the Intel dead, everything Intel DOA, AMD no competition, utter dominance reminds me good old times

Yes, it's like 2006 but the vendors are flipped.

They said high 2000 to be precise, so there is indeed some wiggle room, could be around 40

Bingo!

This thread devolved in to AMD drum beating circle jerk in the last months

Well I mean, they won.
There are many to discuss how so you can try if you want.

because trust me bro intel dead

They're not dead, just poverty brand now.

 

[trexfromouterspace]

lmfao anyone who actually believes this circlejerk is going to be *incredibly* disappointed


This is getting even closer to member call out which is infractable... i suggest you all STOP unless you want to start seeing stars from me swinging the hammer soon...

Moderator Aigo

 

[APU_Fusion]

lmfao anyone who actually believes this circlejerk is going to be *incredibly* disappointed

How about provide facts instead of sarcasm why the circle jerkers are wrong? You feel better mocking people’s excitement? Says more about you then them

 

[adroc_thurston]

How about provide facts instead of sarcasm why the circle jerkers are wrong? You feel better mocking people’s excitement? Says more about you then them

meh it's so easy to bookmark posts here anyway so not like he's getting away scot free.

 

[TESKATLIPOKA]

Huh? 7950X was $699 at launch. As I said, a potential 24C Zen5 (whether there will be such a CPU or not) should be priced the same, i.e. $699 and not $999. And 16C Zen5 lower.

AMD had stayed on max 16C for standard desktop for far too long. They're going to be run over by Intel soon in MT perf if they do not increase core count. They need to bump core count just like they did with Zen (8C) and Zen2 (16C). AMD has stagnated since then.


Of course Zen5 have better perf than Zen4. Why release it otherwise?

But people are not prepared to pay the amount of money for it that you think. They'll buy Zen3/Zen4 or Intel instead. Also, Zen5 will soon after release be competing with Intel Arrow Lake (8P + 32E), which potentially will stomp on Zen5 in MT workloads.

8C16T Ryzen 7 1800X: $499 (100%)
16C32T Ryzen 9 3850X: $749 (150%)
16C32T Ryzen 9 5950X: $799 (160%)
16C32T Ryzen 9 7950X(X3D): $699 (140%) <- CPU is $100 cheaper, but platform cost went up

And now you want
24C48T Ryzen 9 8950X for only $699?
If It had 33% better IPC and 50% more cores you would have 2x higher performance.
Even If AMD released a 24C Zen5 chip there is no reason for It to have the same price as the previous generation.
On the other hand, $999 for the top 16C Zen5 chip is a lot, especially If It doesn't even have 3D cache. It would need to have 43% higher performance just to have the same perf/$ as the previous generation.
If Intel will be behind in performance next gen then I wouldn't be surprised for AMD to set prices a lot higher.

edit:
Just in theory: 16C32T Zen5 vs Arrow Lake 8P+32E
If Zen provided 30% higher IPC but Arrow only 10% and no clockspeed gains for both of them then yes, Arrow would offer better MT performance but only at >16T peaking at 48T. For 1T-16T It would lose to Zen5.
You would have to choose based on how many threads you would use.

 

[TESKATLIPOKA]

You're incorrectly assuming everyone just cares about ST perf and not MT perf.

Ok, so "everyone" only cares about 1T performance? Then why not release 8950X as a 1C CPU?

What he is talking about is not really single thread per se but IPC in case of Zen5.
Meaning that with 30% higher IPC you gain 30% more performance at 1-32T for a 16C32T Zen5 CPU If clockspeed stay comparable to Zen4.

If AMD releases an 8950X with just 16C at $999 or even $699, they're gonna have back track on that with their tail between their legs. Just like they had to with 7950X which was released at $699, but soon after release had to be dropped to $569 when market reality hit them in the face.

The problem with 7950X was that you had to invest in a completely new platform for AM5. You didn't need to buy just a CPU, but also motherboard, DDR5 which were really expensive at the time. Paying that much for only 25-30% higher performance was not really worth It.

 

[trexfromouterspace]

How about provide facts instead of sarcasm why the circle jerkers are wrong? You feel better mocking people’s excitement? Says more about you then them

No and no. Feel free to bring it up later if I'm wrong.

 

[maddie]

This thread devolved in to AMD drum beating circle jerk in the last months.
No technical discussion needed, no debate, no sources, no citations, no disclaimers, because trust me bro intel dead. Sad. What's even more sad is long time members gaslighted to engage in such behaviour.

I use these things to better understand human nature. Have to add that the past few yrs have been astounding.

 

[Fjodor2001]

8C16T Ryzen 7 1800X: $499 (100%)
16C32T Ryzen 9 3850X: $749 (150%)
16C32T Ryzen 9 5950X: $799 (160%)
16C32T Ryzen 9 7950X(X3D): $699 (140%) <- CPU is $100 cheaper, but platform cost went up

And now you want
24C48T Ryzen 9 8950X for only $699?
If It had 33% better IPC and 50% more cores you would have 2x higher performance.
Even If AMD released a 24C Zen5 chip there is no reason for It to have the same price as the previous generation.
On the other hand, $999 for the top 16C Zen5 chip is a lot, especially If It doesn't even have 3D cache. It would need to have 43% higher performance just to have the same perf/$ as the previous generation.
If Intel will be behind in performance next gen then I wouldn't be surprised for AMD to set prices a lot higher.

edit:
Just in theory: 16C32T Zen5 vs Arrow Lake 8P+32E
If Zen provided 30% higher IPC but Arrow only 10% and no clockspeed gains for both of them then yes, Arrow would offer better MT performance but only at >16T peaking at 48T. For 1T-16T It would lose to Zen5.
You would have to choose based on how many threads you would use.

2017:
4C8T Intel 7700K: $339 (100%)
8C16T AMD 1700: $329 (97%)

So with Zen1 in 2017 AMD gave us 2x amount of cores compared to the competition, for 3% lower price.

Now regarding 16C32T 7950X it was effectively launched at $569 (price dropped to that soon after release). If 8950X launches at $699 it'll be 23% higher price than for 7950X. So yes, I don't think it's unrealistic to expect 24C48T for 8950X then (if there actually will be a 24C48T Zen5 variant on AM5). If AMD did like in 2017, they would even double the cores to 32C64T and price it lower than $569 at the same time! But I do know that won't happen unfortunately.

 

[msj10]

lmfao anyone who actually believes this circlejerk is going to be *incredibly* disappointed

How dare you question the 40% IPC prophecy

 

[Fjodor2001]

What he is talking about is not really single thread per se but IPC in case of Zen5.
Meaning that with 30% higher IPC you gain 30% more performance at 1-32T for a 16C32T Zen5 CPU If clockspeed stay comparable to Zen4.

The problem with 7950X was that you had to invest in a completely new platform for AM5. You didn't need to buy just a CPU, but also motherboard, DDR5 which were really expensive at the time. Paying that much for only 25-30% higher performance was not really worth It.

No, he actually meant just 1 core. Did you not read his response?

It's a reused server CCD so they really could not [release an 1C variant] even if they wanted to.

So no reasoning of the type you mentioned. If he actually means MT performance is important too after all, he should say that.

And if MT performance is important, then core count matters as well. A lot. And then ARL-S will give Zen5 some real competition if Zen5 is only 16C, while ARL-S is 8P+32E. But with no official sources confirming any of this, it's all speculation at this point.

 

[HurleyBird]

If AMD could sell 7950x for 699 (and then slash its price like 2 months after the launch), when 5950x was 799, i dont see how they could possibly picture getting away with 999 price for Zen5, if the rumored perf increase is more or less equal to perf increase between Zen 3 and 4.

Honestly, I'm not sure why everyone is giving so much attention to this forum leaker. Besides the fact that the CPU forum has had something like a 0% success rate on previous forum leakers, the $1000 claim for 16 core Z5 is totally a red flag. Firstly because of how unlikely it sounds, and secondly because of the sheer amount of confidence he's projecting with this claim. Pricing is the most flexible, most fluid, most likely to be changed before launch. A real industry insider would know this and wouldn't harp on something that's so likely to leave egg on his face. At most he might say to expect a large price increase.

 

[HurleyBird]

Bergamo is not IFoP limited. It only uses 8 ports although the sIOD has 12 available.

Interesting, in that case maybe AMD has a 192-core version waiting in the wings. But (hate to be this guy) do you have a source on the 8 ports?

 

[adroc_thurston]

So with Zen1 in 2017 AMD gave us 2x amount of cores compared to the competition, for 3% lower price.

Yea slower cores.
Not the case here.

How dare you question the 40% IPC prophecy

Looks 32% to me but I digress.

And if MT performance is important

Cinemememarks are a very secondary thing, and that's all atoms are really good for.

while ARL-S is 8P+32E

This thing hasn't existed for years at this point.

Firstly because of how unlikely it sounds

Why even.

A real industry insider would know this and wouldn't harp on something that's so likely to leave egg on his face.

Products are bracketed in segments and priced into competitive response, to which Zen5 will have none of.

in that case maybe AMD has a 192-core version waiting in the wings.

No.
Genoa sIOD only has 8CCMs which can track cache misses out of 2 CCX each, which limits the design to 16CCX configs tops, which current Bergamo is.

But (hate to be this guy) do you have a source on the 8 ports?

AMD showed unlidded Bergamo package as is.

 

[TESKATLIPOKA]

2017:
4C8T Intel 7700K: $339 (100%)
8C16T AMD 1700: $329 (97%)

So with Zen1 in 2017 AMD gave us 2x amount of cores compared to the competition, for 3% lower price.

Nonsensical comparison, because we were talking about AMD products here.

And you even cherry-picked a much cheaper 8core just to It would look better in comparison and "forgot" about Coffee Lake.

4C8T i7-7700K -> release price: $339 and release date: January 2017
8C16T R7 1700 -> release price: $329 and release date: March 2017
8C16T R7 1800X -> release price: $499 and release date: March 2017
6C12T i7-8700K -> release price: $359 and release date: October 2017
And the fastest was i7-8700K(not in everything) and also had a much higher OC potential.
BTW, at that time Intel also had a lot faster cores compared to zen.

Now regarding 16C32T 7950X it was effectively launched at $569 (price dropped to that soon after release). If 8950X launches at $699 it'll be 23% higher price than for 7950X. So yes, I don't think it's unrealistic to expect 24C48T for 8950X then (if there actually will be a 24C48T Zen5 variant on AM5). If AMD did like in 2017, they would even double the cores to 32C64T and price it lower than $569 at the same time! But I do know that won't happen unfortunately.

You were talking about price at launch, that's $699 not $569.
You want for the same price 50% more cores and also higher performance per core(IPC).
AMD never offered you 2x more cores for less, stop continuing with this nonsense.

 

[aigomorla]

Moderator Stance...

This is an AMD thread.
You will have a lot of AMD supporters.
You will have them throw a lot of hypothetical stuff on top.
You are free to ask for proof and challenge statements and claims.
Infact i support you guys asking for proof in some form of reputable publication.

However Anandtech protects its members, even if they are green and they look and smell like trolls.
So member callout in Tech, of ANY KIND will not be tollerated.
Let us handle the Trolls and do not get baited by them.

I suggest you do not challenge us on that... and instead issue a challenge on the poster to supply reputable publication for his/her/their claims.

Moderator Aigo.