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

[DisEnchantment]

Speculate at will

 

[Markfw]

9950x will most likely be $599.

I hate to say it, but I think $799 or $899

 

[H433x0n]

I hate to say it, but I think $799 or $899

$899 would the most expensive mainstream consumer product AMD ever sold, $799 is more than the MSRP for 7950X.

The 7950X is going for <$500 right now, nobody is going to pay >50% more for a 15% performance increase.

 

[Thunder 57]

$899 would the most expensive mainstream consumer product AMD ever sold, $799 is more than the MSRP for 7950X.

The 7950X is going for <$500 right now, nobody is going to pay >50% more for a 15% performance increase.

Athlon 64 X2 says hi:

Imagine that adjusted for inflation. That said, I think Mark is way off target but who knows? Why must AMD torture us?

 

[Tup3x]

I hate to say it, but I think $799 or $899

Not going to happen. It would get insane amount of bashing in reviews. It's rather unlikely that it can beat the X3D variants in gaming and average performance uplift isn't that large either.

 

[Hans Gruber]

Not going to happen. It would get insane amount of bashing in reviews. It's rather unlikely that it can beat the X3D variants in gaming and average performance uplift isn't that large either.

Let's not forget what AMD did to their Threadripper lineup with Zen 3. They priced gouged to the point where nobody wanted to buy a Threadripper.

 

[ondma]

Maybe AMD has knowledge of the capability of Arrow Lake and is aggressively pricing Zen 5 with a 2 or 3 month lead over an Arrow Lake release. A very smart move by AMD if true.

We have seen no leaks that ARL is going to decisively beat Zen 5. In fact, just the opposite. The consensus, except for a few Intel Die Hards, seems to be that at best, ARL will trade blows with Zen 5. Actually, one would think that AMD would price Zen 5 high initially, and then lower prices if necessary when ARL comes out. Especially with the stability problems of 13th and 14th gen high end cpus, Zen 5 should have a stranglehold on the market at least until ARL is available.

 

[Goop_reformed]

Gaming numbers don't seem to be good

https://twitter.com/x/status/1814593519958274458

 

[Thunder 57]

Gaming numbers don't seem to be good

https://twitter.com/x/status/1814593519958274458

I wouldn't read to much into an "attention grabbing" screenshot. I'd expect the 9700X to be similar to the 7800X3D. Of course you can then cherry pick games to try to prove it either way.

 

[Hitman928]

Gaming numbers don't seem to be good

https://twitter.com/x/status/1814593519958274458

This is the same guy that insists that Rocketlake is better than Zen 3. You probably shouldn’t put any trust in what he has to say.

 

[mpumalanga]

Gaming numbers don't seem to be good

https://twitter.com/x/status/1814593519958274458

I don't see any gaming numbers in your message.

CapFrameX is on the same level as userbenchmark with regards to AMD. Best to ignore both unless you want to read something really wrong and misleading.

 

[Thunder 57]

I don't see any gaming numbers in your message.

CapFrameX is on the same level as userbenchmark with regards to AMD. Best to ignore both unless you want to read something really wrong and misleading.

Quick search seems to suggest that yes, they are Userbenchmark-lite, if you will.

 

[Thibsie]

Yeah, he's pretty much that.

 

[Saylick]

Quick search seems to suggest that yes, they are Userbenchmark-lite, if you will.

For a Xitter handle that appears to be the front face for his CapFrameX software, he sure does spend a lot of time pushing his own agenda rather than use the handle to market the software…

 

[poke01]

Aussie prices list for zen5(my guess) in AUD:

9600X - $499
9700X - $650
9900X - $850
9950X - $1199

 

[RnR_au]

@igor_kavinski maybe something of interest for your friend...

https://twitter.com/x/status/1814824411452879026

 

[branch_suggestion]

Gaming numbers don't seem to be good

https://twitter.com/x/status/1814593519958274458

Ignore grifters.
For whatever reason the CFX media guy is an infamous forum agitator, his biases are very straightforward.

 

[csbin]

599 or 649 USD?

https://twitter.com/x/status/1814966458717467129

 

[Joe NYC]

Aussie prices list for zen5(my guess) in AUD:

9600X - $499
9700X - $650
9900X - $850
9950X - $1199

What is that in fluid ounces?

 

[Joe NYC]

The first quote says AMD ought to release Zen5X3D as soon as the chip is ready. But then the second quote says the regular Zen5 was ready already in early 2023, but is not released until now in July 2024.

So in that case why use differt policy for Zen5 vs Zen5X3D? Why not release both as soon as possible after they are ready?

Zen 5 validation was likely far more involved than previous gens, because of the scope of changes and likely, re-write of most of the internal micro code.

In the meantime, the part of the team designing new silicon likely moved to Zen 6.

Also, since Zen 5 reuses Zen 4 IOD, that team may have moved to new Zen 6 style IOD (or Strix Halo).

(For regular Zen5, I know there is some lead time from A0 silicon until actual release is possible. But it ought to be much shorter than from early 2023 until now.)

Yeah, that's the basis of my speculation

For the first bullet, do you mean just frequency bumps or did you have something else in mind?

Could be frequency, could be something to do with V-Cache, or ability to work with higher power envelope for greater all core frequency.

If these later steppings (B0, C0) exist, they exist for a reason. We just don't know what it is.

BTW, it could very well be that these 3 steppings are for:
- desktop
- server
- Strix Halo
- Mi350a

For the second bullet, would that mean Zen6 release in like 12-18 months from now, so mid-life kicker is not needed?

Yeah, that may be a possibility. If Zen 6 can be released by the end of 2025 or at CES 2026, then there is no reason for any "midlife kicker", since the "midlife kicker" Zen 5 x3d is now speculated to be released in 2024

 

[Joe NYC]

For a Xitter handle that appears to be the front face for his CapFrameX software, he sure does spend a lot of time pushing his own agenda rather than use the handle to market the software…

He is also known as CopeFrameX

 

[Timmah!]

$899 would the most expensive mainstream consumer product AMD ever sold, $799 is more than the MSRP for 7950X.

The 7950X is going for <$500 right now, nobody is going to pay >50% more for a 15% performance increase.

7950x was 699 at launch IIRC and i bought it in December, about 3 months till official release, already for 100 less. There were reasons for that, cpus not selling that well cause of the high price of the new AM5 platform overall, which wont repeat this time around, but as other people said, its not that much faster to separate itself at the top for significantly higher prices to be justifiable. IMO the one reason why the CCDs are the same size as Zen4 and only transistor count increase comes from the newer denser process, is cause AMD dont thinks they can sell more expensive than Z4 and wont be giving any additional performance increase for “free” - as in potentially bigger CCDs could have been even faster, but AMD would expect to be paid for every additional mm2, and people would not be willing to pay….

Or i am completely wrong and its the same size cause any bigger would not fit under IHS

 

[exquisitechar]

Yeah, that may be a possibility. If Zen 6 can be released by the end of 2025 or at CES 2026, then there is no reason for any "midlife kicker", since the "midlife kicker" Zen 5 x3d is now speculated to be released in 2024

Closer to CES 2027 according to rumors.

 

[igor_kavinski]

Or i am completely wrong and its the same size cause any bigger would not fit under IHS

The performance increase in various workloads is great considering the CCD size is about the same. It scales better in MT with additional power, as long as it is properly cooled. They are in a lot better position than Intel. They are dealing with the limits imposed by the laws of physics just like everyone else. Even Apple's design isn't immune to these problems and they have been forced to push their silicon to higher frequencies. Qualcomm can't even get their demoed highest frequency SD Elite X SKUs into WoA laptops, possibly due to yield issues.

The future is going to require ingenious solutions, both at the architecture and process level to get bigger gains. Increasing CCD size is a bad option for AMD since they don't own fabs and don't have "blind" brand loyalty like Intel users so they always have problems selling their CPUs at higher prices, except in the server space where the higher performance at lower power use automatically translates into huge data center cost savings.

 

[Josh128]

Regarding CapframeX's not so cryptic tweet. I can believe it. The main reason I believe why is the same IOD & memory limitations of Zen 4. When looking at Zen 4 memory scaling testing for games, 4800 to 6000 shows a real, tangible difference due to the increase being in the same 1:1 ratio. Increases from 6000 1:2 to 8000 1:2 are either very small, or nothing at all. When you look at Intel scaling, they seem to achieve that same boost that AMD gets from 4800 to 6000 all the way to 8000+. No matter how good the core, its going to be limited in this regard. Until AMD improve mem controller and subsystem, the will always be behind the 8 ball in this regard.

Second reason is that it seems that most of the gains in the Zen 5 core are floating point, not integer ops. I think David Huangs testing (with Strix, so not definitive as to GNR) showed ~10% gains in SPECint testing. Its said that integer perf is more relevant to games than floating point. I think we are going to see ~10% boost in games vs Zen 4 at same frequency, which would put it behind Zen 4 X3D on average by 5% or more. Zen 4, when it released, got such huge bumps in frequency and memory speeds compared to Zen 3 that it slightly beat 5800X3D on average in large sets (50+) of games. HWUB found the 7600X to be 1%-2% faster than the 5800X3D over a 50 game suite in the testing they did. AMD is claiming +12% over 5800X3D for Zen 5, so the numbers actually work out perfectly for 10% over 5800X3D.

 

[Timmah!]

The performance increase in various workloads is great considering the CCD size is about the same. It scales better in MT with additional power, as long as it is properly cooled. They are in a lot better position than Intel. They are dealing with the limits imposed by the laws of physics just like everyone else. Even Apple's design isn't immune to these problems and they have been forced to push their silicon to higher frequencies. Qualcomm can't even get their demoed highest frequency SD Elite X SKUs into WoA laptops, possibly due to yield issues.

The future is going to require ingenious solutions, both at the architecture and process level to get bigger gains. Increasing CCD size is a bad option for AMD since they don't own fabs and don't have "blind" brand loyalty like Intel users so they always have problems selling their CPUs at higher prices, except in the server space where the higher performance at lower power use automatically translates into huge data center cost savings.

Yeah, but limiting the size to be about the same is their deliberate decision, and even though the perf increase might be great for that size, its likely limited by that decision to be possibly significantly more.

Regarding scaling better with additional power, granted , it scales where zen4 stopped to scale. The scaling is still rather insignificant, its not like gain additional 50 percent of perf going from 160W to 230. Its maybe 10~15 percent and while thats more than almost nothing before, its still not something extremely impressive, that would take zen5 out of sight of Zen4/RPL perf-wise. And thats really the most important thing imo.