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

[DisEnchantment]

Speculate at will

 

[Glo.]

Don't believe the rumors spread by... Intel engineers simply.

Same thing happened for RDNA3, same thing happened for Zen 5. Which is why I was cautious believing them.

Still stands.

Its actually third time: Zen 4 was the same.

 

[yuri69]

Unironically great uplift in browsers, what the hell.

Also Python, PHP, Node.js interprets are super fast.

 

[Hitman928]

Turns out it really is a "bulldozer moment" somewhat.
Years of expectation, huge changes to the architecture, negative IPC slower than previous gen, only performs well under specific right circumstances (at least power is good), gives time to the competition to breath and strike back with calm.

Make me fear that Zen 6 and Zen 7 will be Piledrivers and Steamrollers.

​

It's definitely a disappointing release, but where are you seeing IPC being worse?

Unironically great uplift in browsers, what the hell.

Yeah, pretty weird. It seems AMD tried some things architecturally that was very hit and miss, mostly miss but with some good hits here and there. I do appreciate the improved efficiency and AMD's willingness to reduce power, even at the sake of some performance reduction. I do expect the 9950x to have a little better of a showing given that they didn't reduce TDP on that model.

I'm probably in the minority on this, but I think modern CPU power consumption is insane and AMD resetting expectations there (outside of the halo part) is greatly appreciated. With all the talk from reviewers about how crazy high power consumption has gotten, I'm a bit disappointed that this move has been almost ignored by reviewers in favor of highlighting the lack of significant performance improvement. I get it, but still a little disappointed. Like I said, I'm probably in the minority here, but I'm ok with that.

Waiting this long for Zen 5 makes this launch especially disappointing. Had it come out 15 months or so after Zen 4, I think it would have seemed more acceptable, but the extra time combined with seemingly a still shaky release in terms of BIOS readiness just makes it a bit sad, lol. Maybe Zen 5 at least gives them a new foundation with some new lower hanging fruit that they can go after with Zen 6, that's my hope at least.

 

[branch_suggestion]

So uhh, just slapping all the new resource gains on a Z4 style core would've been better than this.
Still no easy fix for the 4 wide decode, maybe 6 wide and cope with the power gains.
But it does give a lot more breathing room for future improvements, assuming they can figure out why Z5 is such a mess.

 

[CouncilorIrissa]

Also Python, PHP, Node.js interprets are super fast.

Truly a server core that somehow found itself on client.

I'm probably in the minority on this, but I think modern CPU power consumption is insane and AMD resetting expectations there (outside of the halo part) is greatly appreciated. With all the talk from reviewers about how crazy high power consumption has gotten, I'm a bit disappointed that this move has been almost ignored by reviewers in favor of highlighting the lack of significant performance improvement. I get it, but still a little disappointed. Like I said, I'm probably in the minority here, but I'm ok with that.

On that we can agree. Both CPUs and GPUs consume embarrassing amount of power on client nowadays.

 

[Heartbreaker]

the gains are there if you look at the comparison with the 7700X-7800X (which consume more) and 7600X.

You consider ~5% to be "gains". For any practical purpose that's essentially identical.

 

[Glo.]

On a serious note.

Its tremendous CPU core for EVERYTHING else, except for gaming.

This should've been realeased as an EPYC/Threadripper CPU first instead of desktop.

 

[SolidQ]

Maybe they trying to get efficiency and improve little perfomance in games for next Consoles CPU

 

[yottabit]

TomsHardware seems to be showing a much bigger geomean gaming uplift than TechPowerUp, especially with PBO enabled on the 9600X / 9700X, but I've never been one to really trust their testing methodology

 

[Heartbreaker]

On a serious note.

Its tremendous CPU core for EVERYTHING else, except for gaming.

It's productivity "gains" essentially match the gaming "gains" at least for HWUB they did.

 

[Glo.]

It's productivity "gains" essentially match the gaming "gains" at least for HWUB they did.

Read Phoronix review.

 

[gaav87]

Well that was kinda expected xD Chinese leak from 10days ago checked out my tuned 5800x3d was 13% faster

 

[leoneazzurro]

You consider ~5% to be "gains". For any practical purpose that's essentially identical.

It depends on the game and how CPU limited you are, in most of the cases even the top CPUs (7800X3D excluded) are not much better. In any case I said "gains" and not "trampling over".

 

[branch_suggestion]

Read Phoronix review.

Phoronix review shows the thing is actually very good, just not in the ways we hoped.
Geomean score is similar to a 13900K with half the power consumption.
As it turns out, FP gains are significantly higher than INT, something I guessed ages ago just from reading the uArch changes.

 

[Hitman928]

AT posted surprisingly high gain on gcc subtest of SPEC, tough compared to 7700, but it's 17 %

Edit:
They have also significantly updated details about the test environment for SPEC, newest CLANG, enabling AVX512 and specifying the WSL version used. Nice

AT SPECint showing 13% improvement overall. The 9700x has a slight clock speed advantage according to boost specs, but without knowing actual running clock speeds, we can't calculate IPC. It is 9% improvement if we assume spec boost speeds, but With Zen 4/5, that's not a safe assumption (could be higher than spec). SPECfp showed significantly higher improvement at 26% overall. For consumer purposes, too many resources were used to improve FP versus INT. For some server customers, maybe AI (?), this will be really good.

 

[Heartbreaker]

Read Phoronix review.

I don't cherry pick reviews looking for results I want. HWUB is my goto. Especially not interested in some esoteric Linux results. Newsflash: Linux is not "EVERTHTHING else".

HWUB non gaming results.

CB: 2% faster
z-Zip: 3% SLOWER
Blender: 0% Identical.
Corona: 11% faster - One significant "win".
Photoshop: 3%
Premiere: 6%

 

[MarkPost]

 

[Panino Manino]

Zen 5% is real.

Btw on Linux it is a different story (just like with Bulldozer): https://www.phoronix.com/review/ryzen-9600x-9700x

Expectation: Zen 35%
Reality: Zen 5%
THE HARSH REALITY: Zen -5%

 

[Glo.]

Phoronix review shows the thing is actually very good, just not in the ways we hoped.
Geomean score is similar to a 13900K with half the power consumption.
As it turns out, FP gains are significantly higher than INT, something I guessed ages ago just from reading the uArch changes.

That is exactly the point I was making here: https://forums.anandtech.com/thread...dge-ryzen-9000.2607350/page-722#post-41270664

 

[branch_suggestion]

Phoronix review shows the thing is actually very good, just not in the ways we hoped.
Geomean score is similar to a 13900K with half the power consumption.
As it turns out, FP gains are significantly higher than INT, something I guessed ages ago just from reading the uArch changes.

Oh, and the biggest takeaway is the most significant gains are in traditionally Intel dominated workloads.
And Linux results in general just look way better than Windows, why exactly is a question for AMD software engineers and whichever intern Microsoft sends as a sacrifice.

 

[Panino Manino]

The worst part is that we’ll be stuck with this total dud of a core for years and Zen6 will only bring iterative improvements over it. At least improvements to the uncore will be something to look forward to.

No way, no way!
AMD can't "survive" selling this for another 2 years! Not for "average consumers".
Zen 6 needs to come in 12 months at most!

 

[FlameTail]

So about that engineering sample from Igor Kavinski...

Do these review numbers line up with that?

 

[yuri69]

And Linux results in general just look way better than Windows, why exactly is a question for AMD software engineers and whichever intern Microsoft sends as a sacrifice.

This was exactly the same case with Bulldozer.

 

[Timmah!]

Phoronix review shows the thing is actually very good, just not in the ways we hoped.

As in the performance? :-D

 

[leoneazzurro]

AT SPECint showing 13% improvement overall. The 9700x has a slight clock speed advantage according to boost specs, but without knowing actual running clock speeds, we can't calculate IPC. It is 9% improvement if we assume spec boost speeds, but With Zen 4/5, that's not a safe assumption (could be higher than spec). SPECfp showed significantly higher improvement at 26% overall. For consumer purposes, too many resources were used to improve FP versus INT. For some server customers, maybe AI (?), this will be really good.

Hwupgrade review found a substantial clock difference between the 7700X-7600X and 9700X-9600X, because the TDP holds back the performance for these SKUs. In Cinebench they foung 500MHz all-core clock difference between the 7700X and 9700X, with the 7700X consuming 40W more, and the 9700X was slightly faster.