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

Page 824 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

MS_AT

Senior member
Jul 15, 2024
270
615
96
One of the theories as to why Zen5 struggles to outpace Zen4 is that there isn't enough memory bandwidth for it to stretch its legs. Maybe the 3d cache will let zen5 stretch its legs a bit and maybe the clock regression for the X3D chips won't be there. If they just fixed the clock regression it could come in at 15-20% faster than zen4X3d, if the x3d cache mitigates some of the memory bandwidth problems it could go even higher.
Who knows? I do think Zen5 is a good chip, it's just not worth upgrading if you have a zen4.
Games are latency not bandwidth bound. The 3d cache is helping with latency too.
 

yottabit

Golden Member
Jun 5, 2008
1,502
543
146
Uhm, system performance, including for a range of game titles, did indeed improve (at least measurably) when Ryzen 5000 increased last-level cache per CCX from 16 to 32 MB, and Ryzen 5000X3D increased it once more to 96 MB. (Still down from Intel Broadwell's 128 MB of last level cache.) Ryzen 7000 and 9000 as well as 7000X3D and 9000X3D merely replicate this. Furthermore, Ryzen 7000(X3D) introduced DDR5 support, and Ryzen 9000(X3D) merely replicates this.

So did you expect some other drastic system level changes from Zen 5/ Zen 5X3D, and feel let down because there are none?
Or did you expect generational increases of CPU core computing speed to bring relevant improvements to games, and feel let down because that's not the case?
He’s upset that the 9800X3D is underperforming in games

Even though we have no idea how fast it is yet
 

itsmydamnation

Platinum Member
Feb 6, 2011
2,940
3,600
136
So i see two things to ponder,

1. clocks , can they get x3d clocks closer to base models relative to zen4?
2. can vcache allow Zen5 to stretch its legs further relative to zen4?

you can see Zen5 is both more data memory bound and front end latency bound then zen4 . Also zen5 is less retirement bound then then Zen4 and Zen4x3d puts noticeably more pressure on retirement.

So zen5x3d could be a banger if both the above occur.
 
Last edited:

jdubs03

Senior member
Oct 1, 2013
900
547
136
X3D news for you all to feast on:
The gaming performance differential compared to the 9800X3D will be interesting. Though I’m not sure how much of an uplift there will be considering most games don’t use more than eight cores.
 

Hail The Brain Slug

Diamond Member
Oct 10, 2005
3,517
2,470
136
X3D news for you all to feast on:
The gaming performance differential compared to the 9800X3D will be interesting. Though I’m not sure how much of an uplift there will be considering most games don’t use more than eight cores.
That source is so questionable I would not believe this for 1 second.

It says "64MB + 128MB Vcache" for the 9950X + 9900X
then it says "32MB + 96MB VCache" for the 9800X

They can't even get the fine details correct.
 

LightningZ71

Golden Member
Mar 10, 2017
1,801
2,161
136
The ONLY real hope that the 9xxx series has from the X3d parts is that their performance is being held back a lot by the bandwidth limits between the CCD and the IOD. If that's the case, then the X3d parts could see notably better uplifts in SOME games. Unless AMD has some real potent "secret sauce " brewed up for the x3d parts, I don't see it.
 

gdansk

Platinum Member
Feb 8, 2011
2,989
4,541
136
The ONLY real hope that the 9xxx series has from the X3d parts is that their performance is being held back a lot by the bandwidth limits between the CCD and the IOD. If that's the case, then the X3d parts could see notably better uplifts in SOME games. Unless AMD has some real potent "secret sauce " brewed up for the x3d parts, I don't see it.
There's actually at least 3 hopes people have mentioned:
1. Bandwidth/lower effective latency helps Z5 more. Maybe, but I doubt it.
2. Clock rate is higher. Maybe 100MHz, but not significantly.
3. Cache chip is different. Slightly lower latency? I doubt it but it's possible.
 
Last edited:
  • Like
Reactions: Elfear and Joe NYC

RnR_au

Platinum Member
Jun 6, 2021
2,068
5,023
106
Shame there are no lows...

1727487868220.png

 

KompuKare

Golden Member
Jul 28, 2009
1,193
1,498
136
Shame there are no lows...

View attachment 108368

There are at least for 1080P upwards, W1zzard just prefers to put them on a separate page:

1000015585.png
 
Jun 1, 2024
177
249
76
X3D news for you all to feast on:
The gaming performance differential compared to the 9800X3D will be interesting. Though I’m not sure how much of an uplift there will be considering most games don’t use more than eight cores.

called it!
 

RTX2080

Senior member
Jul 2, 2018
327
525
136
X3D news for you all to feast on:
The gaming performance differential compared to the 9800X3D will be interesting. Though I’m not sure how much of an uplift there will be considering most games don’t use more than eight cores.
That source is so questionable I would not believe this for 1 second.

It says "64MB + 128MB Vcache" for the 9950X + 9900X
then it says "32MB + 96MB VCache" for the 9800X

They can't even get the fine details correct.
There is this though:

Referenced by HXL.

Even AMD hasn't decided it yet. They had tested various option in lab and successful, like dual layer Vcache and dual CCD Vcache, but nonetheless would increase the cost.
 
  • Like
Reactions: Mopetar

StefanR5R

Elite Member
Dec 10, 2016
5,936
8,881
136
is 96mb per CCD really a hard limit? why would it be?

what is the limiter and why couldn't they change it for a new SKU?
There are certainly several limits which each are more or less costly to bump up (but which may or may not have been reached by the 96 MB 16-way associative config yet). For example, my understanding is that L3$ tags need to be tracked in the IOD, which imposes whatever limit of cache line count they thought of back in the day when they designed this IOD. (Somebody correct me if I got this wrong.)

Edit, or on the hardware side, a 32 + 64 + 64 stack might have worse timings and/or thermals than the 32 + 64 stack, possibly negating whatever diminishing returns would come from 160 MB L3$ compared to 96 MB.
 
Last edited:
  • Like
Reactions: dr1337
Jun 1, 2024
177
249
76
It's better in everything than Phoenix(1T,nT,iGPU, battery life), yet It's a flop? Ridiculous.
Just because you believed the unreasonable IPC gain rumors doesn't mean It's a Flop.

phoenix -> stx is really not worth the upgrade

I also did 5800H -> phoenix and it also wasn't really worth it tbh, main benefit is lower thermals

phoenix -> halo/firerange will be worth it