Info 64MB V-Cache on 5XXX Zen3 Average +15% in Games

Page 78 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Kedas

Senior member
Dec 6, 2018
355
339
136
Well we know now how they will bridge the long wait to Zen4 on AM5 Q4 2022.
Production start for V-cache is end this year so too early for Zen4 so this is certainly coming to AM4.
+15% Lisa said is "like an entire architectural generation"
 
Last edited:
  • Like
Reactions: Tlh97 and Gideon

Topweasel

Diamond Member
Oct 19, 2000
5,437
1,659
136
It should be clear by now that AMD doesn't like to leave high margin profit on the table. Intel can get the quantity profit.
Well they also suffer from much more limited production than Intel. It was one thing when they had nearly no market share and were struggling to sell there production. But they can't compete with Intel on high volume low margin, without turning a significant amount of their wafer allotment to these products. We are several years before AMD can up their allotment to the point where they can focus on low-end low margin components.

Heck even the 6500 isnt about a cheap desktop video card, it's about a mid level laptop graphics ala 1050/1060/3050, that AMD found they could actually sell on desktop for similar margins due to the desktop market. Otherwise they wouldn't have gimped it nearly as bad as they did.
 

eek2121

Diamond Member
Aug 2, 2005
3,100
4,398
136
Saw this on AMD’s site for the 5800x3d:

Max Memory Speed
2x1R DDR4-3200
2x2R DDR4-3200
4x1R DDR4-2933
4x2R DDR4-2667

First time I have ever seen them publish specific numbers.
 

Topweasel

Diamond Member
Oct 19, 2000
5,437
1,659
136
Except that I have 4 x 16 gig 3600 @ 3600 on multiple machines, so I don't believe those stats. However, at cl16, and some at 3200 cl14.
I don't think it's the ceiling but the floor. That the chips, all of them are rated to be able to run at those limits in those configurations. But it doesn't mean you can't run with more ranks/dimms at higher speeds.


Edit: The major issue is going back to first and second gen (specially first) when chances are best case scenario we can hit the official ratings. We tend to pay too much attention to the memory ratings now.
 

moinmoin

Diamond Member
Jun 1, 2017
5,064
8,032
136
Well they also suffer from much more limited production than Intel.
Yes, Intel will continue the king in quantity (the push for IFS 2.0 is mainly for ensuring at least that). The trouble is, for its leading processes Intel naturally wants to be able to command big margins as well. That doesn't work as well if competition regularly outperforms your best output on several parameters, except quantity. Though it does appear TSMC gives Intel some breathing space with N3.
 
  • Like
Reactions: Tlh97 and Joe NYC
Jul 27, 2020
20,040
13,738
146
Assuming that Intel is solving with ease the challenges TSMC is having - or is able to, with ease, leapfrog TSMC to GAA.

While it is within realm of possibilities, I think it is unlikely.
Isn't it possible that the way TSMC is trying to go beyond 3nm is just harder than what Intel has already researched to be feasible and achievable?

TSMC To Fall Behind Both Intel, Samsung By 2024 (NYSE:TSM) | Seeking Alpha

Intel took longer but they did manage to create 10ESF in a unique and innovative way. I'm still not sure what hampered the creativity of Intel engineers during Intel's process manufacturing dry spell but it could have been related to the exodus of engineers during Sohail's reign as TMG head.
 

coercitiv

Diamond Member
Jan 24, 2014
6,677
14,272
136
Isn't it possible that the way TSMC is trying to go beyond 3nm is just harder than what Intel has already researched to be feasible and achievable?
Friendly reminder that Intel's node development woes are not over just because they're talking about an aggressive schedule. The node they currently refer to as "Intel 4" is 12 months late relative to their own internal target, and 6 months late to the market. Granite Rapids was initially spotted in their roadmaps as a 2022 product, later made public as a 2023 product on Intel 4, now slated for 2024 on Intel 3.

And that's just the next node, the one after that is deeper in the fog. It may be doing better, it may be doing worse, there's no way of telling because Intel is on a mission of selling the entire world their vision of bringing balance the the semiconductor world. First hey need to get the money, then we get the truth about their real progress (which may as well be on track).
 

Doug S

Platinum Member
Feb 8, 2020
2,784
4,745
136
Isn't it possible that the way TSMC is trying to go beyond 3nm is just harder than what Intel has already researched to be feasible and achievable?

Everyone is doing GAA as the post FinFET step. They aren't doing it all in exactly the same way, of course, but they are following the same path. Intel "researched" 10nm to be feasible and achievable, but it was not. TSMC also researched N2 to be feasible and achievable, as with Intel having something work in the labs isn't the same as having it work for 100k wpm mass production so there are no guarantees for either of them.

You are being blinded by Intel's rosy roadmaps and showing off an advanced wafer. Have you forgotten the lessons of their 10nm failure so soon? The execs want their underwater stock options to recover, of course they are telling everyone that everything is great. Up until the last possible moment when they are forced by SEC regulations to admit it isn't, if they fail to meet their very optimistic roadmap.