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

Page 828 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Jul 27, 2020
27,981
19,119
146
This could be the last real x86 generation. Nova Lake may feature the proposed x86S and Zen 6 could emulate that in firmware, despite being proper x86 compatible, just so they can pretend that they are entering a new era too.
 

CakeMonster

Golden Member
Nov 22, 2012
1,629
809
136
Well, Z6 could arrive any time from ~18 to ~30 months after Z5 I suppose, so there's at least hope something exciting will happen in not that long (relatively speaking on the short end). I'm in the camp that with everything slowing down, Z6 will use DDR5, not DDR6. Hopefully N3E or whatever it requires will be cheap enough not to delay it more than whatever release window AMD has planned...
 

Mopetar

Diamond Member
Jan 31, 2011
8,487
7,726
136
x86 hasn't been x86 in a long time if you want to look at it that way. Long ago Intel and AMD realized that making a true CISC processor was a nightmare and instead translate the instructions into microcode/micro-ops to simplify the hardware.

x86S just gets rid of some cruft that no one is really using anymore. It's not really entering a new era so much as it is forgetting about some that are no longer relevant or worth supporting. The number of customers who need to run software that old can likely do so just fine on older hardware or aren't even running on x86 in the first place.

If people aren't using some of the legacy instructions in the first place why not ditch them and save transistors on the front-end of the CPU? Particularly considering both Intel and AMD are trying to make wider front-ends that can decode a larger number of instructions.

There are some other reasons to retire part of the ISA outside of this as the older stuff is outside of the duration of any patents and getting everyone to move away from it discourages anyone else from making a CPU compatible with earlier versions of the ISA. Those are probably just nice side effects as opposed to a driving force, but it's still a consideration.
 

inquiss

Senior member
Oct 13, 2010
510
756
136
Well, Z6 could arrive any time from ~18 to ~30 months after Z5 I suppose, so there's at least hope something exciting will happen in not that long (relatively speaking on the short end). I'm in the camp that with everything slowing down, Z6 will use DDR5, not DDR6. Hopefully N3E or whatever it requires will be cheap enough not to delay it more than whatever release window AMD has planned...
It definitely won't use the server-only DDR6...
 

DavidC1

Golden Member
Dec 29, 2023
1,833
2,959
96
If people aren't using some of the legacy instructions in the first place why not ditch them and save transistors on the front-end of the CPU? Particularly considering both Intel and AMD are trying to make wider front-ends that can decode a larger number of instructions.
And are those transistors contributing to power and/or decreasing clocks or efficiency in the cases where they aren't used? Modern processors are far more limited by power not die area.

Or maybe they contribute to design complexity by increasing validation difficulties and time?

If no, then the effects are minimal to negligible.
 

Tuna-Fish

Golden Member
Mar 4, 2011
1,665
2,531
136
If people aren't using some of the legacy instructions in the first place why not ditch them and save transistors on the front-end of the CPU? Particularly considering both Intel and AMD are trying to make wider front-ends that can decode a larger number of instructions.

This has no meaningful impact on making decoders easier. The job of decoding an instruction once you have figured out where it starts and have it lined up with a relevant decoder is completely trivial. It could be done in a single cycle in hardware 30 years ago. The entirety of the problem with x86 decode is that it's byte-aligned variable width, with length between 1 and 15. And x86S does nothing for that.
 

Josh128

Golden Member
Oct 14, 2022
1,319
1,984
106
Very interesting. 9950X seems to be not listed at Amazon anymore, is $799 on the egg, and $900+ on Ebay. WTH is happening??
 
  • Wow
Reactions: Det0x

samboy

Senior member
Aug 17, 2002
223
94
101
Very interesting. 9950X seems to be not listed at Amazon anymore, is $799 on the egg, and $900+ on Ebay. WTH is happening??
Could simply be that the initial allocation has sold out.......... it wasn't that long ago that this happened within the first couple of days of a popular new processor release. Given the under whelming release for this new generation it has simply stayed in stock longer before this happened?
 

Joe NYC

Diamond Member
Jun 26, 2021
3,634
5,176
136
Very interesting. 9950X seems to be not listed at Amazon anymore, is $799 on the egg, and $900+ on Ebay. WTH is happening??

Whenever the primary seller (Amazon, Newegg) is out of stock (at normal prices) the price switches to 3rd party sellers, who may have high prices.
 

Joe NYC

Diamond Member
Jun 26, 2021
3,634
5,176
136
Some more nontent on 9800x3d as far as capabilities and performance, but says launch may be imminent (within weeks).

 

coercitiv

Diamond Member
Jan 24, 2014
7,355
17,424
136
Is it the case that when MLID says something will happen it means that it will not happen?
Nope, when MLID says something will happen it means you're not wiser after hearing his words: you do not know more about whether it will happen, won't happen, was planned, was not planned. You're just ad revenue.

When MLID says that Schrödinger's cat is dead, you don't even know if Schrödinger is real or not.
 

Josh128

Golden Member
Oct 14, 2022
1,319
1,984
106