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

Page 660 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

jdubs03

Senior member
Oct 1, 2013
803
408
136
Well, look what we have here :)
Same boost clock as the 370. Would be nice if it had an extra 100 MHz.

 

leoneazzurro

Golden Member
Jul 26, 2016
1,051
1,711
136
Well, look what we have here :)
Same boost clock as the 370. Would be nice if it had an extra 100 MHz.

They probably only overclocked a bit the NPU because of HP's requests. Nothing more than that.
 

Josh128

Senior member
Oct 14, 2022
350
476
96

Reviewers definitely have ZEN5 in Hand right now.
Gary Colomb is a laptop reviewer-- he is likely speaking about Strix.

**EDIT: Its exactly 7 days before "official" launch date and we still have no pricing announcement from AMD-- I very much doubt that Zen 5 desktop reviews are dropping today. I did read somewhere that today there may be Strix laptop reviews available.
 

Det0x

Golden Member
Sep 11, 2014
1,262
3,991
136
Seems like some reviewers are still waiting for samples
And i have a sneaking feeling there will be few 9950X for sale at launch
 
  • Like
Reactions: inf64

Abwx

Lifer
Apr 2, 2011
11,536
4,323
136
Some info from AMD, lot of slides, among others this one wich state configurable 256/512b datapath in Zen 5 :


3-1080.770b0664.png


12-1080.d611528e.png



 
Last edited:

Hail The Brain Slug

Diamond Member
Oct 10, 2005
3,497
2,443
136
Seems like some reviewers are still waiting for samples
And i have a sneaking feeling there will be few 9950X for sale at launch
Time to setup the bots/autorefreshers/stock notifs from Falcodrin
 

MS_AT

Senior member
Jul 15, 2024
219
521
96
Some info from AMD, lot of slides, among others this one wich state configurable 256/512b datapath in Zen 5C :


3-1080.770b0664.png


12-1080.d611528e.png



AMD marketing is making lot's of typos, while L1/L2 BTB might just sound weird and be the actual value, then FP pipes is definitely wrong as Zen4 had 4 pipes so far according to AMD itself, unless they will now disable one pipe over the net;)
 

Abwx

Lifer
Apr 2, 2011
11,536
4,323
136
AMD marketing is making lot's of typos, while L1/L2 BTB might just sound weird and be the actual value, then FP pipes is definitely wrong as Zen4 had 4 pipes so far according to AMD itself, unless they will now disable one pipe over the net;)

The L1/L2 BTB is indeed curious, as for the FP pipe methink that it s related to the 3/1 L/S ratio that limit the effective pipes utilisation to 3 ops/cycle, if the operands cant be loaded more pipes will have no effect for the thoughput.
 

MS_AT

Senior member
Jul 15, 2024
219
521
96
The L1/L2 BTB is indeed curious, as for the FP pipe methink that it s related to the 3/1 L/S ratio that limit the effective pipes utilisation to 3 ops/cycle, if the operands cant be loaded more pipes will have no effect for the thoughput.
They could mark that on the slide, like 3 (effective) with explanation at the end of the slide deck. I might be nitpicking but I think the quality of AMD press materials is getting worse instead of getting better. Actually that was 1 FP512 load per cycle and 0.5 store per cycle. But still the execution pipes won't go to waste if you do something more than shuffle memory around and keep data in the registers.
 

mostwanted002

Member
Jun 16, 2023
33
66
61
mostwanted002.page

gdansk

Platinum Member
Feb 8, 2011
2,890
4,363
136


It's really joever for intel.. is it??
I really don't want to talk about Intel products in this thread.

In any case I say both x86-64 vendors have been caught slacking with desktop.
 
Last edited:

exquisitechar

Senior member
Apr 18, 2017
683
940
136


It's really joever for intel.. is it??
Really want to see the efficiency of both, but Granite as well as ARL-S seem to be a disappointment in terms of performance. Don’t care who wins by a tiny margin. Well, the silver lining for AMD is that they’re on N4P instead of N3, at least…
 
  • Like
Reactions: Kryohi