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

Page 487 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,740
14,772
136
I said ~32% in SIR which this isn't.
Either Turin perf is down or client workloads somehow map worse to Zen5 which is a first for AMD.
sorry, its been like ages for me, I thought it was 39% in something.
If true, it's going to get stomped by ARL-S. Worse node, comparable ST at best, likely worse MT.

kekw
With NO SMT ? Lets wait until they are both out.
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,740
14,772
136
Not for 128c SKU.
Mind that AMD upped Turin numbers last year.
My Genoa (360 watt 9554) gets. 3.5 ghz, but my retail 9654 only. gets 2.7 all core. And I mean 100% load with SMT all cores, sustained with what I consider only a middle of the line HSF.
96c skt perf at max cTDP is 50% moar SIR score.
Assuming conservative clock bumps from moar power and a bit better node, it kinda goes to around ~32% PPC.
Yes, at any rate, it looks killer to me, but final judgement is after a full suite of benchmarks. Right now Genoa is untouchable by anything Intel has, regardless of TDP. Once you include that Intel is in the toilet. Turin will just be embarrassing.
 

Doug S

Platinum Member
Feb 8, 2020
2,483
4,041
136
Yeah.
LLMs cost actual money to run.
Gotta pay.


What I was talking about should be able to be done on device. Heck transcription has been available for a while - certainly doesn't require an LLM. Apple has AI related patents to do with running stuff in a memory limited mobile device, we'll have to see what comes of that but that's a capability they care about more than other companies and will focus more of their investment there.

So no I don't believe Apple will charge for something simple like that. At worst it will be like iCloud, where you get the feature for free but the more heavily you want to use it the more you have to pay.

Dude, unless you say to yourself "AI" hundreds of times a day or like hearing it, just look at the slides and skip the rest. The entire industry outside of publicly traded companies is sick of AI being constantly pinned on everything.

Atom is great. It is one of the few things Intel has going for it that could make a difference.

Now you understand Apple's approach to AI and why they have tried to be very careful about execution. Local, device only AI is the actual future, not all that cloud based subscription nonsense and expensive data center stuff that NVIDIA is making tons of money on.

For every 100 billion companies are throwing away "investing", they are seeing only about a billion in revenue in return, and revenue != profit. The bubble pop is going to seriously kill economies worldwide. You can call me a pessimist and ignore me, but unless things changed drastically, the collapse is coming. Bookmark this post.


You can only do so much on device, and not just because devices are limited in the size of LLMs they can run. For example, if you ask your phone something to do with searching the internet that can't be done on device, because you can't store an index of the entire web on your phone. So Apple is going to have to have some sort of cloud based AI for queries that cannot be done locally, whether they get that from Google, OpenAI, or build their own AI cloud with their own silicon.
 

poke01

Golden Member
Mar 8, 2022
1,392
1,603
106
My Genoa (360 watt 9554) gets. 3.5 ghz, but my retail 9654 only. gets 2.7 all core. And I mean 100% load with SMT all cores, sustained with what I consider only a middle of the line HSF.

Yes, at any rate, it looks killer to me, but final judgement is after a full suite of benchmarks. Right now Genoa is untouchable by anything Intel has, regardless od TDP. Once you include that Intel is in the toilet. Turin will just be embarrassing.
It’s looking like Zen 5 is great for DC but not for desktop. Looks like Intel might have chance with ARL
 

Panino Manino

Senior member
Jan 28, 2017
846
1,061
136
16% ipc increase... if we take one lesson from this it's don't trust leakers


The Sonic Cycle is real, exactly HALF the increase I "expected"/wanted.
Yes, clock speeds, even so I expected more with all the changes to the architecture. Not to mention AMD needs much more than this to stay competitive.
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,740
14,772
136
It’s looking like Zen 5 is great for DC but not for desktop. Looks like Intel might have chance with ARL
See my post below yours. And remember, ARL is not supposed to have SMT

EDIT: I can't believe this, so it doesn't appear to have 30% average gain, and we don't even have benchmarks, and people are saying RIP...

If you twist the stats ADVERTIZED you could take the 7 to 56%, average that and say its 31.5% lol.
 
Reactions: lightmanek

SarahKerrigan

Senior member
Oct 12, 2014
599
1,460
136
Hype train coming back to earth. People always seem to over hype AMD's new stuff thats coming out. Looks like a solid product but the people expecting +30% IPC are always disappointed.

Yeah, I mean, they've been hanging around +15% for multiple microarchitecture generations on top of a pretty solid baseline. That's not bad, especially since they've done it on a reasonably quick cadence. (But it was kind of funny watching this thread whip itself into a frenzy about how +25% or higher was a done deal and +40% was plausible...)
 

poke01

Golden Member
Mar 8, 2022
1,392
1,603
106


If we remove the Geekbench AES test which I think uses AVX-512 its not good. These have to be fake or old but thts looking unlikely
 

Doug S

Platinum Member
Feb 8, 2020
2,483
4,041
136
Zen 5 and X5 - both ended up about the same.

The "Apple is standing still, AMD, Qualcomm and ARM are going to leave them in the dust and Intel is going to lose their entire market and go bankrupt" ideas that seemed to be popular in the Anandtech forums the last few months sure haven't panned out the way some people were thinking/hoping it would lol!

When we will see independent reviewers like Anandtech able to run their own benchmarks, measure power, and all that good stuff? This week, or will that wait until the "official" release in July?
 
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |