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

Page 952 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

SteinFG

Senior member
Dec 29, 2021
709
830
106
Interesting thing - there's 2 new skus at the bottom: 4-6 cores, 2-4 CUs, 50 tops, 12-14MB cache, and most importantly - just 14 PCIe lanes. I'm 90% sure it's a new die, a successor to small phoenix. 4 CU of RDNA 3.5 is awful though, just a bit better than several years old 5600G/5700G
 

fastandfurious6

Senior member
Jun 1, 2024
443
593
96
it's just like hawk point over phoenix, pointless "refresh" to fill the gap until zen 6 comes out

the slides look off, how is +95% MT possible when it's the same CPU by 99% lol
 

SteinFG

Senior member
Dec 29, 2021
709
830
106
it's just like hawk point over phoenix, pointless "refresh" to fill the gap until zen 6 comes out

the slides look off, how is +95% MT possible when it's the same CPU by 99% lol
It's not +95%, the chart is just not readable

It's +95% at 45W, in relation to 100%, which is CBR23 MT score of HX 370 at 15W. HX 370 itself gets +83% to the score when going from 15 to 45W. So the difference between CPUs is at most 5-6%
 

jpiniero

Lifer
Oct 1, 2010
15,980
6,470
136
Interesting thing - there's 2 new skus at the bottom: 4-6 cores, 2-4 CUs, 50 tops, 12-14MB cache, and most importantly - just 14 PCIe lanes. I'm 90% sure it's a new die, a successor to small phoenix. 4 CU of RDNA 3.5 is awful though, just a bit better than several years old 5600G/5700G

Probably just a very deep cut that has the NPU still intact.... so they can go AI AI AI on a cheaper product.

That's probally why AMD isn't doing anything like upgrading the IGP to RDNA 4... there's no point when the OEMs only care about AI.
 

Ghostsonplanets

Senior member
Mar 1, 2024
773
1,227
96
Ryzen AI 3 with 2CUs is awful. Assuming it clocks at ~3GHz, that's ~0.77 TFLOPs of FP32 performance. Even Ryzen 5 210/Ryzen 3 8440U has the same 4 CU IGP.

And all of that for the NPU. Copilot was a big mess from Microsoft.
 

Ghostsonplanets

Senior member
Mar 1, 2024
773
1,227
96
Still should be OK for 2D usage.
Should run PS4/XOne generation 3D games fine at 720p.

But my point isn't even this. It's just that you're cutting down so much of CPU and GPU prowess in exchange for an NPU that will be useless 99% of the time.

Of course, given it's presented as a mainstream SOC in the slide, AMD probably want to fight off Snapdragon X in $599 devices, availability allowing.

But I really think the NPU is a waste of silicon.
 

LightningZ71

Platinum Member
Mar 10, 2017
2,077
2,525
136
It's already there on the SoC. If it failed, then the chip isn't marketable to the west. If it didn't, then it keeps the product relevant for full featured Windows, which is still the biggest target market available.

Stuff with a failed NPU will likely just get sold overseas in secondary markets.
 

jpiniero

Lifer
Oct 1, 2010
15,980
6,470
136
It's already there on the SoC. If it failed, then the chip isn't marketable to the west. If it didn't, then it keeps the product relevant for full featured Windows, which is still the biggest target market available.

Stuff with a failed NPU will likely just get sold overseas in secondary markets.

I think the failed NPUs are being used by the handhelds. The yields at TSMC are good enough that the NPU doesn't get busted that often.
 
Reactions: Ghostsonplanets

eek2121

Diamond Member
Aug 2, 2005
3,277
4,825
136
That's not possible. Even on regular dual CCD CPU's one chiplet is binned better and runs at a higher frequency.
This is correct. In fact, all of the review samples I've seen have better average clocks on CCD1 than my vanilla 7950X does.
Bad schedulers like in Windows?

I had to disable half of my 7950X3D to get it all working in Win 11. Unless AMD puts V cache on both chiplets next time it will be obvious no-buy from me, 12 cores with extra cache will do just fine then.
Sounds like a "you" problem, as in, if this is an issue, you need help with it, and aren't out to attack AMD, post a thread. I have a non-X3D dual-CCD chip, and I've used a dual-CCD X3D chip rather extensively. Things for both generally work "as expected". As in, "99% of the time, the proper CCD is used, with it only not being used 100% of the time". It does require you to update your drivers, windows, and such, however. If you don't do that, or mess with BIOS settings in an unsupported way, that is a "you". If this is indeed an issue, I encourage you to open a new thread and be prepared to have an objective conversation, so that said issue can be troubleshooted/fixed (so as to rule out user error, because if it IS happening, it will warrant an RMA)
It's not only about clock speed. The V$ die isn't universally faster than the non-V$ die even at the same clock, because larger cache incurs (well, at least on Zen 4 it did) 4-cycle penalty for accessing L3. So if your workload fits within 32MB, it would be faster on the non-V$ die even if it clocked the same.
View attachment 120318
Only for things that don't benefit from more Cache.
Question.. Why is a 16 core 4004 EPYC(4.5 ghz) + motherboard($399+$299 $698 total) close to the price of a 9950x(4.3 ghz, $542) and the 4004 runs faster ????
In addition to what has been mentioned, It's a server part, which comes with things like better support, etc.
View attachment 120635
So, it's no zen6, it's just the "old" zen5 cores. Renaming and that's it.
AMD does this with mobile parts quite often. Unsure why folks are making this a big deal, this has been happening for years.
 

Win2012R2

Senior member
Dec 5, 2024
795
795
96
Sounds like a "you" problem
I did all by the book - even installed Windows 11, which I regret, that stupid Xbox bar, drivers etc - games were not giving me the FPS I expected on 4090, so had to disable non-3D chiplet to get them work very nicely. Funnily enough 18 months later new BIOSes introduced new "optimised mode" for chips like 7950X3D, which in effect did the same as what I did, so clearly even past that time things were still not very good.

It's AMD problem because now when 3D Zen 6 is out I'll only get single chiplet version, rather than two - their loss.
 
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/    |