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

Page 919 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

MS_AT

Senior member
Jul 15, 2024
869
1,763
96
Classic AMD - the cheapo company.
Well, while I personally was looking forward to wider link to CCD in Halo, to be fair, there is an iGPU for whatever they need. It's not like on Granite Ridge, where 9950x has more flops than the integrated GPU iirc;) The only downside to iGPU, it is that its AMD one, so software support, well, let's not talk about it. While on CPU side, you can use Intel libs quite easily most of the time so the increased BW could come in handy in some corner cases;)

Still I am curious how low the idle power draw will go.

I think he simply misspoke and meant that 32B/cycle behavior is similar to that of Strix Point.
But do we have any source on Strix Point beyond C&C diagram. They sometimes make mistakes. I mean there is already precedent for Strix Halo CCD to have 32B read/write as Turin CCDs does that [and I would guess Granite Ridge CCDs as well would be capable if not for the old iOD, but this is only my guess, no data to back that up].

NV AI+Gaming Halo range
Well, their only proposed candidate is still far way, will lack Windows support most likely, and is locked down to one mini-workstation SKU, so I guess AMD might enjoy some wins this year, and the assault from the ARM competition will come in 2026, once again pure guesswork, but they are people lurking on this forum that have some access to actual roadmaps so maybe they would be able to give more accurate info.
 
  • Like
Reactions: Tlh97

Keller_TT

Member
Jun 2, 2024
158
186
76
Well, their only proposed candidate is still far way, will lack Windows support most likely, and is locked down to one mini-workstation SKU,
I guess you're talking about a completely different product category. I'm not talking about Project Digits or AI focused 'compact' workstations.
I'm talking about NV's upcoming laptop custom ARM SoC for Windows with integrated Blackwell graphics. They're meant to cover the mainstream as well as the higher end. NV could also make chipsets with a basic APU and mobile dGPU from Ada & Blackwell.
Windows on ARM is very capable now, and Nvidia is the one name that can remake the whole PC segment and push gaming forward on ARM.
 

DZero

Golden Member
Jun 20, 2024
1,625
629
96
Why would you use a GPU with this part outside of eventual upgrades (when games become too demanding to run, which won’t happen for a long time)

Meh, bandwidth is fine.

My 7950x won’t do +50 even. It jumps off the motherboard and runs away if I even think about enabling PBO.

Fallen, Intel has.

Ignore early benchmarks, and ignore benchmarks from a single product. Consider it a ballpark.

Sadly, best ASUS has given me is AGESA 1.2.0.2b

Lame

They have to save some gains for next gen. 🤣

They never know if Intel might pull another Core(tm) moment.

Cheers🥂
Spoiler: they won't. Core moment only happens once and Intel pulled it. AMD pulled with Ryzen and Apple with the A9 series.
 

StefanR5R

Elite Member
Dec 10, 2016
6,675
10,568
136
[Strix Halo, fabric bandwidth to CCDs]
Well, while I personally was looking forward to wider link to CCD in Halo, to be fair, there is an iGPU for whatever they need.
Plus, as the interviewee hinted at:
  • In a client CPU, memory latency is generally much more of a concern than memory bandwidth.
  • In an APU, one of the big problems to solve is to retain reasonable memory latency for the CPU even when the iGPU is utilizing the memory interface a lot. (Remember, bandwidth and latency are not decoupled. As you approach bandwidth limits, latency eventually spikes. Earlier if the system architecture is bad.) He claims they achieved good memory access latency for the CPUs even at certain iGPU load levels. 3rd party tests will have to verify that.
They have to save some gains for next gen. 🤣
Strix Halo is not just Client segment, it particularly is Mobile APU segment.
The Zen 6 generation is when upgraded packaging will also be adopted in the Desktop segment, where memory bandwidth to the CPU plays a somewhat different role. (At the high end of the segment, that is.)

Edit: In this mobile SoC, they use a wide but low-clocked fabric between CCD and IOD in order to save Watts, and Watthours. Which makes a lot of sense. (In contrast, in a high-end desktop chip, it would make sense to use a wide and high-clocked fabric in order to gain computing throughput.) It also makes sense that they merely evolved the existing link a little: They just got rid of the serdes's but didn't change it otherwise. I look at that more like Igor does, less like Yury does.
 
Last edited:

MS_AT

Senior member
Jul 15, 2024
869
1,763
96
I guess you're talking about a completely different product category. I'm not talking about Project Digits or AI focused 'compact' workstations.
I'm talking about NV's upcoming laptop custom ARM SoC for Windows with integrated Blackwell graphics. They're meant to cover the mainstream as well as the higher end. NV could also make chipsets with a basic APU and mobile dGPU from Ada & Blackwell.
Ah yes, I have assumed HALO competitor. But I think this still apply that whatever NVidia and Mediatek are cooking together it rather won't go for sale till the end of 2025, seeing we know almost nothing about the chip(s).
 

LightningZ71

Platinum Member
Mar 10, 2017
2,508
3,191
136
NVIDIA has two methods of attack on the market. They can choose to try to enter and evolve the WoARM world which they certainly have the software development capabilities to do. This would be the most seamless approach to enter the broader market. They can also choose to come at the market from the steam deck route. They can throw software development hours after making SteamOS run perfectly on their platform, (which would take buy in from valve) and evolve it a bit to be a better desktop OS experience, making it finally the year of the Linux desktop. Since so many apps are either web based now, or are well handled by Proton/Wine, they could get sufficient market penetration to make the headaches worth it.

Either way, they are losing the bottom, high volume end of their dGPU market and need something to continue to offset gaming GPU development costs with volume.
 
  • Like
Reactions: biostud and Tlh97

MS_AT

Senior member
Jul 15, 2024
869
1,763
96
Nice to see confirmation that they will ship 128GB version, I am just afraid what the price will be;p
They can throw software development hours after making SteamOS run perfectly on their platform, (which would take buy in from valve) and evolve it a bit to be a better desktop OS experience, making it finally the year of the Linux desktop
That would certainly be a more surprising development but also maybe something that would allow them to fill a niche for more performant devices using SteamOS.
 
  • Like
Reactions: lightmanek

Det0x

Golden Member
Sep 11, 2014
1,465
4,999
136
Wonder what programs is supported and how it affects performance..
Some info on what games thats supported

Application Compatibility Database File 蠁 DXMD.exe 蠁0 Deus Ex Mankind Divided 蠁 <Unknown> 蠁 * 蠁$ ProcessorCountLie 蠁 8 蠁8 DyingLightGame_x64_rwdi.exe 蠁 Dying Light 2 蠁 FarCry6.exe 蠁 Far Cry 6 蠁 16 蠁 MetroExodus.exe 蠁 Metro Exodus 蠁& Three_Kingdoms.exe 蠁2 Total War Three Kingdoms 蠁 Warhammer3.exe 蠁0 Total War Warhammer III 蠁* Youngblood_x64vk.exe 蠁. Wolfenstein Youngblood
 

yottabit

Golden Member
Jun 5, 2008
1,671
874
146
Well it looks like I’ll finally be taking the time to mess around with Parallels or similar VM on my wife’s M1 Macbook Pro. At those price points I don’t think Strix Halo is going to be worth it for me over a MBP for the couple Windows-only apps I need to run. Just need to confirm they work through emulated x86. Most of the software I use on my laptop has Mac OS native ports and I don’t use it for gaming anymore now that handhelds are a thing.

The real killer on the value proposition of Macs is storage though, my aging HP Omen has a modest i7-9750H and 32 GB RAM, but I was able to put in a 4 TB NVMe SSD for like $200? Definitely not paying Apple prices for that much storage. Maybe 2 TB can cut it :tearsofjoy:

I’ll still be very curious to see battery life for Strix Halo under mixed loads
 
Last edited:

fastandfurious6

Senior member
Jun 1, 2024
761
957
96

disappointing.....

asus g16/g18 are older chassis and only up to 5070ti 12gb, official site says max 5070 8gb... $2k

legion 5 is the new horrible chassis and only up to 5070 8gb... at least it has OLED

the 3 MSI models are too heavy 3.6kg and def VERY expensive, the 5090 model will be like $5k
 
Last edited:

biostud

Lifer
Feb 27, 2003
19,910
7,016
136
NVIDIA has two methods of attack on the market. They can choose to try to enter and evolve the WoARM world which they certainly have the software development capabilities to do. This would be the most seamless approach to enter the broader market. They can also choose to come at the market from the steam deck route. They can throw software development hours after making SteamOS run perfectly on their platform, (which would take buy in from valve) and evolve it a bit to be a better desktop OS experience, making it finally the year of the Linux desktop. Since so many apps are either web based now, or are well handled by Proton/Wine, they could get sufficient market penetration to make the headaches worth it.

Either way, they are losing the bottom, high volume end of their dGPU market and need something to continue to offset gaming GPU development costs with volume.
Yup, it shouldn't be to difficult for Nvidia to build a good gaming focused arm SoC, with much better driver support than Qualcomm.

Whether it is worth it is another matter.