• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Question Zen 6 Speculation Thread

Page 234 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Doug S said:
They haven't even STARTED mass production of N2 yet. You think there will be finished N2P wafers in Q3 next year - which would be six months or less after the first finished N2 wafers? Dream on!
Click to expand...
I agree! Unless N2P is such a trivial change to N2 .... but then , what would be the point?
marees said:
Bumblebee (MDS-3): FP10 and/or FP8 socket: TSMC N3C monolithic
2-core Zen 6 + 2-core Zen 6c + 2-core Zen 6 LP = 6 total cores (2+2+2)
2-4 CU RDNA 4 or 3.5 (+) iGPU (both architectures are referenced as possibilities)
128-bit LPDDR5X memory controller
"MDS-3" is specifically listed as targeting the "budget laptop" market


There is also a new one "Medusa point little" aka MDS-2 — maybe this is scrapped now ? (Edit: maybe 2028?)

Medusa Point Little (MDS-2): FP10 socket, TSMC N3P monolithic
2 or 4-core Zen 6 + 4-core Zen 6c = 8-10 cores total (2 or 4+4+2) (AMD has not yet decided if they want to go with 2 or 4 "full" Zen 6 cores)
4 CU RDNA 4 or 3.5 (+) iGPU (both architectures are referenced as possibilities, and there are different CU counts referenced between documents of different variants)
128-bit LPDDR5X memory controller
"MDS-2" is specifically listed as targeting "AI 5" and "AI 3" class products
Click to expand...
It does seem like AMD have abandoned any thought of sharing die across products. They seem perfectly happy to share modular architecture and make a butt ton of die.
fastandfurious6 said:
all these are useless 😭 I'm 100% sure amd is losing money with all these choices

market only needs:
- medusa range (top end, x3D etc)
- medusa halo
- medusa baby
- phoenix++++
Click to expand...
LOL. Not sure I agree with your individual markets; however, I do agree they don't need so many SKU's!
 
Kepler_L2 said:
Yeah, SoundWave too. There's also a few things using Warhammer 40K codenames which haven't leaked yet.
Click to expand...
I used to have a Soundwave transformer toy back in the '80s. He had a cassette player in his chest and a little cassette that transformed into a pterodactyl/ condor. I used to have quite a few transformers, it seemed the toy quality was much better back in the 80s than the trash you generally see on the big box store shelves today. The dinobots were my favorite.

1756954451786.png
 
Some days ago ML said ...

2027 is too long ...
Click to expand...

... now you can guess how long 2028 will be ...

Maybe AMD needs more consistent competitors ... ... but it doesn't look good : they're going to be the absolute kings at DC ... and regarding client, well ... that's NOT their main interest ... ... ... ... so ... 2028 ...

🙂
 
Magras00 said:
So it's exclusively N3 based IODs for entire stack?

Not a cost optimized lineup for sure. Really looks like AMD is going all out but a N2P 12 core CCD and N3P IOD for lowest SKU can't be cheap. Either lower margins or everything gets a price increase nextgen.
Click to expand...
I think their line of thinking is that people who want cheap can buy Zen 4 or Zen 5. Or Zen 2 Mendocino if you are really strapped. They need to release an AM4 version of that. 🤣 🤣

Adroc is right, they going to make bank on Zen 6 unless Intel somehow pulls a rabbit out if its hat and forces them to lower pricing.
 
CakeMonster said:
This was widely discussed with Z3, people speculated that the 5900X would have a theoretical advantage in some scenarios over the 5950X. It was tested (I can't remember where) and nobody could find any scenario where it benefited.
Click to expand...
Those have the same L3. Not what is being talked about.

They're saying even if an application doesn't use more cores there is more shared L3 available in Z6 than Z5. This should help games a bit. Like a poor man's vcache.

I am replying in the appropriate thread if one wants to continue the discussion.
 
Josh128 said:
I think their line of thinking is that people who want cheap can buy Zen 4 or Zen 5. Or Zen 2 Mendocino if you are really strapped. They need to release an AM4 version of that. 🤣 🤣

Adroc is right, they going to make bank on Zen 6 unless Intel somehow pulls a rabbit out if its hat and forces them to lower pricing.
Click to expand...
I am pretty certain that IS their thinking.... mostly.

There has to be some overlap though. A 12c CCD that only yields 8 good Zen 6 full cores is STILL worth more to most people than a Zen 4 16c .... especially if it clocks higher than the Zen 4.

It will be interesting to see how the product lineup works out. AMD definitely has that roadmap out there for mobile showing exactly this kind of product mix... but the details (generation of cpu, # cores, # CCD's, amount of L3 etc) will be interesting.
 
gdansk said:
Those have the same L3. Not what is being talked about.

They're saying even if an application doesn't use more cores there is more shared L3 available in Z6 than Z5. This should help games a bit. Like a poor man's vcache.

I am replying in the appropriate thread if one wants to continue the discussion.
Click to expand...
AMD project +10% IPC for Zen 6 over Zen 5 in MLIDs now famous (but real) AMD tick-tock cadence leak slide. I think we can take that to the bank (plus or minus 1 or 2 percent). That is average geomean IPC uplift over a range of workloads. The kicker is what IPC uplift will we see for gaming. It could differ significantly from the geomean of 10%. Unfortunately, we probably wont know anything more than that for another 12 months or so.
 
Josh128 said:
AMD project +10% IPC for Zen 6 over Zen 5 in MLIDs now famous (but real) AMD tick-tock cadence leak slide. I think we can take that to the bank (plus or minus 1 or 2 percent). That is average geomean IPC uplift over a range of workloads. The kicker is what IPC uplift will we see for gaming. It could differ significantly from the geomean of 10%. Unfortunately, we probably wont know anything more than that for another 12 months or so.
Click to expand...
Not +10% but 10%+, wich is not exactly the same thing, even if the end result should be very close since 11% would be enough to fulfill the promise.
 
Abwx said:
Not +10% but 10%+, wich is not exactly the same thing, even if the end result should be very close since 11% would be enough to fulfill the promise.
Click to expand...
Stay calm, I did say plus or minus 1 or 2 percent. Things can always be better or worse than projections...

Jensen Calm Down.png
 
Tigerick said:
My prediction
Ryzen 7 - 12C, 48MB L3, N3P: ~$399
Ryzen 5 - 10C, ?, N3P: ~$299
Ryzen 3 - 8C, 32MB L3, N4P: ~$199
Click to expand...
Thunder 57 said:
$200 is too low. Bump all those by $50 would be my guess. Maybe even more (hopefully not).
Click to expand...
My guess is
$449 12-core R7
$389 10-core R7
$329 8-core R5
$279 8-core R5 Zen5 (9700X refresh)

Basically a small reduction in $/core, nothing crazy.
This pricing pushes people with less than 300 dollars to buy last gen 8-cores, which is what AMD really wants.
 
adroc_thurston said:
oh no they're not doing that.

then want you to buy Zen6 and pay respective ASP premiums on it
Click to expand...
maybe they can just keep selling it as R7 9700, but AMD likes to do pointless refreshes (5000 GT, XT), so not out of the question. No new silicon obviously, just renaming.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top