- Mar 3, 2017
- 1,777
- 6,791
- 136
No desktop will be a small, least relevant extension of mobile starting with Zen6.
Won't ever have any relation to server anymore.
Period.
No it's good, you're getting a real system design step-change for Zen6 DT for the first time since 2019.That's very sad trend
No AMD would get into commercial too if they could, but that's kinda tough.The real desktop is really only left for Gamers who use dGPU
No it's good, you're getting a real system design step-change for Zen6 DT for the first time since 2019.
No AMD would get into commercial too if they could, but that's kinda tough.
Well CPU + northbridge setups worked well before for AMD.It is surprising that the server chiplet technology worked as well as it did in desktop
Well it's a late'25 thing so yea.Zen 6 is really not that far away as it seems
venice?TR is not a priority but EPYC yes, Venice is another core count bump.
more and more I'm convinced he's an amd subcontractor or employee. he didn't correct me when I implied zen 5 is a go for a hard q2 release.Zen 6 is really not that far away as it seems. Zen 5 is all locked in on path to release, and most of the work has now likely shifted to Zen 6.
Venice = Zen 6 servervenice?
That's very sad trend, but I can see it.
Zen6 AMD server.venice?
I just talk to OEMs.more and more I'm convinced he's an amd subcontractor or employee. he didn't correct me when I implied zen 5 is a go for a hard q2 release.
But it's still chiplet.There’s no reason that Zen desktop has to be configured as a chiplet processor.
No.It would benefit as either a monolithic processor
RDL pile doesn't quite fit the definition of that but might do.or as tiled processor on top of an interposer.
more and more I'm convinced he's an amd subcontractor or employee. he didn't correct me when I implied zen 5 is a go for a hard q2 release.
It’s a very welcome change and pretty exciting to hear.
There’s no reason that Zen desktop has to be configured as a chiplet processor. It would benefit as either a monolithic processor or as tiled processor on top of an interposer.
Why? A flavor of foveros is what everybody will be doing in 5 years anyway.Intel's entire client side bet on the interposer may turn out to be one of the two:
- a fatal mistake
- a sub-optimal decision that made Intel's newest and latest parts less competitive in important client side segments.
There is no performance benefit to having 7700X as a chiplet processor. It's a tiny amount of silicon, as a monolithic processor it'd only be like 150-160mm^2 of silicon - there'd practically be no difference in costs.
There is.there'd practically be no difference in costs
Why? A flavor of foveros is what everybody will be doing in 5 years anyway.
This is my opinion (so take it for what it's worth) but Intel's problem isn't packaging. Both MTL/ARL will have the best packaging tech available for client products. They're not really behind in packaging - at all.Suppose you are right, but consider what happens in the first 4 of the 5 years:
In many segments of the client market, Intel will have to use older monolithic parts, that will be stuck in time (Intel has no money to fund another parallel development).
In the meantime, AMD will be releasing a new refresh every year, every year outdistancing Intell's old, legacy parts. (Raphael Lake ).
don't worry, you can name drop asus here. it's a safe zone.(And from Zen 4 launch, we see that a lot of things can go wrong, things that are of AMD making or outside influences)
That hasn't been remotely confirmed and was suggested to be a rumor based on a 3 year old rumor now.Lake 8P + 32E cores?
intel is dumping ht going forward. amd's design choice according to adrock thuston is that Zen will maintain 16 cores at the top end for now for zen 5 but zen 6 will include dense cores of the same type, these will have less cache but also have ht. Cores will increase with this zen 6 processor according to adrock a few hours ago.Seems like in MT, AMD will be overrun given the available power consumption budget. Because "efficiency" cores will have better perf/watt. So was AMD late to the party with regards to big.LITTLE on desktop? Since the rumors are they will go big.LITTLE with Zen 6, I get the impression they agree it's a good design for desktop too and not only mobile/laptop.
No such thing as ARL-S 8321.Just wonder what everyone's thought are with regards to how a likely Zen 5 top-of-the-line desktop AMD 8950X 16C will be able to compete against Intel Arrow Lake 8P + 32E cores?
36um 2.5D Si interposer is like 2020 tech.Both MTL/ARL will have the best packaging tech available for client products
They're just worse at making cores.Their issue vis a vis AMD is core design IP and their validation process takes much longer
Never said that.zen 6 will include dense cores of the same type
Okay, so a quasi PCIe (IF) connection through the substrate is 2023 tech?36um 2.5D Si interposer is like 2020 tech.
That’s basically what I said.They're just worse at making cores.
Plain and simple.
Low-swing single-ended links aren't PCIe...so a quasi PCIe (IF) connection through the substrate is 2023 tech?
Here are some recent sources mentioning 8P + 32E for desktop Arrow Lake:That hasn't been remotely confirmed and was suggested to be a rumor based on a 3 year old rumor now.