What I am reading about the High NA machines looks very promising as well. From 40 passes to less than 10 for 18A samples. Of course, it is possible (highly likely) that to achieve more density than 18A, 14A will have way more than 10 passes.... but it might be worth it.That's what Novalake is for, brother.
Neither 18A or 14A are hNA.What I am reading about the High NA machines looks very promising as well. From 40 passes to less than 10 for 18A samples. Of course, it is possible (highly likely) that to achieve more density than 18A, 14A will have way more than 10 passes.... but it might be worth it.
I believe 18A is only being used to prove out the equipment. It is my understanding that 14A is targeting High NA.Neither 18A or 14A are hNA.
Found this through ServeTheHome:Neither 18A or 14A are hNA.
I found this and a maybe another that said the same thing; however, in all fairness, this was April 2024. I think the target for 14A is like 2027 - 2030?Found this through ServeTheHome:
Yeah.Zen 6 DT will feature a whole new IOD, right?
The cheapest one which is 3.5 and you can plaster display/media cores irregardless of the underlying shader microarch.What kind of GFX IP will be there? RDNA 3.5 or RDNA 5? Given the RDNA 4 vastly improved new Media Engine, would be cool if it was GFX13.
No, its Zen.Is this an intel thread? I hope zen6 moves to 24/48 giving us a 12 core single with x3d cache.
I thought that it was pretty much decided that Zen 6 would have a 12 core CCD .... nearly guaranteeing a 12 core single with X3D in the future.... and of course an 2 CCD 24/48.Is this an intel thread? I hope zen6 moves to 24/48 giving us a 12 core single with x3d cache.
If Zen 6 is better at gaming that's what moves DIY. E cores add scheduling issues and yeah it's good for cinebench but I don't get the impression many people care. Best single thread and very healthy multi core performance is best. What's moving the DIY market now is the best gaming performance.I thought that it was pretty much decided that Zen 6 would have a 12 core CCD .... nearly guaranteeing a 12 core single with X3D in the future.... and of course an 2 CCD 24/48.
As good as that sounds though, Zen 6 will be facing a 52 core Nova Lake (16P, 32E).
It kinda seems like things are getting a bit silly on the core counts though. Don't you think? Sure, for DC and Workstation, more cores please.
With this level of core count, doesn't the desktop part begin to encroach on the workstation part (Threadripper)? Of course, maybe not since a Zen 6 Threadripper would be a monster with 12 core CCD's!
I wish it weren't so damn far in the future though. It's hard being a computer enthusiast these days. I think I am going to need another trip to my therapist to ease my anxiety on the slowing development trend .
I don't think 48 cores on a client 2ch socket is a good idea.As good as that sounds though, Zen 6 will be facing a 52 core Nova Lake (16P, 32E).
Not sure you've been following the forum that closely adroc. It's a fairly common opinion that desktop platforms should have double the cores of current platforms and double the memory channels. I understand this is a non-niche use case.I don't think 48 cores on a client 2ch socket is a good idea.
I highly doubt that will matter much, if at all.As good as that sounds though, Zen 6 will be facing a 52 core Nova Lake (16P, 32E).
I don't think 48 cores on a client 2ch socket is a good idea.
High-clocked 24 cores of Zen 6 would also posses a b/w problem for 2ch DDR5.
I think for most AM5 users, they won't even look at Nova Lake cause you need to change your motherboard. So gamers will move to 10800X3D and workstation users will move to 10950X3D
I mean it makes sense.intel is definitely gearing up 3D cache research aren't they? I read something about some intel spokesperson neither confirming nor denying they have some kind of 3d prototype going on
but something tells me panther/nova lake will have to make significant compromises so altogether they still cant compete with 9800x3d
Agreed. Fair to reasonable chance of it being cancelledI highly doubt that will matter much, if at all.
That thing will
- be rather expensive to make compared to Medusa Ridge, so I don't see Intel pricing it any lower than $999, possibly higher
- either clock fairly low (all-core at least), or have a workstation/server-like TDP (or even both)
- be of little additional value for gaming and most everyday tasks
- either need its own 4ch socket (expensive + possibly short-lived), or be fairly bandwidth-constrained
- still lose to proper Threadripper models in real workstation/server tasks
Frankly looks like an expensive-yet-low-margin, neither-fish-nor-flesh product to me, just so they can claim the MT win when Medusa 12C-X3D and 24C beat the regular 8+16 NVL in about everything relevant to desktop users (let alone mobile).
It kinda seems like things are getting a bit silly on the core counts though. Don't you think? Sure, for DC and Workstation, more cores please.
Well, that is kina my point.Moving to quad channel would mitigate any memory bandwidth concerns.
Honestly, I think desktop would be great with 32 cores. It would allow AMD to resolve the odd HEDT problem that Threadripper has.
Example:
Athlon - Dual channel memory - Up to 8 cores -20 PCIE 4.0 lanes
Ryzen - Quad Channel Memory - 8 - 32 cores - 40 PCIE 5.0 lanes (example prices: $1299 32c, $999 24c, $699 16c, X3D +$100)
Threadripper - 8 channel memory - 32+ cores - 80 PCIE 5.0 lanes
EPYC - 12+ Channel memory 32+ cores 80+ PCIE lanes.
Of course it probably won’t ever happen, but one can dream. 🤣
My old Zen 1 Threadripper had quad channel boards for $300-$500. While that was DDR4 and before inflation, I imagine they could get it done for $600-$800. The X chipset could be quad channel while the B chipset could be dual channel. The A chipset could be similar to B except designed for Athlon.
uh no no one's paying for that dawg. get real.Ryzen - Quad Channel Memory - 8 - 32 cores - 40 PCIE 5.0 lanes (example prices: $1299 32c, $999 24c, $699 16c, X3D +$100)