If you see it on the market, you know it'll have a low-ish margin compared to whatever else they could do with the same silicon. They would have to price it outside of its own market otherwise.
Not if they raise prices compared to Vermeer's MSRP.
Not sure that they'll do that. It's possible. But really, do you think they can make the same profit per mm2 off of Zen3D as they can Milan? They're already sacrificing wafers to Vermeer refresh (presumably to maintain OEM contracts and keep the market supplied).
Not sure that they'll do that. It's possible. But really, do you think they can make the same profit per mm2 off of Zen3D as they can Milan? They're already sacrificing wafers to Vermeer refresh (presumably to maintain OEM contracts and keep the market supplied).
I fully expect AMD to increase the MSRP according to the performance improvement Zen 3D brings. That may well be still a lower margin than with stock Zen 3, but to call that a loss leader definitely is an exaggeration.If you see it on the market, you know it'll have a low-ish margin compared to whatever else they could do with the same silicon. They would have to price it outside of its own market otherwise.
I'm kind of assuming that it is quality binned and Vermeer is the leakiest of the 6/8 core dies.
I fully expect AMD to increase the MSRP according to the performance improvement Zen 3D brings. That may well be still a lower margin than with stock Zen 3, but to call that a loss leader definitely is an exaggeration.
Depends what kind of volume discounts they give to OEMs and hyperscalers.
Those are the dice I expect them to throw at Vermeer Refresh.
Zen3D is just Vermeer Refresh with some extra cache bonded to it (more or less). It's the extra cache that eats up wafer supply that isn't being used for CCDs where AMD may be losing money (unless they're going to have extra cache dice left over from Milan-X?).
You have to include Threadripper Pro in that too.
Unless they're going to launch a Milan-generation version of TR Pro soon?
Yeah that's what I'm expecting. Could be tomorrow even.
Nobody knows when AMD is going to release any product at all. We don't even know if Zen3D will hit the market, period, of if Zen4 will drop in April, or both, or . . . whatever.
Yeah Zen3D would be neat, but personally I would rather see them focus on Zen4 on the desktop.
Meanwhile, the idea that AMD has to have Zen3D out now, instead of November or December, is a bit stupid.
Those are English words grouped together with a period at the end. I'm not clear on what thought they are trying to convey.
Unless you are an investor margins and market share should be of little concern. The price/performance of the products is all that consumers should be worried about. Intel and AMD have wildly different business models at this point in time. TSMC is intel's real competition. AMD's resurrection has as much to do with Dr. Su's leadership and mission statement (executing consistent incremental performance gains each generation) as it does with Intel's mistakes. Intel soiling the bed for several years certainly helped but Dr. Su and team still deserves a lot of credit for the massive performance gains in consumer desktop we've seen over the last few years.
The problem with Zen 4 is that it uses TSMC N5 node, the same as Genoa and RDNA 3 cards. So it will be a crowded field, and the one that is likely going to be crowded out of it is desktop Zen 4.
Zen 3D, OTOH, is on N7, which is plentiful, and if AMD didn't shoot itself in the foot (by missing the window to release it), could have been a successful product for gamers, and ADL would have been another RKL...
I fully expect AMD to increase the MSRP according to the performance improvement Zen 3D brings. That may well be still a lower margin than with stock Zen 3, but to call that a loss leader definitely is an exaggeration.
MSRP will be aligned with the equivalent Intel parts. If Alder Lake had not performed as well as it does, they could have set MSRP to whatever they wanted to. Now? The 5900X with v-cache will be priced close to the 12900k more than likely. Other chips will be priced higher or lower depending on where they are in the stack relative to Intel.
MSRP will be aligned with the equivalent Intel parts. If Alder Lake had not performed as well as it does, they could have set MSRP to whatever they wanted to. Now? The 5900X with v-cache will be priced close to the 12900k more than likely. Other chips will be priced higher or lower depending on where they are in the stack relative to Intel.
If they make sure that it works on B450 there are going to be plenty of people on a 2xxx part that might consider upgrading to the top of the line AM4 part to see them through for a few years before jumping onto a DDR 5 platform and in that situation even an overpriced CPU can be comparatively cheap to the alternative due to saving on motherboard costs.
Agree on lower end parts/low core parts. 16 core and more for desktop needs bandwidth. Looking forward to see what 3d cache extra bandwidth does for 5950x.I am of strong belief that 5950x is more bottlenecks by bandwidth than power.It's going to take a long time for the benefits of DDR5's additional bandwidth start to make a big impact on performance, especially on mid to low end parts that don't have iGPUs, and especially when compared to the higher end DDR4 kits that are available.
Who knows, the one layer of 3D cache AMD launches may well be comparable in performance improvement as the doubled L2$ in Zen 4.I don't see how they can release a next gen zen4 Epyc without this 3D cache.
Sure they can just add it all on the die but it's cheaper and more dense 3D stackedWho knows, the one layer of 3D cache AMD launches may well be comparable in performance improvement as the doubled L2$ in Zen 4.