- Mar 3, 2017
- 1,687
- 6,243
- 136
N7p is DUV and is fully design compatible with N7 g1.N7P and N7+ were EUV nodes with different design rules than N7
Then single V-cache die shared between two CCDs as a huge L3 cache and remove the L3 from each CCD.If you people watched cpu Jesus tour of amd labs you will know why we won't see consumer dual ccd with 3d cache on both ccds.
Hint cache coherence between ccds...
You love to kill dreams, don't you?Welcome to the future™.
So completely re architect coherency. That l3 holds the tags for all the l2s so now probes need to go to every core.Then single V-cache die shared between two CCDs as a huge L3 cache and remove the L3 from each CCD.
I am honestly surprised we haven’t seen another XT chip from AMD. I imagine at least some chips are capable of hitting 6ghz.
They don't care.
Refresh cycles are X3D now.
Yeah this would be the holy grail, unfortunately this means putting the cache die under the core and that IMO really doesn't make sense unless they're redoing the packaging anyway - so most likely we'll have to wait for Zen 6 for this limitation to be removed.
Unless AMD surprises us with a super-secret project at X3D release as they did with the initial 3D cache CPUs ...
I must say, I'm a bit disappointed that AMD decided to keep the Zen 2 packaging for the Zen 5 generation. Event on the high-end consumer CPUs, cmon, It's been 4 generations already!
If someone would have asked me in 2019, when will AMD move to more advanced consumer packaging on premium parts, I would have quessed with Zen 4 (with a new socket) or with Zen 5 at the absolute latest ...
N7p is DUV and is fully design compatible with N7 g1.
I'm confused, are you saying an XT refresh would have a bigger halo effect than an X3D refresh? The Zen 3 refresh is still selling pretty well I heard.Halo effect matters, and AMD should have cared when Zen 3 lost the performance crown to Alder Lake.
Going forward, when it looks like Zen 5 will be in comfortable lead in single thread, it will be sufficient to have just regular Zen 5 and then refresh with X3D.
I'm confused, are you saying an XT refresh would have a bigger halo effect than an X3D refresh? The Zen 3 refresh is still selling pretty well I heard.
Gonna save myself the headache of thread priority again and get the vanilla Z5 16c part. Bonus is more value over time as its (probably) released before the X3D part. Downside is probably BIOS immaturity.
Absolutely.You love to kill dreams, don't you?
No Zen3 was still selling like hotcakes past ADL.Halo effect matters, and AMD should have cared when Zen 3 lost the performance crown to Alder Lake.
No.But with the new packaging, it would also be possible to connect 2 Zen 5 dies in Strix Halo to each other.
oh man.If it translates to the new Zen 5 RDL packaging, and the 2 ports where the GMI is attached can be exposed to this new packaging, it might be a possibility.
Yes, TSM marketed it as 2nd gen N7 iirc.It is?
It's a mature platform, nothing changes but CCDs themselves.Downside is probably BIOS immaturity.
wasting time on absolutely irrelevant products is not how AMD operates.AMD could have done both, N6 refresh of Zen 3 and also X3D
They could have handed the job over to interns. No precious time of actual engineers wasted.wasting time on absolutely irrelevant products is not how AMD operates.
128MB X3D. 64MB seems pretty low.Dream Strix Halo?
8C Zen 5P, 16C Zen 5C/D, 40 CUs, 64 MB X3D cache on the that is available for the CPUs and GPU.
Yes.128MB X3D. 64MB seems pretty low.
No Zen3 was still selling like hotcakes past ADL.
Not like a 100MHz bump changes anything.
wasting time on absolutely irrelevant products is not how AMD operates.
No chance that a project that would end up as an actual AMD product would be handed over to interns.They could have handed the job over to interns. No precious time of actual engineers wasted.
If they had started on it when Zen3+ was originally decided, the interns would have completed it by now and we would be seeing a CPU launch on AM4 right about now. Nothing exciting but at least, it would be something new for the AM4 crowd. I'm not saying that they would have done it all by themselves but I think they could have managed 80% of the work, with help from their seniors. But of course, that's just me assuming that AMD interns are far above in capability than the interns everywhere else.No chance that a project that would end up as an actual AMD product would be handed over to interns.
No, interns are never given that much responsibility. They barely know how to install a toilet correctly (metaphorically speaking) let alone build the entire house, even if given the blueprints and all the right materials to start. The senior engineers would be spending more time fixing their mistakes than had they just done it themselves.If they had started on it when Zen3+ was originally decided, the interns would have completed it by now and we would be seeing a CPU launch on AM4 right about now. Nothing exciting but at least, it would be something new for the AM4 crowd. I'm not saying that they would have done it all by themselves but I think they could have managed 80% of the work, with help from their seniors. But of course, that's just me assuming that AMD interns are far above in capability than the interns everywhere else.
ughhh no.200-300 MHz would have been enough for Zen 3 to somewhat tie Alder Lake, it would prevented Pat from making his "rear-view mirror" comment
It didn't tie ADL-S, it vaporized it, lmao.And then Zen 3 X3D would have been again a Halo product, instead of a sort of a tie with Alder Lake.
what.The loss Halo effect cost AMD dearly in sales. One of the contributors to AMD loss of 66% of revenue in client.
that's all laptops, -S has no relevance to AMD's revenues or bottom line.Loss of 66% of client market is a clear indication that the way AMD operates results in sub-optimal outcome.
Ongoing effect of this is multibillion loss of revenue. AMD lost far more in client revenue than gained in server revenue since AMD lost Halo effect to Alder Lake.
what.
The client decline is all laptops since channel was stuffed with garbage.
TIL I learned a 1% lead is vaporizing the competition.It didn't tie ADL-S, it vaporized it, lmao.