Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
Will this come to am4?
Nah, Zen 4 should still be released second half of 2022. This will be Zen 3+ on AM4So this must push Zen 4 that much further way then in terms of a time scale.
I don't understand. Zen3 chiplets give AMD the easiest route to AM5.Can't see AMD releasing a new socket for zen 3 chips, so yes.
Unless they have a need to release a pipe-cleaner to test out TSMC's 6nm I think that is unlikely.I doubt this will become a new generation of CPUs before Zen 4 is released next year, because they are still releasing Zen 3 CPUs (APUs) in August this year.
I think they will release 1 or 2 top high end products on socket AM4 (with high clocks), aimed at competing with Alder Lake CPUs in early 2022. Most Alder lake systems will still be using DDR4 RAM.
They will just but another X or T in the title, that seems to be how they market new CPUs of the same generation. There seems to be room for a '5900 XT' and '5950 XT' in the lineup.
I wonder if it would be worth fabbing them with TSMC's 6nm process?
Why do you think AMD needs to achieve Parity with Alderlake. AMD are already ahead of Intel, the 3D stacking coming in Zen 3+ will give AMD an extra 15% on average which should be enough to keep the performance crown. If you are referring to parity in using DDR5 then i do not see that as a problem. Being first to the party is not always a good thing, early DDR5 will have little to no difference in performance to DDR4 so AMD have time on their side, wait a year till DDR5 is a little more worth while.I don't understand. Zen3 chiplets give AMD the easiest route to AM5.
Glue Zen 3 chiplets to a DDR5 IO die and bam, you have a next generation platform. Then, to keep customers excited about it, glue a massive SRAM cache pool to chiplets (btw maybe AMD knows something we don't about DDR5 latency?).
That achieves parity with Alder Lake, both performance and release date. And buys some time to finish/polish 5nm production Zen 4.
AMD is ahead at this moment, but I expect Alder Lake is not a complete dud like 11th gen. There will be an IPC uplift, and Intel will be loud about it. DDR5, USB4 are nice buzzwords too. In any case, as Zen4 is currently projected end of 2022, AMD needs something earlier than that.Why do you think AMD needs to achieve Parity with Alderlake
You must be expecting some massive gains from Alderlake if you think Intel will catch up on this Gen + the 15% of the new AMD variants. PCIE 4 came out with Zen 2 and it didnt really mean much to the average Joe buying a PC and still doesn't for that matter, Heck i do not even think it means that much to the enthusiasts at this moment in time either, in fact even in terms of NVME most of us would never notice the difference between PCIE 3 and 4 except for some extreme usage scenarios. So i doubt a buzz word will make much difference to be honestAMD is ahead at this moment, but I expect Alder Lake is not a complete dud like 11th gen. There will be an IPC uplift, and Intel will be loud about it. DDR5, USB4 are nice buzzwords too. In any case, as Zen4 is currently projected end of 2022, AMD needs something earlier than that.
For the well-informed, yes.You must be expecting some massive gains from Alderlake if you think Intel will catch up on this Gen + the 15% of the new AMD variants. PCIE 4 came out with Zen 2 and it didnt really mean much to the average Joe buying a PC and still doesn't for that matter, Heck i do not even think it means that much to the enthusiasts at this moment in time either, in fact even in terms of NVME most of us would never notice the difference between PCIE 3 and 4 except for some extreme usage scenarios. So i doubt a buzz word will make much difference to be honest
Thats what I was saying. Unless AMD releases a +cache model, there is a danger Intel ends up on top until Zen 4.You must be expecting some massive gains from Alderlake if you think Intel will catch up on this Gen + the 15% of the new AMD variants
Intel said +20% for Rocket Lake over Comet Lake and everyone said to the critics of that: No no no.... this is a brand new architecture, Intel will crush AMD.Thats what I was saying. Unless AMD releases a +cache model, there is a danger Intel ends up on top until Zen 4.
Intel targets +20% IPC for Alder Lake. Meanwhile bigger cache +15% perf was shown only in select games (that are already known for liking large cache). For most benchmarks it won't be +15% IPC.
Zen3+ on AM5 makes sense for feature parity and to reduce pressure to release Zen4.
More digging around on the AT forums:
https://forums.anandtech.com/threads/speculation-zen-4-epyc-4-genoa-ryzen-6000.2571425/post-40514844
quoting this:
So basically saying the Zen3 CCDs already were designed for this from the beginning.
Possible just took longer to get to stacking: maybe the B2 revisions are something to do with getting this working?
Things like release cycles, TSMC's and their packaging ability schedule all can slip.From the Twitter string, There are TSV's all over the Zen 3 chip, yes Zen 3 was designed to be 3D stacked right from the start, and not just the Cache.
Things like release cycles, TSMC's and their packaging ability schedule all can slip.
So another big takeaway from this is great risk management from AMD since they could ship Zen 3 CCDs without that part being ready.
https://www.anandtech.com/show/16535/intel-core-i7-11700k-review-blasting-off-with-rocket-lake/7Intel said +20% for Rocket Lake over Comet Lake