No one is sure. But it's been rumored for a while now. Add on the fact that Intel confirmed something on ARL is going to be using external 3nm, well it at the very least is a possibility that it could be the CPU tile.
I really hope desktop MTL if it gets released, is not good enough to make the RPL refresh look bad. But the more I read about the huge L4 cache, the more my hopes seem en route to being shattered.
mtl will have a frequency regression and missed cores. it's a cleaning generation. People get hung up on the gen naming but it doesn't matter if rpl refresh gets it's own generation number or not. what matters is performance. mtl-s may be a very short lived generation before arl s comes along whether very late this year or very early next year alongside mtl-s. if I had to throw a month I'd go with March and then arrowlake to follow before end of summer. none of this is aided by the intel fab problems being brought up all the time. Intel's length 10nm debacle has half the people on the internet convinced intel can't get their fabs in order and never could.
Hulk
Diamond Member
- Oct 9, 1999
- 5,138
- 3,727
- 136
If I had to guess, and I do, then I'd postulate that MTL is a way to increase profits by reducing cores and process size. If they can squeeze a lot of these on each wafer then they can get really competitive with pricing when it comes to mobile and low to midrange desktop with Raptor Refresh taking care of the high end.
I'm getting more and more convinced that 14th gen will be MTL/Raptor Refresh as indicated above.
But of course there will be a new rumor tomorrow that will throw a wrench into everything. And that's the fun of this
I'm getting more and more convinced that 14th gen will be MTL/Raptor Refresh as indicated above.
But of course there will be a new rumor tomorrow that will throw a wrench into everything. And that's the fun of this
RPL Refresh could also mean that Intel 4 volumes are low so Intel needs something like a 14600K to be their best selling gaming CPU.
no unless you attend the school of cheech and chong would that make sense. not accusing you specifically mister hulk. arrow lake will cover the full spectral range of of i3 to i9. your argument only makes sense if raptor's refresh is stuck to the i5 to i9s on k models whereas mtl will server as a lower end product again not making a whole lot of sense with its own i5 put on offer unless raptor's refresh is stuck to i7 and i9, because the lower end current raptor normals are alderlake carryovers.
my prior post timeline for arrowlake makes some sense for a 1h or trailing 2024 release only because amd is due to release their zen 5 around that same time window.
my prior post timeline for arrowlake makes some sense for a 1h or trailing 2024 release only because amd is due to release their zen 5 around that same time window.
coercitiv
Diamond Member
- Jan 24, 2014
- 7,355
- 17,425
- 136
It already kinda' threw a wrench. Raptor and Meteor are not as easy to marry as one would think. We can still get an arranged marriage, but they clearly were not meant for each-other: memory support, chipsets, sockets... the paperwork alone will be a nightmareBut of course there will be a new rumor tomorrow that will throw a wrench into everything. And that's the fun of this
It should be pretty interesting to see how they tweak the RPL for the refresh to offer more performance. Maybe lower cache latency? Base DDR5-6400 support? Or best of all, RPL shrunk to Intel 4! They should have done that as a backup anyway, coz I read that going forward, Intel was going to have multiple backup plans. It would also mean that maybe there's a MTL design backported to Intel 7, all ready to go as soon as the Intel fab engineers give up on Intel 4
Arrow Lake cores will be built on Intel fabs. The 3nm Intel mentioned was for the GPU and other bits.
Intel’s GPU IP is built on a TSMC process. Porting it to an Intel process would cost a lot of money for no gain. By leaving it at TSMC Intel will have ample capacity to make it’s next-gen chips.
Unless Intel suffers a really large setback (they haven’t), you will never see Intel client cores produced on an outside process.
Intel 4 with backside power delivery. Seems like a good sign that Intel may be ahead of schedule on delivering BPD technology if they are testing on Intel 4, since they originally claimed that they would test backside power delivery on Intel 3.
Obviously these are internal tests and almost certainly won't show up in actual products until Intel 20A, but still interesting.
Specifics about the results of BPD are in the excerpt and will prob go into more depth at the actual event (VLSI)
I don't think Intel 4 vs 3 is a real difference here. The key detail is pairing p1276 finfets with p1278 metal stack (or just backside?)
Intel 4 with backside power delivery. Seems like a good sign that Intel may be ahead of schedule on delivering BPD technology if they are testing on Intel 4, since they originally claimed that they would test backside power delivery on Intel 3.
Obviously these are internal tests and almost certainly won't show up in actual products until Intel 20A, but still interesting.
Specifics about the results of BPD are in the excerpt and will prob go into more depth at the actual event (VLSI)
I.m confident they.ll reach 7ghz with this technology.. 10ggz not far away lol
Intel 4 with backside power delivery. Seems like a good sign that Intel may be ahead of schedule on delivering BPD technology if they are testing on Intel 4, since they originally claimed that they would test backside power delivery on Intel 3.
Obviously these are internal tests and almost certainly won't show up in actual products until Intel 20A, but still interesting.
Specifics about the results of BPD are in the excerpt and will prob go into more depth at the actual event (VLSI)
Nobody is sure, however if they plan to bring ARL-S in H2 2024 it is unlikely they can do this with their own 20A, this is the problem. Furthermore Raichu claimed that ARL-S is on TSMC 3nm and ARL-P on 20A...and he has a very good track record. He was the first who claimed MTL-S being limited to lowend-midrange almost a year ago. Also we knew Intel planned Lion Cove+Skymont on TSMC 3nm from the leaked old roadmap ARL-P, means the IP is certainly compatible with TSMC 3nm. Up to now they couldn't simply use TSMC for their core architecture.
DrMrLordX
Lifer
- Apr 27, 2000
- 22,903
- 12,974
- 136
All links must be accompanied by your own commentary.
Last edited by a moderator:
Arrow Lake on Intel 3? It must really be Intel 4+ in essence if they can follow up MTL with a high performance architecture on a new node so soon. E-core on 18A sounds nice. Wonder if they will be restricted to 4 GHz speeds or if they will be the first E-cores to hit 5 GHz?
So looking at this, he calls MTL 7nm, it just says ARL 3nm. This entire thing just seems confusing, I wish I could see the entire thing unredacted. This could be interpreted as like 3 things:
Does he mean Intel 3nm (which is actual Intel 20A/18A?)
Does he mean TSMC N3?
Does he mean Intel 3?
SteinFG
Senior member
- Dec 29, 2021
- 732
- 869
- 106
ARL is on "Intel 3"
MTL is on "Intel 4", which is a rename of a node previously known as "7nm"
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 22K
-
News NVIDIA and Intel to Develop AI Infrastructure and Personal Computing Products- Started by poke01
- Replies: 384
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
Facebook
Twitter