WYP
News Guru
Last edited:
Alder Lake is Intel's secret weapon for late 2021 - Huge IPC improvement leaked
- Thread starter WYP
- Start date
AngryGoldfish
Old N Gold
Totally. I see no reason to buy Intel's next CPUs.
Alder Lake could be absolutely incredible.
AngryGoldfish
Old N Gold
It's correct. There are 16 cores, 8 of which do not have hyperthreading. It's this 'dual Core' thing Intel is moving ahead with. It's still one monolithic die, but technically there are two 'architectures' or 'Cores' inside the chip, one for performance and one for efficiency.
Dawelio
Active member
looz
Active member
They already released Lakefield, which has 1 strong core and 4 weak cores. Similar approach to phones, which have a mix of strong and weak cores. Latter can be used for background tasks for decent power savings.
https://ark.intel.com/content/www/us/en/ark/products/codename/81657/lakefield.html
https://ark.intel.com/content/www/us/en/ark/products/codename/81657/lakefield.html
Dicehunter
Resident Newb
So 8 for big computational tasks and 8 "smaller" cores for background things or low power tasks.
AngryGoldfish
Old N Gold
That's right. And only 8 of those cores have hyperthreading. Which means, 8 cores x 2 = 16 threads + 8 cores with no hyperthreading = 24 total threads. 16 of those 'cores' are real cores and 8 of those 'cores' are SMT.
AlienALX
Well-known member
It's about die space. 10nm is still pretty far from 7. So they will basically fit everything they can onto it, so they can compete with AMD. Pretty sure it will still be monolithic too (as in one die, not multiple).
Problem is by the time they do that AMD could shrink again and potentially double the cores on their desktop models. Intel's core technology is faster IMO* but with AMD having Zen so small now the potential to cram loads more on it is there.
* what I mean is imagine if Intel had their tech at say, 7nm, with 5ghz speeds. It would clearly be very fast indeed. The problem however to counterbalance that is the reason why they have not released any of their shrinks. The clock speed falls off a cliff (see also Broadwell E desktop CPUs that sort of never happened. Well did, for about a month, then didn't).
According to Intel, their 7nm is roughly equivalent to TSMCs 5nm.
The problem is, TSMC's 5nm is already used for large monolithic dies, particularly CPUs, in lots of shipping devices, and would be feasible for a run of consumer desktop CPUs of Zen's transistor count this year (Not that we should really expect that with the current shortages though), while Intel's 7nm is nowhere to be seen, and their 10nm won't come to desktop until the end of this 2021
So really, at this point I think you could say TSMC have the lead even by Intel's own admission, as they are practically not one "marketing node" behind, but two or three. We've have desktop-class 5nm TSMC CPUs shipping long before even Intel's next round of 14nm Desktop CPUs, and not tiny 5nm CPUs, Apple's M1's were huge by modern CPU standards.
The problem is, TSMC's 5nm is already used for large monolithic dies, particularly CPUs, in lots of shipping devices, and would be feasible for a run of consumer desktop CPUs of Zen's transistor count this year (Not that we should really expect that with the current shortages though), while Intel's 7nm is nowhere to be seen, and their 10nm won't come to desktop until the end of this 2021
So really, at this point I think you could say TSMC have the lead even by Intel's own admission, as they are practically not one "marketing node" behind, but two or three. We've have desktop-class 5nm TSMC CPUs shipping long before even Intel's next round of 14nm Desktop CPUs, and not tiny 5nm CPUs, Apple's M1's were huge by modern CPU standards.
Last edited:
