While others are still trying to conceal the lack of real innovation with AI logos and colorful bars, Intel is apparently currently relying on an old-fashioned virtue: learning from mistakes. This is new. Ori Lempel, Senior Principal Engineer in the Core Design team at Intel, has confirmed to KitGuru that they are now well beyond Cougar Cove and are working on something that is – get this – three generations ahead. Griffin Cove is the codename, and it certainly doesn’t come from a marketing department.
Of panthers, lions and other predators
As usual, the naming follows the zoo principle. After Lion Cove comes Cougar Cove, which is to make its debut in Panther Lake. What then follows – at least according to logic – is Griffin Cove, presumably as part of Razer Lake. This would also be the birth of another buzzword that can be backed up with bold letters and even bolder promises in films. But beyond the zoological playground, there is actually something substantial to report. Because: Intel has apparently understood that by consistently using its own production, i.e. IFS (Intel Foundry Services), it has tripped itself up. The admission may come late, but at least it is public. According to Lempel, the core design team is now “99% process node agnostic”. In other words: They now also take a look at what their neighbor can do. And if it works better, you just take that.
Agnostic instead of dogmatic: the reality check at Intel
What sounds like technical sophistication is actually a return to common engineering sense. Why commit to a process node that may sound good in theory, but in practice offers neither yield nor clock rates? Right, it makes no sense. This is precisely what Intel has now apparently recognized. And so future designs such as Griffin Cove will no longer be optimized exclusively for a single production node, but will remain flexible. Sometimes TSMC, sometimes IFS, depending on who is currently delivering – or at least not failing. In concrete terms, this means that CPU designs are simulated and synthesized on several nodes at the same time. Then they look to see where there is the least amount of noise – in terms of performance, efficiency or simply the schedule. The decisive factor is no longer whether it is the in-house node, but whether it works at all. This is new at Intel and could – if they are serious about it – actually make a difference.
Cougar Cove is through – now comes the future
According to Lempel, Cougar Cove is already finished and will be used in Panther Lake – planned for the second half of 2025. So anyone toying with a desktop upgrade should be patient. What will follow is the ominous Griffin Cove, which is far enough away to leave plenty of room for optimization (or excuses). There is one interesting rumor that is better handled with gloves at the moment: Razer Lake – presumably with Griffin Cove – could do away with e-cores completely. In other words, a purebred P-core architecture. Sounds reasonable, especially in view of the fact that many applications would prefer a fast core instead of many small would-be computing units anyway. But as I said: as long as nothing is official, it’s just coffee grounds.
Less faith, more physics
If Intel has finally understood that progress cannot be forced through marketing and in-house production alone, then this is not a technological leap, but a step towards reality. The process-node-agnostic approach is – soberly considered – not an innovation, but an overdue reaction to years of overconfidence. It remains to be seen whether Griffin Cove will really be the big hit or just another code name on the road to the unknown. But the approach of no longer forcing designs onto its own nodes at all costs is at least a realistic one. Intel is to be congratulated: not on a milestone, but on the realization that physics, yield and clock speed are non-negotiable – no matter how many times you try. Until then, everything will remain as it is: AMD continues to push, Apple calculates consistently, and Intel hopes that the future really will be better this time. They are already working on it. Allegedly.
Source: Youtube
15 Antworten
Kommentar
Lade neue Kommentare
Urgestein
Urgestein
Urgestein
Urgestein
Urgestein
Mitglied
Urgestein
Urgestein
Urgestein
Veteran
Urgestein
Urgestein
Urgestein
Veteran
Urgestein
Alle Kommentare lesen unter igor´sLAB Community →