Intel Upcoming Lunar Lake-U 17W Processors to Enhance Multithreaded Performance Significantly
Click here to post a comment for Intel Upcoming Lunar Lake-U 17W Processors to Enhance Multithreaded Performance Significantly on our message forum
schmidtbag
Kind of odd how they seem to be implying HT was slowing them down... Really gets you to wonder why they stuck with HT for so long if getting rid of it was so obvious of an improvement.
H83
The usual marketing BS.
tunejunky
when you have crappy e-cores (as opposed to AMD slo-cores) that have always been a nightmare for Windows Scheduler, you "do what you need to do".
but marketing has been the curse at Intel. at least Gelsinger has his priorities straight. guiding Intel is like piloting a supercarrier - it takes miles to make a turn.
H83
Although i dislike BigLittle implementations, specially on desktops, i think it`s unfair to say that E-cores are crappy.
For me, they are the most interesing stuff Intel released lately, i expect Forevos packaging to be even better, but they shine more alone than as sidekicks for super hungry CPUs...
Undying
I really hope Intel shows something this year in GPU and CPU department lately they are just a talk. I see no no reason to leave amd platform anytime soon.
tunejunky
swarm computing... i still like the AMD design better in that....it's better at everything
but there is definitely a market for Lunar Lake
tunejunky
for me definitely for at least three years. then their node advantage may sway me, but their efficiency must go up, way up for me to switch
schmidtbag
Well, their node advantage is another thing they keep making statements about yet we don't see much progress. Considering their struggles with 10nm, I'm not feeling great about future nodes.
In my opinion, Intel is in a pretty bad situation. They don't have a single product that is the most appealing in any market. Right now they only thing they've got going for them is early AVX10 support (if you even care about that), a decent NPU (but perhaps not as good as a GPU or ASIC), and good single-threaded performance at high clock speeds (in a world that is starting to care less and less about single-threaded performance).
Neo Cyrus
I've said it before, and I still think it's true, if Pat Gelsinger, a real engineer, can't save Intel, then I don't know if anyone can. Putting Pat in charge was the best thing Intel have done in an eternity, let's just hope it wasn't too late. For all our wallets' sakes.
tunejunky
i understand that you're looking from afar.
just last week Intel made a show of "unboxing" in Beaverton, a complete ASML cutting edge EUV lithography machine that no one else is getting for five years unless ASML doubles their output (which ain't happening in the timeframe).
the Beaverton Fab is the "academy" or "teaching fab" where all of the training for new processes and working out the uArch wrinkles that are inevitable at one point of a new process. this means that Arizona will most likely get the new set unless Intel has other plans
schmidtbag
That is true and a good point, however, I get the impression it's not so much the machine that matters but how you use it. As far as I understand, the machines Intel had before weren't really any different than what TSCM or Samsung had, yet TSMC managed to reach 7nm years sooner than Intel reached 10nm (and yes, I am aware their sizes are comparable since it depends on how you measure the area). Meanwhile, I wouldn't be surprised if Intel's 14nm was better than anyone else's equivalent by a significant margin. From what I recall, it was better than GloFo's 12nm.
So - while I think Intel will have an easier time getting below 10nm on this new lithography, I suspect they won't get as small as they'd like when they'd like. When they do reach their target size, I suspect they won't hold that lead for very long. Of course, I very much could be wrong - perhaps they learned something with their 10nm process that applies to smaller nodes and therefore will achieve their goals sooner than expected.
tunejunky
yes, they were different. they were running a lot of Applied Materials lithography, like GloFlo, but more recent make.
part of their issues was the fact that they had to search for different tooling, and ya' know, the bureaucracy of a mature corp where marketing had been winning all fights for 10 years.
that's the backstory to their embrace (damn near partnership) of ASML after buying some of the last gen tooling
tunejunky
it will be awhile. TSMC has been making brave faces, after all they have their processes in addition to the most modern across-the-board fabs. they've sold on their equipment for old nodes so they're not interested in legacy fabbing.
however, Intel has secured new business worth billions to start and potentially hundreds of billions - the US DoD
this is not just the "secure computing" initiative, it's DARPA, and it's also both cutting edge and legacy nodes.
you can guarantee that any involvement with DARPA will entail node superiority as a subject of research.
in any case TSMC and their CoWoS are well ahead of Foveros, which is good just not as versatile. so non-defense foundry clients aren't going anywhere anytime soon
Glottiz
Intel new gen with new socket coming in 6-7 months. I see no reason to leave Intel platform any time soon. This CPU bickering doesn't even matter at the end of the day. Intel or AMD all within few percentage of each other. X3D better for gaming, but terrible for productivity. Intel overall better for mixed usage. Realistically GPU is the only thing that matters if you have somewhat decent CPU from last 3 years or so.
H83
Securing the US DoD contracts might not be related to a case of having the better nodes but a case of national security because there`s an eery feeling that China may invade Taiwan sooner or later, so it`s better for the US to move the fabrication of essential hardware indoors. And it will also boost the national economy.
So for me, this move is more political than technical, not a critique because i would do the same even if Intel fabs and CPUs were total crap.
Personally, i don`t put too much stock on Intel after all their failed promises, so i will for their future releases to make my judgement.
Undying
Im talking about people building a gaming PC now. Like @pegasus1 said there is no reason to go Intel atm. AMD has a platform longevity, CPUs are alot more efficient, more performant in games, less expensive. Everyone who aksed me i always recommend going AMD. Many of my friends went that route becouse of that.
mackintosh
You can't go wrong with either for gaming right now. It's all a matter of how much you're willing to spend to get a motherboard with the features you need. CPUs literally don't matter. Any of the X3Ds or the 13600s/14600s and above will do. Productivity is a different matter, and there are many different use cases to consider there that require a tailored approach.
Glottiz
Maybe from your extreme diehard fanboy eyes it appears that way. But there are plenty reasons to go Intel right now. AMD doesn't even have something comparable to offer in 12400F, 13400F, 14400F price range. And in higher end CPUs maybe not everyone enjoys dealing with AMD jank and relying on 20 beta BIOS updates to fix something.
Undying
I can tell you didnt used AMD system for a long time so you certainly must think so. I never had problems since zen+ and im on zen4 now.
Glottiz
Well I never had problems with Intel. So why do you have to keep pushing AMD so hard? Why does it matter so much for you? It's like saying "well there is no reason to buy anything except Toyota". All cars cost similar and do similar thing. Just let people buy what they want. When you push one brand so hard it just makes you look annoying. And if it makes you feel better I also have a laptop with Ryzen 5500U, so I used AMD plenty (including both consoles which also run AMD).
