I know I'm being cynical when my first thought upon reading this comment was that clearly the real return to form for Intel was adding in more planned obsolesce to force more consumer hardware turnover. Because while Intel isn't completely hopeless in the GPU market nowadays thanks to Arc, without driver support all that extra GPU horsepower is completely worthless.Improved integrated graphics is wonderful and all, but they just discontinued mainstream driver support from chips they're still selling as 14th gen. Everything with integrated graphics that's not Lunar Lake or Arrow Lake is already on a legacy branch.
The one they fired last year?Is this the work of Pat Gelsinger coming to fruition? He seem to be the right kind of CEO to get the company back on track from a tech perspective. Assuming this is truly a return to the Intel of old.
Yeah well, this part is fairly typical. But I really doubt ONLY Panther Lake will get Core 300 branding... so any "finally no more weird mix" is premature.which will presumably come to market under the "Core Ultra 300" banner, but the chips will no longer use a hodgepodge of mixed-and-matched technologies. All Panther Lake chips are still assembled with Foveros;
Yes, new architectures take 36 to 48 months to bring to market. New CEO has been in charge for less than 12 months. Y'all do the math on which CEO influenced these chips the most.The one they fired last year?
I will wager pennies to nickels that when Win12 arrives, an NPU will be mandatory.I feel....nothing. I find myself these days completely unfazed by Intel's CPUs, I feel like the well has run drier than a bedouin's sandal.
...no, I do actually feel something; I am annoyed by the forced inclusion of NPUs. I mean, sure, everyone's including them now, but I, for one, wish they'd just leave those out entirely or used the space for something more useful.
Silicon wafers are restricyed, but advanced nodes like IN3, IA18 or TSMC N2 are "even much moar more" restricted/limited.I though that silicon wafers were expensive and limited, does Intel has an overproduction of silicon wafers that they can waste all that area in the filler tiles?
NVM, all the parts aren't made from the same wafer.
The one they fired last year?
Yes as Williamyf said there is a fair amount of lead time involved for a new process.Yes, new architectures take 36 to 48 months to bring to market. New CEO has been in charge for less than 12 months. Y'all do the math on which CEO influenced these chips the most.
Can that be considered even more reason to abandon Windows? Not that it matters - because of Windows, all of the laptop/desktop x86 chips (and the similarly targeted ARM chips) will have an NPU regardless.I will wager pennies to nickels that when Win12 arrives, an NPU will be mandatory.
Will 50TOPS be enough? Who knows. But a processor with an NPU has a slim chance to be compatible with Win12, a processor sans TPU will not be compatible from the get-go.
It might; Arrow Lake Refresh is basically just going to be minor clock speed bumps it seems, and Nova Lake is supposed to be late enough into next year that it probably gets Core Ultra 400 branding.Yeah well, this part is fairly typical. But I really doubt ONLY Panther Lake will get Core 300 branding... so any "finally no more weird mix" is premature.
Drowning in lakes, indeed.
They did fire him, but this will actually be the first product he would have been responsible for based on Intel's timelines if he hadn't been fired.The one they fired last year?
Interesting and thank youIf this was Pat's work they would have already outsourced the whole thing to TSMC. 18A was announced years ago, but the determination to actually use it is new management trying to get their fab business back in order.
After switching to AMD, I kind of wish they would drop the P-Core/E-Core nonsense. I don't feel like it has provided any value at all and honestly has caused more problems that it has solved. If (or until) it becomes a x86 architectural standard, and modern operating systems universally support it, just drop it from the mainstream.
What’s ailing Intel will take 5-10 years to fix…… It won’t be quick, and there will be no shortcuts. And without an in-house OS, they will be a second rate chip company because times (technology) have changed and moved on, should have taken Steve Jobs seriously 18-20 years ago.Yes, new architectures take 36 to 48 months to bring to market. New CEO has been in charge for less than 12 months. Y'all do the math on which CEO influenced these chips the most.
Nope. Device side AI inference has its uses, even (or rather, more so) on linux. One needs the NPU to do that power-efficientlyCan that be considered even more reason to abandon Windows? Not that it matters - because of Windows, all of the laptop/desktop x86 chips (and the similarly targeted ARM chips) will have an NPU regardless.
They can turn things around the problem is no one wants to hear it’s 5 to 10 years.Weirdly enough, MacBook Airs provide better value for money than most, if not all, ultrabooks (especially with the educator discount). Windows has annoyed me enough to the point where I'm planning on sticking with macOS for the foreseeable future. And AMD is eating Intel's lunch in the desktop space as the article points out. So for the time being, I am not terribly interested in Intel's products. We'll see if they can begin to turn things around.
If this was Pat's work they would have already outsourced the whole thing to TSMC. 18A was announced years ago, but the determination to actually use it is new management trying to get their fab business back in order.
Why would they need their own OS? Both AMD and Qualcomm are successful, cutting-edge chip companies and neither has their own OS.What’s ailing Intel will take 5-10 years to fix…… It won’t be quick, and there will be no shortcuts. And without an in-house OS, they will be a second rate chip company because times (technology) have changed and moved on, should have taken Steve Jobs seriously 18-20 years ago.
Problem is, you can say that exact same sentence 5-10 years ago and it would be perfectly applicable.They can turn things around the problem is no one wants to hear it’s 5 to 10 years.
OK, so the chip uses less power, but what about the entire laptop/PC? I thought one of the major benefits of Lunar Lake was that the RAM was integrated and that brought significant power savings and longer battery life.And Intel says the chip consumes 10 percent less power than Lunar Lake
The package-mounted RAM on Lunar Lake didn't save any energy compared to other ways of mounting LPDDR5x. It saved power compared to SO-DIMM slots. If you look at any modern laptop implementation the DRAM ICs are millimeters from the SoC, hardly any different from mounting them on the package.OK, so the chip uses less power, but what about the entire laptop/PC? I thought one of the major benefits of Lunar Lake was that the RAM was integrated and that brought significant power savings and longer battery life.
we can expect up to a 10 percent improvement in single-core CPU performance compared to Lunar Lake
NPUs are useful. They’re just matrix math processor units. ML has been a mainstream feature for over a decade now, used anywhere pattern recognition or prediction is used today.I feel....nothing. I find myself these days completely unfazed by Intel's CPUs, I feel like the well has run drier than a bedouin's sandal.
...no, I do actually feel something; I am annoyed by the forced inclusion of NPUs. I mean, sure, everyone's including them now, but I, for one, wish they'd just leave those out entirely or used the space for something more useful.
You think it only took 5 months to get 18A back on track, to get designs taped out, and the fabs rolling, I have a bridge to sell you... This work was ALL started under Gelsinger, not Tan. Gelsinger was hardly perfect, but I think Tan is a crappy choice. He seems too eager to abandon fabs altogether, and too eager to fire more people.Actual lead time on manufacturing is ~2-3 months, and this is a product that won't be available in volume until 2026.
FWIW Tan started in March 2025. He would certainly have made the call for products in 2026.
Why would you abandon Windows over an NPU? You sound like the people complaining about Vista requiring a GPU, when today no one bats an eye over needing one to accelerate 3D because fundamentally 3D graphics is a superset of 2D graphics.Can that be considered even more reason to abandon Windows? Not that it matters - because of Windows, all of the laptop/desktop x86 chips (and the similarly targeted ARM chips) will have an NPU regardless.