Wednesday, March 31st 2021
Intel Could Rename its Semiconductor Nodes to Catch Up with the Industry
In the past few years, Intel has struggled a lot with its semiconductor manufacturing. Starting from the 10 nm fiasco, the company delayed the new node for years and years, making it seem like it is never going to get delivered. The node was believed to be so advanced that it was unexpectedly hard to manufacture, giving the company more problems. Low yields have been present for a long time, and it is only recently that Intel has started shipping its 10 nm products. However, its competitor, TSMC, has been pumping out nodes at an amazing rate. At the time of writing, the Taiwanese giant is producing the 5 nm node, with a 4 nm node on the way.
So to remain competitive, Intel would need to apply a new tactic. The company has a 7 nm node in the works for 2023 when TSMC will switch to the 3 nm+ nodes. That represents a marketing problem, where the node naming convention is making Intel inferior to its competitors. To fix that, the company will likely start node renaming and give its nodes new names, that are corresponding to the industry naming conventions. We still have no information how will the new names look like, or if Intel will do it in the first place, so take this with a grain of salt.
Source:
Oregon Live
So to remain competitive, Intel would need to apply a new tactic. The company has a 7 nm node in the works for 2023 when TSMC will switch to the 3 nm+ nodes. That represents a marketing problem, where the node naming convention is making Intel inferior to its competitors. To fix that, the company will likely start node renaming and give its nodes new names, that are corresponding to the industry naming conventions. We still have no information how will the new names look like, or if Intel will do it in the first place, so take this with a grain of salt.
60 Comments on Intel Could Rename its Semiconductor Nodes to Catch Up with the Industry
It gives manufacturers some leeway on what to pick as their metric and not end up being called a liar on technicalities ))) In which universe? Even if that were somehow true, it's a dead-end platform, so as anything up to Kaby Lake. Not only do you lose modern features, there's also no prospect for upgrade. And I'm Lisa Su's nephew. No need to retract to laptops. Let's keep going with overclocked 2600K that beats Zen+ in 99.9% of games and applications.
Logic gate size (which is what the X nm denotes after all) matters very much, because that dictates power consumption and thus cooling requirements ...
Just to be clear, this is also for everyone who thinks that ICL is on the same 10nm as intended for CNL. It is not, ICL is the first 10nm+ product & probably wouldn't make it to the market if it was on the original 10nm node!
Compared to the Intel promo material for 10nm it's quite a bit less dense:
Hopefully Alderlake or nextlake shake's things up.
hexus.net/tech/news/cpu/119699-intel-10nm-density-27x-improved-14nm-node/
High Density variant is canonically used for mobile and other dies aiming at low power (and lower frequencies), High Performance variant are used - as the name says - high performance dies running at higher voltages, higher frequencies and higher power consumption.
Anandtech's 10nm Cannon Lake article laid out the Intel 10nm spec, scroll down to 'There's more to it' section - 100 MTr/mm² for High Density, 67 MTr/mm² for Ultra High Performance:
www.anandtech.com/show/13405/intel-10nm-cannon-lake-and-core-i3-8121u-deep-dive-review/3
This is not Intel-specific. The same applies to TSMC, with N7 official numbers at 96 MTr/mm² and high performance variant in the 65 MTr/mm² range. N7+/N7P/N6 should be 15-20% more dense.
All the official numbers are standard cells - or something more dense like SRAM if manufacturer wants to show even larger density numbers. The actual dies (like CPUs or GPUs) are usually less dense than that due to number of reasons. Logic circuitry does not lend well to being tightly packed together, there are areas like IO that do not scale down that well etc.
But it fits the trend of a loser: every gen you add a new nudge towards more misleading specs. Its what Intel has been doing since they abandoned their eternal quad core. Their introduction of a half dozen turbo states fits that category too. It was never about being helpful with more specs. Its about creating headroom by twisting reality. Meanwhile, it gets ever harder to keep an Intel CPU cool and thus performant.
If you're winning, your design and your product speaks for itself. Consider AMD's Zen. Right now, its never even about the node or whatever TDP It has and when it does, stuff either gets fixed or placed in perspective. Its just about the product: what do you get for your money and why is it better than most of everything else. The funny thing is, that discussion is also alive for Intel product... but no longer in their high end offerings, but the midrange and lower end. They're still competitive there - usually not even just 'situationally'. It reminds me of AMD's GPU stack not too long ago.
Basically more marketing is inversely proportional to a worse product. Its very refreshing to look at the market and its offers that way across the board, a pair of glasses I'm very fond of tbh, and it rarely proves to be wrong.
Side note: it'll be interesting what they'll do when they went small enough to run out of numbers, seeing as they're way ahead of reality in racing to the number 1 :)
Negative node sizes? -2nm? A rift in space and time? :ohwell: