Rumor: Intel Intentionally Holding Back 10nm

Subject: Processors | December 3, 2017 - 03:16 PM |
Tagged: Intel, Cannonlake, 10nm

According to Fudzilla’s unnamed, “well-placed” sources, Intel could have already launched a 10nm CPU, but they are waiting until yields get better. This comment can be parsed in multiple ways. If they mean that “yeah, we could have a 10nm part out, but not covering our entire product stack and our yields would be so bad that we’d have shortages for several months” then, well, yeah. That is a bit of a “duh” comment. Intel can technically make a 10nm product if you don’t care about yields, supply, and intended TDP.

View Full Size

If, however, the comment means something along the lines of “we currently have a worst-case yield of 85%, but we’re waiting until we cross 90%” then… I doubt it’s true (or, at least, it’s not a whole truth). Coffee Lake is technically (if you count Broadwell) their fourth named 14nm architecture. I would expect that Intel’s yields would need to be less-than-mediocre to delay 10nm for this long. Their reactions to AMD seems to be a knee-jerk “add cores” with a little “we’re still the best single-threaded tech” on the side. Also, they are looking like they have fallen behind the other fabs, which mostly ship 10nm in mobile.

I doubt Intel would let all that stigma propagate just to get a few extra percent yield at launch.

Of course, I could be wrong. It just seems like the “we’re waiting for better yields” argument is a little more severe than the post is letting on. They would have pushed out a product by now if it was viable-but-suboptimal, right? That would have been the lesser of two evils, right?

Source: Fudzilla

December 3, 2017 | 06:41 PM - Posted by hood6558 (not verified)

Anyone remember 2013? They had the same problems/delays with 14nm (Broadwell), and it was pushed back until 2014. As the process node gets smaller, low yields will be more of a problem. Intel can't afford to raise prices in this market, not with AMD dropping theirs, so good yields are needed.

December 4, 2017 | 01:41 AM - Posted by zme-ul

horrible title

December 4, 2017 | 01:41 AM - Posted by Vanessa (not verified)

Meh, tendentious post, PCPER.
They prob could do a 8nm chip.. don't mean that they should. AMD dropped the ball to stop competing with them before Ryzen (and even Ryzen aren't top of the top CPU yet). Why would they release a new nm architecture if they need to let the competition to catch up (i mean, or they would have monopoly issues)... also, the better the tech, the more expensive it will be for us to pay.. Intel 14nm CPUs are expensive (and high end) enough... before AMD come up with a i7 and i9 killer (both in performance and price).

December 4, 2017 | 01:55 AM - Posted by Nonnymous (not verified)

"They would have pushed out a product by now if it was viable-but-suboptimal, right? That would have been the lesser of two evils, right?"
Depends on how good the process is for clocks and power efficiency.

December 4, 2017 | 03:29 AM - Posted by John H (not verified)

Cannonlake will probably be cancelled due to yield issues, or being worse than 14++ perf wise. Long live icelake.

Also other mobile 10nm processes are on par with Intel 14nm.. not better.

December 4, 2017 | 07:09 AM - Posted by willmore

Yeah, that will be the solid way to know if 10nm really was that bad--if they've already come up with the improved architecture before the process is ready.

December 4, 2017 | 12:38 PM - Posted by Luthair

Isn't process node effectively a marketing term at this point and may not be measuring the same thing?

December 4, 2017 | 04:24 PM - Posted by willmore

Yes, that is correct. There have been quite a few articles--some based on actual data--about this topic.

The downside is that you have to understand the low level details of semiconductor manufacturing to make much sense of them.

December 4, 2017 | 04:58 PM - Posted by HoldCardsForTheBigChipInterestsAsUsual (not verified)

Intel's 10nm may not be that bad but it may be too costly currently to Intel's margins so Intel may just put it's 10nm off until Intel is forced by the competition to go with 10nm. Maybe Intel can sneak in a 14nm half node to 12 and still call it 14nm+++ but there is curently no market pressure on Intel to go smaller than 14nm with tweaks until either GF's 12nm process gets Zen+ up safely up into the 4.5GHz+ range or The GF 7nm rumors are actually true and GF is getting Great results with its 7nm testing yields and process performance.

It's Just like Nvidia's options with Nvidia's many base GPU die designs with their varying degrees of ROPs that can be used like the more ROP heavy GP102 base die design that tops out at 3840:240:96, shader cores:TMUs:ROPs, where Nvidia can get a Pascal base core design that has more ROP's than its GP104 base die design provides.

Nvidia has GP100, GP102, GP104, GP106, and GP108 base die designs all with different ratios of ROP to TMUs to Shader cores for Nvidia to drop some new gaming products on the mmarket to compete with only one AMD Vega 10 base die design for AMD's entire Vega micro-arch based discrete GPUs. And AMD's Vega Mobile integrated 10 nCUs, and Vega Mobile Integrated 8 nCUs designs currently for Raven Ridge APUs do not have any HBM2 currently so no HBCC/HBC in the laptop Vega offerings currently.

AMD's Vega 10 based die has a maximum of 64 ROPs available for gaming and way more shader cores for compute where Nvidia's many designs top out with quite a bit more ROPs with which Nvidia has the options to create SKUs like the GXT 1080Ti and retain it's Gaming FPS leadership.

AMD's Vega 10 is a compute first base die design and is very shader heavy even in its Vega 64 and 56 gaming derivatives. Vega 10 is also good with besting Nvidia's competition in raw texture fill rates but AMD is behind Nvidia in that one FPS metric that gamers pay the most attention to, and that is the one FPS metric that loves what larger numbers ROPs can throw out there.

So both Nvidia, Intel, and everybody have exhausted the low hanging froot of process node shrinks/scaling and Intel's 10nm may allow for more transistors per MM2. But that Intel 10nm process is currently rumored to be less efficient than Intel's current 14nm process node at getting higher clocks so maybe Intel's 10nm can come on mobile first where the clocks are lower and Intel will have to wait and see about 10nm for everything else unless Intel is forced to do so by the competition.

December 4, 2017 | 05:04 PM - Posted by HoldCardsForTheBigChipInterestsAsUsual (not verified)

Edit: froot
to: fruit

Edit: GXT 1080Ti
to: GTX 1080Ti

December 5, 2017 | 02:51 AM - Posted by Nikki (not verified)

Because if intel announce the 10 nm today, people would no longer buy the 14nm products on shelf today. People would just wait for 10 nm stocks to come in.

And if intel does not have enough 10 nm stock to sell, they would have not much revenue.

Post new comment

The content of this field is kept private and will not be shown publicly.
  • Lines and paragraphs break automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <code> <ul> <ol> <li> <dl> <dt> <dd> <blockquote><p><br>
  • Web page addresses and e-mail addresses turn into links automatically.

More information about formatting options

By submitting this form, you accept the Mollom privacy policy.