Intel has just lost their process advantage

[CataclysmZA]

Thought you ought to know.

https://www.intc.com/investor-relat...-Corporation-Earnings-Conference/default.aspx

News Summary:
• Record first-quarter revenue was $16.1 billion, up 13 percent year-over-year1 on strength of Intel's data-centric* businesses, which accounted for 49 percent of first-quarter revenue.

• Data-centric growth and operating margin leverage boosted earnings-per-share (EPS), which rose 53 percent year-over-year; non-GAAP EPS was up 32 percent year-over-year.

• Intel is raising its full-year revenue and earnings outlook based on this strong start; expecting 2018 revenue of $67.5 billion, up $2.5 billion from prior guidance.

Earnings Call:
10nm HVM in "2019," no time given. Yields not improving as fast as they expected.

Formally announced Whiskey Lake and Cascade Lake.

28, 20 and 14nm FPGAs demand up 40%

FPGA had more design wins this quarter than any other quarter before it

NOTES:
Intel is shipping low volume 10nm, HVM delayed till 2019

Source: https://www.reddit.com/r/hardware/comments/8f6cta/intel_q1_2018_financial_results_record_q1_results/

In addition to this, they're reshaping their Technology and Manufacturing Group (TMG) to fall under CISA, which is now under the command of Jim Keller, while the TMG head is now the chief of Technology, Systems Architecture & Client Group (TSCG) and oversees Keller.

 

[Bryn]

Eh, would have bought AMD anyway when I finally upgrade my system.

 

[HavocXphere]

Intel has just lost their process advantage

Doubtful. They're still miles ahead.

If they're running into issues I fully expect AMD to hit the same ones 2x as hard.

That said...I'm cheering for AMD...monopolies are good for nobody

 

[ponder]

Doubtful. They're still miles ahead.

If they're running into issues I fully expect AMD to hit the same ones 2x as hard.

That said...I'm cheering for AMD...monopolies are good for nobody

AMD are busy lab testing their 7nm vega gpus, with sampling to follow later this year. Says nothing about yields though.

 

[Polymathic]

With Windows 10 running on Qualcomm's chips this makes an Intel acquiring AMD merger a possibility

 

[Johnatan56]

With Windows 10 running on Qualcomm's chips this makes an Intel acquiring AMD merger a possibility

That's definitely not going to happen.

 

[ponder]

That's definitely not going to happen.

+1

They've got those annoying little anti-trust laws that might get in the way.

 

[DrJohnZoidberg]

Doubtful. They're still miles ahead.

If they're running into issues I fully expect AMD to hit the same ones 2x as hard.

That said...I'm cheering for AMD...monopolies are good for nobody

AMD have options though, kind of, as both TSMC and Global Foundries have working or near-working 7nm processes (they have a contract with GF that requires them to produce a certain amount with them which is why I say kind of). It would look really bad if Intel had to outsource their fabrication which is why it almost certainly won't happen.

There is some speculation though that GF is not quite ready yet as their Vega prototype is from TSMC.

 

[CataclysmZA]

AMD are busy lab testing their 7nm vega gpus, with sampling to follow later this year. Says nothing about yields though.

TSMC is fairly set for 7nm at this point, the next Snapdragon platform will be using it. I think they're already in ramp-up, and AMD taped up Vega 7nm a while ago.

 

[Polymathic]

That's definitely not going to happen.

Apple going to making their own desktop processors also, the market is rapidly changing

 

[HavocXphere]

AMD are busy lab testing their 7nm vega gpus, with sampling to follow later this year. Says nothing about yields though.

AMD is really rocking it lately.

And I gather they might beat Intel to the punch on Zen 2 on having undergone the painful process of setting up a new architecture.

...but overall Intel throws so much money at R&D that I don't see them easily losing out to AMD in a "oops we lost our process advantage" way. Don't recall the numbers but it's a hell of a lot more than AMD. That's why I'm saying...if Intel stumbles so will AMD.

AMD have options though, kind of, as both TSMC and Global Foundries have working or near-working 7nm processes

Fun fact - the 7nm means nothing. It's not a physical measure of anything in the chip (despite the nm name). Literally marketing. So one company's 7nm isn't necessarily on the same level as another 7nm. And from the stats I've seen it's all over the place in terms of thermals etc.

 

[Solarion]

Intel has just lost their process advantage

How many times this month?

AMD in the lead next week, Intel the week after and so forth.

 

[ponder]

Fun fact - the 7nm means nothing. It's not a physical measure of anything in the chip (despite the nm name). Literally marketing. So one company's 7nm isn't necessarily on the same level as another 7nm. And from the stats I've seen it's all over the place in terms of thermals etc.

Please elaborate. So '97 250nm is no different to '15 7nm...

 

[HavocXphere]

Please elaborate. So '97 250nm is no different to '15 7nm...

It is.

The 7nm no longer refers to seven nanometres in the physical sense. It refers to whatever the fk the marketing department wants. It has no basis in physical measurements of anything in the physical chips.

A bit like 3g 4g lte and all that sht.

From what I gather it seems to be based on a broad consensus in the industry on what level of advancement constitutes the next "node" in nanometer numbering. So it's sorta in sync by player but not really

 

[Lord Farquart]

How many times this month?

AMD in the lead next week, Intel the week after and so forth.

I had to scroll up to see if Staff Writer was not involved.

 

[access]

..yeah ok

 

[CataclysmZA]

How many times this month?

AMD in the lead next week, Intel the week after and so forth.

No, this is really it. 10nm was supposed to be done in late 2015, and then they tried to do it over focusing on EUV, and then they had to scale back to non-critical paths with EUV while attempting to hit some design milestones to give them a density edge. But now they're not sure when it'll be done. At the very earliest, they can start risk production in early 2019. 2018 is a no-go, and AMD is already sampling 7nm Zen 2 chips.

At the start of 2019, Intel will have a 10nm process that has barely any useful products on it, while TSMC will have been producing 7nm chips for months beforehand, and Qualcomm will be eating up their market share in low-end laptops.

Whatever lead Intel had is gone, because the rest of the year is additionally dedicated to launching a Coffee Lake refresh that is focused on fixing the branch prediction, among other things.

 

[ponder]

Intel also just hired Jim Keller as senior vice president. Every good CPU AMD ever made he was behind it.
Think Athlon K7, K8, x86-64, Ryzen etc.

So they got him and Raja Koduri from Radeon GPU fame...

 

[CataclysmZA]

Intel also just hired Jim Keller as senior vice president. Every good CPU AMD ever made he was behind it.
Think Athlon K7, K8, x86-64, Ryzen etc.

So they got him and Raja Koduri from Radeon GPU fame...

Keller's role within Intel is to fix their process and production issues as part of the TSCG, though. Nothing architectural this time around.