AMD Thinks Intel's 10nm Stumble will Open Doors

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
55,601
There has been a lot of talk in the press by analysts and others in-the-know, that think Intel is positioned to lose a lot of business due to its bungled transition to 10nm lithography. This article is paywalled at TheInformation, but it will give you one peek for free with your email address, which it seems can be totally made up. Forrest Norrod, Senior Executive with AMD, was quite candid with his words in suggesting that Intel might have just got caught with its pants down. Of course, Intel's bungling in Taiwan has done nothing to shore up its position of continued dominance either.


Now, though, AMD may have a shot at coming out with a faster, more powerful chip than Intel for the first time. Intel in April said it was delaying the release of a more advanced chip manufacturing process until sometime in 2019. AMD has its own new, advanced chip, which it will now be able to release earlier than Intel, potentially giving it an edge in the market for high-performance chips for PCs and data center computers.

Aaron Tilley, the author of this article even exposes some of the ugliness going on behind the scenes, from I don't know when, but I did find it odd that AMD's Norrod had some jabs about Raja Koduri and him working out at Intel. And I thought that was our job. He also suggests that Intel will not be in on the GPU game for another 3 years, which is something we fully agree with.

Mr. Norrod played down the significance of the hiring. “Raja is a brilliant guy, a brilliant architect, a brilliant marketer, but he lives in the future,” he said. “He’s running a massive organization, and based on my perceptions of Raja and my experience working with him, that doesn’t play to his strengths. We’ll see what happens.”

He also claimed that AMD’s graphic chip business is doing a little better since Mr. Koduri’s departure. Under Mr. Koduri, AMD changed product plans for new graphic chips several times. “We had some wastage on the GPU roadmap where he changed his mind,” he said. “I think that since we’ve really locked that down, we’re making a little bit more rapid progress.”
 
Last edited:
"a faster, more powerful chip than Intel for the first time."
First time in a long time maybe, but not the first time. Athlon K7 was faster than the PIII when it was released and Athlon X64 smoked anything Intel had to offer until Core 2 came out.
 
"a faster, more powerful chip than Intel for the first time."
First time in a long time maybe, but not the first time. Athlon K7 was faster than the PIII when it was released and Athlon X64 smoked anything Intel had to offer until Core 2 came out.

Yeah wasn't the Athlon the first (mass market X86 cpu) to break the Ghz barrier?
 
Please let's insert arguing about stupid shit from the past into this conversation about the future. Oh, I see you already did that. You should get your achievement badge this week.

If you care to discuss history of processors, please go do that in one of the hundreds of other threads that discuss that or make your own.

On topic please.
 
Last edited:
Underestimating Intel is the worst that AMD did before. That did not end well. If Raja was the only problem at RTG then why was there several changes in personnel?

Lets say something silly in 3 years Intel comes out with a GPU product that matters then the hollow words from Norrod mean nothing. As much as he blames Raja from having his head in the clouds you could say that Norrod has his head up his ass, because when Intel has a product guess how many OEM will choose between AMD,Intel and Nvidia. If you find this pretty bold, check which OEM signed the GPP agreement ...

Short sighted or narrow minded AMD management need to make sure that if Intel is coming back they better still have a 3 year lead. If not then fighting Nvidia and Intel on 3 fronts is not going to end well.
 
I still don't think AMD will "unseat" Intel as king of the hill, and I'm a yuge investor in AMD (skin in the game baby). I do however see the massive potential for them to steal market share, and that's really all that matters. Hell, if Intel is unseated as King, even better, but I just want to see them deliver on their product line and nail their turn around. I'm a yuge fan of Dr. Su and everything she's done for the company.

I would stop worrying about who is better than who and get some equity in these bad boys (Nvidia and Micron as well). The competition is only going to benefit your bottom line, and semi-conductors and GPUs are going to power the next generation of cloud computing, video game streaming, machine learning, self-driving automobiles, and crypto-currency mining.

Get some skin in the game, people. Fanboyism isn't lucrative.

It's not for everyone, and it's been one hell of a ride (AMD stock has been manipulated so many times over the past two years) but those of us who have stayed long AMD (and long semi's & GPUs) are finally earning profits on their investments. Hell, if you've been long Nvidia since 2015, you probably don't need to work for a living any more.

It's a good day to be in the market.
 
Still AMD's fault we're in this situation, for bungling chip after chip after chip, letting Intel go nearly a decade of barely increasing clockspeed and IPC. Today's 4C/8T CPU is not significantly faster than the 4C/8T of Sandy Bridge. AMD is only now catching up. The only thing Intel bungled was misjudging that AMD would suddenly find competence.
 
Underestimating Intel is the worst that AMD did before. That did not end well. If Raja was the only problem at RTG then why was there several changes in personnel?

Lets say something silly in 3 years Intel comes out with a GPU product that matters then the hollow words from Norrod mean nothing. As much as he blames Raja from having his head in the clouds you could say that Norrod has his head up his ass, because when Intel has a product guess how many OEM will choose between AMD,Intel and Nvidia. If you find this pretty bold, check which OEM signed the GPP agreement ...

Short sighted or narrow minded AMD management need to make sure that if Intel is coming back they better still have a 3 year lead. If not then fighting Nvidia and Intel on 3 fronts is not going to end well.

I think the whole vega launch was handled so badly in all regards, maybe the engineering decisions in it were a bit before rajas time but the thing was his first big product outing, it went real bad and by all accounts he's a bit of a dick. That's enough to get you booted in plenty of places.
 
Still AMD's fault we're in this situation, for bungling chip after chip after chip, letting Intel go nearly a decade of barely increasing clockspeed and IPC. Today's 4C/8T CPU is not significantly faster than the 4C/8T of Sandy Bridge. AMD is only now catching up. The only thing Intel bungled was misjudging that AMD would suddenly find competence.

IF ONLY INTEL PLAYED FAIRLY IN THE k8 & P4 ERA... THE CRUX OF WHAT INTEL DID WAS DENY AMD ACCESS TO THE MARKETS THAT WOULD OF MADE IT THE MOST MONEY. NOT SAYING THAT WAS AMD'S ONLY ISSUE BUT IT SET THEM ON THE PATH.

INTEL DENYING OEM/ODM MARKET SHARE VIA KICKBACKS = LESS MARKET SHARE = LESS R&D SPENDING = DOING LESS WITH MORE AND THAT IS WHAT BIT AMD THE MOST ALONG WITH MANAGEMENT MISSTEPS TO COMPENSATE FOR THE SITUATION.
 
Still AMD's fault we're in this situation, for bungling chip after chip after chip, letting Intel go nearly a decade of barely increasing clockspeed and IPC. Today's 4C/8T CPU is not significantly faster than the 4C/8T of Sandy Bridge. AMD is only now catching up. The only thing Intel bungled was misjudging that AMD would suddenly find competence.

There was only one blunder chip really, then they were saddled with it in various forms for 5 years
 
Still AMD's fault we're in this situation, for bungling chip after chip after chip, letting Intel go nearly a decade of barely increasing clockspeed and IPC. Today's 4C/8T CPU is not significantly faster than the 4C/8T of Sandy Bridge. AMD is only now catching up. The only thing Intel bungled was misjudging that AMD would suddenly find competence.

I will never get this sentiment. You ignore the obvious and blame the victim of unfair competition, namely cash incentives to sell only intel. Yes, AMD fumbled but you act as if Intel had every right to do what it did, to not innovate on their own and offer their customers better value.
 
The fact that Intel is lagging behind now with competent solutions in both the HEDT and SERVER markets is almost unfathomable. I don't see them competing on a price/performance standpoint anytime in the next year. How many people are lining up to spend THREE THOUSAND dollars on a 28 core intel desktop chip? If AMD can deliver Threadripper refresh in the same monetary range as their initial offering, Intel is going to be in an extremely precarious situation.
 

As a business student I feel confident to say that there has never been and there won't be a free market ever.

Intel have power and if they say they won't let AMD take more than 15%, you can be sure they are willing to do some pretty shady stuff to hold them, the fines from the FTC and EU are jokes anyway.
 
Last edited:
i think the real question here is how many pluses can intel add to the end of 14nm++

The irony here, besides the ++ thing being an AMD ploy, is that intel has had 5ghz capable chips since sandy bridge. While they have "kept up" their newer stuff doesnt clock as well, so it all ends up a wash. I wonder what would happen if you put sandy bridge on a current DDR4 memory controller, and just let it rain 5GHZ+ (i just clicked on a little thing that popped up saying 5ghz+). The quest for clocks has historically failed (p4, BD), but i recall some pretty high clocked P4's that did well for themselves, heat and power aside, and cost, they were stupid expensive iirc.

Seriously this is AMD's change to un-bulldozer this shit. Realistically, i think AMD knows their 7nm is good shit, and its working, nothing can wreck this. So the next 16 months are AMD. Intel 10nm may realistically beat the living shit out of AMD 7nm, and then? Carbon nanotubes? Hardware as we knew it for the last 20 years is going to change soon.
 
Underestimating Intel is the worst that AMD did before. That did not end well. If Raja was the only problem at RTG then why was there several changes in personnel?

Lets say something silly in 3 years Intel comes out with a GPU product that matters then the hollow words from Norrod mean nothing. As much as he blames Raja from having his head in the clouds you could say that Norrod has his head up his ass, because when Intel has a product guess how many OEM will choose between AMD,Intel and Nvidia. If you find this pretty bold, check which OEM signed the GPP agreement ...

Short sighted or narrow minded AMD management need to make sure that if Intel is coming back they better still have a 3 year lead. If not then fighting Nvidia and Intel on 3 fronts is not going to end well.
AMD isn't really fighting. Just building chips. As long as sales outpace costs they aren't going anywhere.
 
Still AMD's fault we're in this situation, for bungling chip after chip after chip, letting Intel go nearly a decade of barely increasing clockspeed and IPC. Today's 4C/8T CPU is not significantly faster than the 4C/8T of Sandy Bridge. AMD is only now catching up. The only thing Intel bungled was misjudging that AMD would suddenly find competence.

Intel was not worried about AMD not because they had a crap product.

Was AMDs product up to par since the Athlon hay day... no. Still not the issue.

The folks running Intel believe they are in a position to slap down all comers forever.

Intels history has lead them to believe that.

Even before all the crap with AMD they tried to shut Cyrix down... failed. However after their engineers pulled those clean roomed Cyrix chips apart they realized they had fixed a bunch of Intel mistakes in regards to register naming, and vastly improved x86 power management. So they stole it for the Pentium Pro design... and got sued. Sure they had to give Cyrix an official licence to end that (and I guess that is how China is now making x86). Still the law suits and other back room crap kept a lot of OEMs away from Cryix. If they had gotten more sales during those Intel law suit years their later chips could easy have been on the market 2-3 years earlier and we would now have a 3 way race.

When AMD launched Athlon... Intel crushed them. It didn't matter that AMDs product was basically 2 generations better. Intel using illegal contracts and back room deals crushed them... the fines they got hit with well later where well worth paying and a complete joke. (and I would say now Intel would have zero worries doing the same shit again if need be.... no Judge is going to break up a company the US Gov views as a national security asset)

So why would Intel push performance, there is no need. They honestly believe even if AMD releases better product they can just out hustle them... skirting the law. Heck breaking it even if need be... the fines are no real deterrent.

As far as x86 goes... and as great as x86 Ryzen and Epyc are it makes no difference. Frankly AMD will never really push themselves into a potential 50/50 marketshare or anything even close to it situation. Intel will pull out all the tricks and are more then willing to copy steal cheat and break laws if need be.

Long term the only way AMD really puts the hurt on Intel... is if the ARM infrastructure continues to grow. Its no secret AMD has K12 (ARM Ryzen) designs sitting waiting to be fabbed at some point. Hopefully AMD continues updating those designs with their Ryzen core updates and refreshes. If the ARM server chip companies the Qcomms Samsungs and Apples push into the laptop market... its very possible that in a couple years AMD would be more then ready to drop Epyc ARM chips, I believe that is AMDs best road to beating Intel in the server market.
 
Last edited:
AMD Can't get cocky now. They need to pour a TON of resources into RnD and get their products to a level where they can't be ignored. They can't settle for being 2nd best at ANYTHING, because 2nd best in a two-horse race is also known as 'worst'.
 
Man intel CEO is straight up saying they expect to lose market share server side, 15-20% that will be a giant win for AMD if that happens. As long as AMD doesn't take it for granted they seem like they are on role right now. I think intel is back to drawing boards right now. I mean bulldozer was mess now AMD went with the same set up just with faster IPC and boom they can put together many more cores than intel. 7nm is only going to get better. I think the module design is finally paying off and all they needed was IPC and better interconnect. Amazing what you can do with tiny ass budget compared to intel if you set your mind to it and just execute. Lisa is gonna be paid well and I don't think AMD share holders are letting her go anywhere lol.
 
AMD Can't get cocky now. They need to pour a TON of resources into RnD and get their products to a level where they can't be ignored. They can't settle for being 2nd best at ANYTHING, because 2nd best in a two-horse race is also known as 'worst'.

in time. They are on right track and looks like they have the CEO in Lisa. Lady is an executioner! 7nm EPYC already in labs. She is for real!
 
This is the perfect combination of laziness and corporate greed.

They were way ahead for years and got complacent, allowing them to feel like they didn't need to dump tons of money into 10nm R&D thus making shareholders happy these past years.

Too bad they had a lack of vision and now Darth Shareholder, who is already upset about product delays, will be really pissed about losing sales / market share.
nmp-1-gif.gif


If they had invested more up front then 10nm would probably be fine now, but instead they're going to let AMD eat their lunch, again, and deservedly so.

I guess it's been too long since they've learned their last lesson.


Funny enough, in this analogy the overconfident Emperor would be Intel and Luke would be AMD, but the lack of vision worked too well here.
 
Now I'm under the impression that Raja is one of those fast talking bullshitters that combine just enough knowledge to confuse management for a while.
I've worked with those, it sucks.
If he is, look for Intel GPU sucking again and again.
 
This is the perfect combination of laziness and corporate greed.

They were way ahead for years and got complacent, allowing them to feel like they didn't need to dump tons of money into 10nm R&D thus making shareholders happy these past years.

Too bad they had a lack of vision and now Darth Shareholder, who is already upset about product delays, will be really pissed about losing sales / market share.
View attachment 80613

If they had invested more up front then 10nm would probably be fine now, but instead they're going to let AMD eat their lunch, again, and deservedly so.

I guess it's been too long since they've learned their last lesson.


Funny enough, in this analogy the overconfident Emperor would be Intel and Luke would be AMD, but the lack of vision worked too well here.
Intel falling behind in process...or even being so close to each other, is unacceptable and a freaking mayor market shift in my eyes... To me is a holy shit moment.. i know noise is being made about, but doesn't feel like nearly enough. As far as cpu design and all that, I think they will figure it out its a matter of how long, plus they can always copy a bit here and there probably (?) ... They are still in a position they should never have been.. but hey if they can sell 6000$ xeon based on 10y old tech, why not right? Right?..
 
Still AMD's fault we're in this situation, for bungling chip after chip after chip, letting Intel go nearly a decade of barely increasing clockspeed and IPC.
I don't think "bungling" is the correct word to use. In a situation where your main competitor financially dominates you to such an incredible degree. Not only in R&D, but in established business contracts and marketing.
I think you should give your head a shake and understand it is a near miracle that AMD is still in the game. Oh, and on top of this they are fighting a two front war. Yet they are still in both.
 
I don't think "bungling" is the correct word to use. In a situation where your main competitor financially dominates you to such an incredible degree. Not only in R&D, but in established business contracts and marketing.
I think you should give your head a shake and understand it is a near miracle that AMD is still in the game. Oh, and on top of this they are fighting a two front war. Yet they are still in both.

Two fronts where the competition are both willing to play dirty.
 
Should be interesting to see how it plays out. The stars may be aligning for AMD being that they're getting competitive right when Intel appears to be stumbling.

in time. They are on right track and looks like they have the CEO in Lisa. Lady is an executioner! 7nm EPYC already in labs. She is for real!
Yeah, AMD has been lagging behind the last decade and a half simply because the previous CEOs forgot to tell the engineers to whip up better chips. Thank god they passed the major hurdle of sending out an email to the engineers.
 
Still AMD's fault we're in this situation, for bungling chip after chip after chip, letting Intel go nearly a decade of barely increasing clockspeed and IPC. Today's 4C/8T CPU is not significantly faster than the 4C/8T of Sandy Bridge. AMD is only now catching up. The only thing Intel bungled was misjudging that AMD would suddenly find competence.

Intel's IPC gains came at the expense of breaking security, that's why we have all the Intel-specific Spectre/Meltdown BIOS updates with huge performance hits. With that in mind AMDs Bulldozer lineage was actually the superior set of chips, especially for anything heavily threaded.

AMD isn't perfect but you can't blame them for Intel's anti-competitive behavior.
 
Yeah, AMD has been lagging behind the last decade and a half simply because the previous CEOs forgot to tell the engineers to whip up better chips. Thank god they passed the major hurdle of sending out an email to the engineers.

The dude she replaced had a bachelor's degree in Information Systems... he did a fine job diversifying AMD, but he wouldn't know if the chip guys where doing anything groundbreaking or just blowing smoke and collecting a pay stub.

Lisa Su has a PhD from MIT in electrical engineering. There are very few people in the industry (at any company) on her level. She is a legit world class top of her field chip designer.

Her MIT Thesis from 1991.... if you understand even the first few pages of that you can see she was well ahead of her time in her thinking as a student. I bet she has forgotten more about chip design then anyone working at Intel these days... including the traitors. :)
https://dspace.mit.edu/handle/1721.1/11618

EDIT:
Some pre AMD Lisa Su;

In 1998 IBM moved to cooper interconnects.. and yes Lisa was the project manager that made it happen (all current chips are fabbed using Cooper pre Su it was Al... Intel is moving to cobalt I believe for their 10nm parts perhaps they should have hired Su lol);
https://www.electronicsweekly.com/n...nection-on-soi-semiconductor-process-1999-04/
"For the past couple of years our process technology has been very showy,” said Dr Lisa Su, project manager for CMOS Logic Technologies at IBM Microelectronics, pointing out that the 0.11?m features are drawn using a 248nm UV wavelength.

In 2001 IBM Toshiba and Sony partnered on a little chip called cell;
https://www.eetimes.com/document.asp?doc_id=1143130
Lisa Su, director of emerging products at IBM Microeletronics, said, "We started with a clean sheet of paper and sat down and tried to imagine what sort of processor we'd need five years from now."
 
Last edited:
Devil's advocate: Who's to say Intel intends to accomplish this by nefarious means? They may just mean that they'll step up marketing, promotion, vendor outreach etc. in order to reach this goal.

I agree though, it's seems difficult to hold your competitor to an exact percentage of a market where customers are totally free to choose which product they purchase.
 
I think it's debatable whether AMD has the better product right now. In the Athlon 64 days, AMD clearly had the better product. From what I see, a Ryzen 2700x is comparable to 8700k in most workloads and metrics. That's an amazing accomplishment. But the 8700k has been out longer and doesn't have hardware fixes for meltdown, etc. It still clocks faster and has less cores. I'd argue Intel may still have a slight tech/process lead. I suspect the Zen 2 will nearly even this out.

If AMD's next process node goes well and intel's doesn't, and Ryzen doesn't have any limitations on clock speed other than heat, and they can hold that lead for at least a year or two, then I suspect intel will lose up to 20% of it's market share. Liquid nitrogen overclocks that I've seen suggest intel's chips can clock much higher than Ryzen. Businesses will upgrade on their regular cadence. Home users will upgrade on their regular cadence. Intel will lose a much higher share of enthusiasts and much faster - but that's a small overall market share.
 
Back
Top