Question Susquehanna's second chat with Charlie Demerjian

Page 5 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

lobz

Platinum Member
Feb 10, 2017
2,057
2,856
136
Intel's 10 nm were supposed to have been launched in either 2015 or 2016 IIRC, meaning they would be up against Zen 1 CPUs: VS THOSE, Intel's 10 nm is supposed to be more efficient.

Since then AMD has launched Zen 2 CPUs and those are more efficient than Intel's 10 nm.
That's why I specified Zen 2, I didn't get 100% how you meant it :)
 

lobz

Platinum Member
Feb 10, 2017
2,057
2,856
136
Yeah, not great at all actually.
9986060459e6261e0c07873dda1f2d41.jpg

This looks even more concerning when you think about that ICL-U doesn't perform better than Vega 11 in games at all.
 

jpiniero

Lifer
Oct 1, 2010
15,164
5,695
136
Why is it obvious? Not trying to be pedantic but want to see if you have anything beyond general assertions and the word of Charlie

They are doing better than I thought actually. I thought even Icelake client would be MIA. But it's obviously still terrible, see the 14 nm shortage. Even with the need to sell XCC dies due to Rome, Charlie's also right that the shortage goes away if they are able to supply enough Icelake.

The bad clock speeds (as seen on that chart) is also a sign. They are having to bin extremely loosely and jack up the voltage to ensure everything that comes out completely clean is sold as the i7 and all four cores are clean as an i5.

The shortage btw is only going to get worse once they start Cooper production, assuming OEMs can still sell it.
 

RetroZombie

Senior member
Nov 5, 2019
464
386
96
If company A does 7 dumb things, and 3 good ones, and company B does 8 good things, and 2 stupid ones, then I want someone that tells it honestly, without a vendetta or a case of the shills.
But do you have a link of one site that says the opposite of him, how wonderful the intel 10nm process is, how soo rosy is intel new designs, how amazing well have been their execution in late years, ... do you have some fairy tale website link?
 
  • Like
Reactions: DarthKyrie and lobz

moinmoin

Diamond Member
Jun 1, 2017
5,064
8,032
136
But do you have a link of one site that says the opposite of him, how wonderful the intel 10nm process is, how soo rosy is intel new designs, how amazing well have been their execution in late years, ... do you have some fairy tale website link?
Intel's stock price currently appears to be the highest since November 2000, so Wall St. may believe all of the above.
 

jpiniero

Lifer
Oct 1, 2010
15,164
5,695
136
Intel's stock price currently appears to be the highest since November 2000, so Wall St. may believe all of the above.

That's the magic of stock buybacks. It's just... when the magic runs out it won't be pretty.
 
  • Like
Reactions: DarthKyrie and lobz

lobz

Platinum Member
Feb 10, 2017
2,057
2,856
136
Intel's stock price currently appears to be the highest since November 2000, so Wall St. may believe all of the above.
I'd say that intel's current stock price is just as realistic as amd's $1.89 stock price was.
 

Gideon

Golden Member
Nov 27, 2007
1,769
4,126
136
There were some Icelake-SP entries on Sisoft 8 days ago running on a Supermicro Board, it tells us that Intel is indeed sampling Icelake-SP for testing end evaluation. They don't have to announce a sampling start to the public.
6-core samples - still goes to show that yields cannot be that terrific . You can't find rome samples anywhere near as low
 

Gideon

Golden Member
Nov 27, 2007
1,769
4,126
136
Rome is shipping. The 64 core variants can even be had on newegg for a decent (compared to MSRP) price.

Are you comparing to Rome when it was sampling 1-2 years ago ?
My point was that everything that AMD sampled (a long time ago) had way more cores. I don't know about milan core count, but i doubt it's any different
 

Arkaign

Lifer
Oct 27, 2006
20,736
1,377
126
But do you have a link of one site that says the opposite of him, how wonderful the intel 10nm process is, how soo rosy is intel new designs, how amazing well have been their execution in late years, ... do you have some fairy tale website link?

Huh? No, that's not what I was even talking about.

It's clear that Intel has had some major speedbumps for donkey's years now, and no clear answer in sight for most of their segments.

It's obviously deeper than even 10nm, there is a clear lack of planning in effect for what happened to their roadmaps given the realities of the current market. Pricing, model naming, announced products, it is all a bit confused at the moment.

All that aside, listening to CD for me is like listening to the ramblings of a madman. If you took the exact product stack of AMD and Intel, and swapped them right now in a hypothetical reality. Intel Ryzen / Threadripper etc, AMD 10980XEOMG 14+++++++ rehashes lol, and so on, Charlie would be saying how crap Intel and Ryzen were, and how AMD would soon be kicking ass with some magic new product, and how *ackshually* Ryzen sucks for everything.

There are simply enough good and great sources and journalists who treat these subjects with respect. Hell, they treat themselves and their readers with respect! Actual rational and honest people who call it as it is with no malice and no shilling. We don't NEED biased sources.
 
  • Like
Reactions: Zucker2k and OriAr

ksec

Senior member
Mar 5, 2010
420
117
116
It is the reason why Amazon and ARM accelerated their Server CPU development. The hardest part isn't designing a competitive CPU uArch anymore, we have done that with POWER and SPARC, it is the Fab.

Everyone knew what they were paying for buying Intel, part of that premium was Industry leading Fab. Since they constantly lied, Amazon is big enough to take matters into its own hand.

I doubt Intel GPU will save them either, their drivers wont be anywhere as good as AMD and Nvidia. Although an Intel engineers said this is complete bollox, we will see. I dont have much faith.
 

RetroZombie

Senior member
Nov 5, 2019
464
386
96
There are simply enough good and great sources and journalists who treat these subjects with respect. Hell, they treat themselves and their readers with respect! Actual rational and honest people who call it as it is with no malice and no shilling. We don't NEED biased sources.
That's the problem we have these days and not only in the tech press.
The press in my country was always very truthfully normally they only reported 'the news', today not really, now they have to give an opinion and if it's related to mr x lets attack him, if it's about mr y let's defend him.
And we surely don't need biased sources, and specially biased reviews, and there's a lot of those around.
 
  • Like
Reactions: Arkaign

Glo.

Diamond Member
Apr 25, 2015
5,802
4,776
136
There are simply enough good and great sources and journalists who treat these subjects with respect. Hell, they treat themselves and their readers with respect! Actual rational and honest people who call it as it is with no malice and no shilling. We don't NEED biased sources.
You do realize that behind the names are actual products, and not just names? Like actual, delivered products?

Also, when has judging anyones credibility became based on one's assumption over that person?

Charlie already has proven that when credit is due, he is way more than capable of giving it. Even to Intel.

There are reasons why he uses this language. Partially is to be sensationalistic for his paywall articles to sell, partially because when a donkey is a donkey you do not call it a horse.
 

lobz

Platinum Member
Feb 10, 2017
2,057
2,856
136
Huh? No, that's not what I was even talking about.

It's clear that Intel has had some major speedbumps for donkey's years now, and no clear answer in sight for most of their segments.

It's obviously deeper than even 10nm, there is a clear lack of planning in effect for what happened to their roadmaps given the realities of the current market. Pricing, model naming, announced products, it is all a bit confused at the moment.

All that aside, listening to CD for me is like listening to the ramblings of a madman. If you took the exact product stack of AMD and Intel, and swapped them right now in a hypothetical reality. Intel Ryzen / Threadripper etc, AMD 10980XEOMG 14+++++++ rehashes lol, and so on, Charlie would be saying how crap Intel and Ryzen were, and how AMD would soon be kicking ass with some magic new product, and how *ackshually* Ryzen sucks for everything.

There are simply enough good and great sources and journalists who treat these subjects with respect. Hell, they treat themselves and their readers with respect! Actual rational and honest people who call it as it is with no malice and no shilling. We don't NEED biased sources.
I'm sorry but no. Charlie was the only one who stuck to his story all through the 2 years of desperate denial. His information wad not biased, you still don't understand. Maybe you don't need information from sources you may not like but I think 90% of the tech enthusiast community looks at this from a more objective point of view.
If you get lost in the way something js presented, you might miss the whole point altogether.
 

lobz

Platinum Member
Feb 10, 2017
2,057
2,856
136
That's the problem we have these days and not only in the tech press.
The press in my country was always very truthfully normally they only reported 'the news', today not really, now they have to give an opinion and if it's related to mr x lets attack him, if it's about mr y let's defend him.
And we surely don't need biased sources, and specially biased reviews, and there's a lot of those around.
Please refresh my mind again: how is something a _biased source_ when what he says is true?
 
  • Like
Reactions: DarthKyrie

lobz

Platinum Member
Feb 10, 2017
2,057
2,856
136
You do realize that behind the names are actual products, and not just names? Like actual, delivered products?

Also, when has judging anyones credibility became based on one's assumption over that person?

Charlie already has proven that when credit is due, he is way more than capable of giving it. Even to Intel.

There are reasons why he uses this language. Partially is to be sensationalistic for his paywall articles to sell, partially because when a donkey is a donkey you do not call it a horse.
Finally a sober voice - plus you can iterate this much better than I can :)
 

Arkaign

Lifer
Oct 27, 2006
20,736
1,377
126
There was a lot more than Charlie to tell us 10nm was not going well, it was widely analyzed and reported.

Biased sources only tell you the truth when it aligns with their biases (and even then may be the part of the truth which best serves their agenda). It's like Rush Limbaugh or Rachel Maddow, the agenda just oozes like an irrepressible slime.

If I have a broken clock, once in a while it will look perfectly accurate. Still doesn't mean I should rely on it.

I won't and don't tell others what they can or can't listen to, but I choose to pay attention to people that seem most capable of calling the truth the truth no matter what. Anand Shimpi before moving on to other things was that guy. Now it's people like Stephen Burke.

YMMV. Best holiday wishes :)
 

Glo.

Diamond Member
Apr 25, 2015
5,802
4,776
136
I won't and don't tell others what they can or can't listen to, but I choose to pay attention to people that seem most capable of calling the truth the truth no matter what. Anand Shimpi before moving on to other things was that guy. Now it's people like Stephen Burke.
Im sorry to say this to you, but Stephen Burke has nowhere near the same type of sources, that Charlie has.

And its not only about Intel. Its about a lot of this tech industry.