Intel Tells Us Charlie has got it WRONG

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
51,074
File this under "Reports of my death have been extremely exaggerated." Intel made a direct comment in response to SemiAccurate's story yesterday telling us that 10nm is DOA for Intel. Not sure what to believe anymore.


Media reports published today that Intel is ending work on the 10nm process are untrue. We are making good progress on 10nm. Yields are improving consistent with the timeline we shared during our last earnings report.
 

The Mad Atheist

[H]ard|Gawd
Joined
Mar 9, 2018
Messages
1,187
Almost makes you wonder if Intel leaked false info to Charlie intentionally to discredit him after he embarrassed them.
"[Editor’s note: Charlie is currently off schedule traveling in Asia. Any requests related to this article will have a delay in response. Please contact the usual suspects for scheduling help. He will be back in the US timezone range next week.]."
 

trick_m0nkey

Moderator
Staff member
Joined
Oct 11, 2005
Messages
4,115
bagdad-bob.jpg
 

LostMF1

n00b
Joined
Sep 20, 2018
Messages
48
no conspiracy, just bad info all around. 10nm is definitely coming from Intel. Probably mid to late 2019. it seems misdirection supersede's information anymore. And many are confusing opinion with knowledge.
 

mdburkey

Limp Gawd
Joined
Jan 19, 2007
Messages
498
Intel only responded so quickly because they don't want their stock to dip

When they respond this quickly, the answer had to come from the marketing, management and PR departments -- NOT engineering or manufacturing.

Which begs the question, have they even talked to manufacturing before putting out the press release?

My personal guess would be that the rank and file down in manufacturing are saying "put a fork in it, it's done" but that hasn't filtered up yet -- or management is in the process of telling manufacturing to keep beating the dead horse until hell freezes over, regardless of whether or not it ever works right, just to save the stock price and keep up appearances while they work on a replacement process.

That Intel will eventually have a 10nm process is probably not even in question. How soon, and whether or not it is in any way based on the one they have been working on previously, is a completely different question however.
 

psyclist

Gawd
Joined
Jan 25, 2005
Messages
844
It will be called 10nm, but dropping a bunch of features originally intended for this node, will be interesting to see when they finally release what the performance and density will be like and comparison to the i3 8121u that is currently shipping
 

knowom

Limp Gawd
Joined
Aug 15, 2008
Messages
425
Sensationalism with sprinkles of truth most likely. Intel probably is making a bit of headway, but it's a matter of speed and efficiency at their 10nm progress that thus far has been plagued with setbacks.
 

DrBorg

Gawd
Joined
Jan 22, 2005
Messages
555
This is about right; deny while the guy is away for a week, just to muddy the waters while they release another Shitshow review for more processors that don't exist in quantity, and are having issues if you can get one.

They're clocked so close to the edge, I wonder about electromigration from any overclocking...

Kinda like Nvidia's new cards are having issues, so we'll see something similar by them this week as well.

Judging from the artifacts, NV is having soldering problems again, imho.

Lead free soldering works so well, lol.
 

Brackle

Old Timer
Joined
Jun 19, 2003
Messages
7,715
Well for what its worth his website is called semi-accurate laugh. I have found Charlie to be about 50% right.....

But I mean I would expect Intel to deny the claims, well because you know investors don't want to hear bad news! Not saying Intel isn't lying....anyway now im rambling.
 

dragonstongue

2[H]4U
Joined
Nov 18, 2008
Messages
3,162
This is about right; deny while the guy is away for a week, just to muddy the waters while they release another Shitshow review for more processors that don't exist in quantity, and are having issues if you can get one.

They're clocked so close to the edge, I wonder about electromigration from any overclocking...

Kinda like Nvidia's new cards are having issues, so we'll see something similar by them this week as well.

Judging from the artifacts, NV is having soldering problems again, imho.

Lead free soldering works so well, lol.

IMO the problem with Nv is they often do not go "all the way" to make best possible version, they find ways to "cheat" the system often enough to lower cost as much as possible and "hopefully" only lasts as long as warranty period.

granted what AIB do is not always in their control, but Nv is very "iron handed" about anything bearing their name, so the fault ultimately lies at their feet, bet you dollars to doughnuts them cheaping out on Vreg/caps for the lowest "reasonable" spec does not help anything either.

As for the Intel bit, they HAVE to say SOMETHING otherwise investor relations throw a shit fit, likey for the sheer amount of problems they have had up till now with their 10nm stuff it will be nothing close to what it should have been for anything but the lowest power/performing chips.

Heavy is the head that wears the crown. They were so focused on getting power consumption as low as possible (in most cases) little things were swept by the wayside, I have a funny feeling by them doing that it is finally coming to bite them in the ass real [H] ^.^
 

DrBorg

Gawd
Joined
Jan 22, 2005
Messages
555
10nm* is here now.

* +4nm

:)

Actually TSMC is doing very well with their 10nm finfet process, from what I hear.

Intel is having serious issues, I hear from the same people. :)

Almost 30% more leakage across the board; and this is some Very special silicon.
It's been processed such that there's a preferred orientation of current flow; it flows better in x dimension than Y dimension.
Chips have been made of 'Strained Silicon" for a while; this isn't new.

I have not heard that TSMC uses this type Si; that could be the difference.

The next big speed bump is going to be Silicon Carbide semis; they are making some chips, but mostly transistors for high power use now, due to difficulty and expense, but that will change.

Here's the early stuff; Yes, it's expensive.

No, that's Not 50 degrees, it's 500 degrees C.

Just replace the boiling water, and the chip won't go much past 100C, lol.

I have a 300A power transistor soldered to a copper can that is an infinite heat sink, as long as the Garden hose is on, lol.
Sometimes you need to power something up, just to see what's wrong with it; melting is for wimps. :)
 
Last edited:

Unoid

[H]ard|Gawd
Joined
Feb 4, 2003
Messages
1,049
What if the report is somewhat accurate that Intel 10nm from initial designs is a fail and they scrapped it for an altered 10nm, still labeled 10nm for legal reasons.

But loosening some of the pitch sizes and altered germanium usage in order to get successful dies out? Probably true seeing as how 10nm is 2 years behind schedule. That had to of transitioned 10nm into 10nmv2
 

Pieter3dnow

Supreme [H]ardness
Joined
Jul 29, 2009
Messages
6,785
The problems for Intel are yields on 10nm. The joke is somewhat on Intel, if they don't stop products as server cpu are not going to work ever on the current 10nm process unless it is a 2 core part.
The things they make on 10nm is not viable for large volumes and it is worse (power/performance) then their 14nm process

Even if you have the money for it the whole process has been worse then their 14nm track record. So what is it Intel wants a charade on 10nm while nothing of their high end cpu will ever work on current 10nm or just call it a day.

Their 14nm++ is not as bad as people think and will suffice.

Everyone can read the denial by Intel what everyone can not do is buy current Intel cpu on 10nm and that is a rather obvious tell.
 

Jim Kim

2[H]4U
Joined
May 24, 2012
Messages
3,760
"[Editor’s note: Charlie is currently off schedule traveling in Asia. Any requests related to this article will have a delay in response. Please contact the usual suspects for scheduling help. He will be back in the US timezone range next week.]."
Charlie was last seen going into a consulate seeking a wedding license.
 
  • Like
Reactions: Meeho
like this

N4CR

Supreme [H]ardness
Joined
Oct 17, 2011
Messages
4,687
:)

Actually TSMC is doing very well with their 10nm finfet process, from what I hear.

Intel is having serious issues, I hear from the same people. :)

Almost 30% more leakage across the board; and this is some Very special silicon.
It's been processed such that there's a preferred orientation of current flow; it flows better in x dimension than Y dimension.
Chips have been made of 'Strained Silicon" for a while; this isn't new.

I have not heard that TSMC uses this type Si; that could be the difference.

The next big speed bump is going to be Silicon Carbide semis; they are making some chips, but mostly transistors for high power use now, due to difficulty and expense, but that will change.

Here's the early stuff; Yes, it's expensive.

No, that's Not 50 degrees, it's 500 degrees C.

Just replace the boiling water, and the chip won't go much past 100C, lol.

I have a 300A power transistor soldered to a copper can that is an infinite heat sink, as long as the Garden hose is on, lol.
Sometimes you need to power something up, just to see what's wrong with it; melting is for wimps. :)

Saw those Cree SiCs few years back, they are phenomenal and incredibly efficient. The temp ratings blew my mind too - if I do a solar setup, I know what I will be using... let alone future cpus - the parameters of them will just make things so much less stressful for overclocking hard, as long as the materials can handle the thermal cycling and expansion rates etc. I will overkill heatsink the hell out of it to avoid any of that for my use be it solar or computers lol, even have a 30kg extruded bus bar alu heatsink in the garage somewhere for a mega LED or solar/etc project.

Wasn't aware strained silicon performed this way electronically, because it's same sort of factor for light and is used in various non-linear crystal manufacturing solutions.
Did you know they used to cool old stationary engines with a 40gal drum equivalent of water? The evaporation, surface area and natural circulation+convection alone will cool an engine making 12hp 24/7..

People keep forgetting Intel already rushed a 10nm laptop to market via Leonovo.
As you said, it is slower, hotter and less functional than 14nm equivalent before it. They didn't even get a 50% yield, they had to cut the GPU (about half the die) and 2 of 4 core modules!!!! That's potentially as bad as 20-30% yield, all while sucking profusely. An absolute shit-show.
Love your post.
 

Meeho

Supreme [H]ardness
Joined
Aug 16, 2010
Messages
5,196
no conspiracy, just bad info all around. 10nm is definitely coming from Intel. Probably mid to late 2019. it seems misdirection supersede's information anymore. And many are confusing opinion with knowledge.
Is it coming just like the 9900K came?
 

Pieter3dnow

Supreme [H]ardness
Joined
Jul 29, 2009
Messages
6,785
no conspiracy, just bad info all around. 10nm is definitely coming from Intel. Probably mid to late 2019. it seems misdirection supersede's information anymore. And many are confusing opinion with knowledge.

You are wrong :)

Lenovo Ideapad 330 is a Cannon Lake based 10nm CPU from Intel .

So can you explain it to whom here has a confusing opinion rather then knowledge ?
 

ChadD

Supreme [H]ardness
Joined
Feb 8, 2016
Messages
4,914
Throwing good money after bad... is not good business. Sometimes you gotta rip the bandaid off. Bottom line is Intel has revised their density targets a few times already... to a point where their "superior" to 7nm is no longer superior to nothing. (assuming they could ship functioning product) At their current target densities Intels 10nm will be inferior to TMSC and Samsungs 7nm... loosing the performance, yield and marketing war. I don't believe their is any scenario where spending billions more on 10mn is a sound decision.

Sure they pushed out the one 10nm sku... with the GPUs disabled, which is telling. Intel can claim its just that the GPUs aren't translating well to the 10nm process but there is nothing different in terms of a GPU or CPU gate. Chances are over 50% of their 10nm waffers are completely worthless with functioning GPUs and non functioning CPUs. I wouldn't be shocked if those crap low end Laptop chips they had to pay OEMS to take... had a waffer yield in the single digits. I can only imagine how many hundreds of dollars Intel lost on each one of those. But at least they where able to tell investors they where shipping an actual 10nm part.

Intels best play right now as they waited too long and likely threw multiple billions of good money after bad. Is do what they are doing... clock the crap out of what they got and hope AMDs 7nm parts underwhelm. While refocusing all their real R&D money on 7nm... with a REALISTIC density target this time, not 2.7x what they where already doing.
 
Top