Discussion Intel current and future Lakes & Rapids thread

[TheGiant]

LGA1159? Thanks no Intel, at least 8C CFL-Refresh was Z370 compatible.

the MB makers must be really sick of all of those sockets and power specs/chipset and BIOS updates over last 5 years....

 

[JoeRambo]

the MB makers must be really sick of all of those sockets and power specs/chipset and BIOS updates over last 5 years....

Isn't that's how they make money out of customers? ( or the proper English word here is "milk" ).

 

[DrMrLordX]

Isn't that's how they make money out of customers? ( or the proper English word here is "milk" ).

They do until they have to support old products that are otherwise EoL.

 

[birdie]

@ApTeM Where did you find that?

Some obscure non-English 4chan. Looks quite real I have to admit that's why I posted it. And I have to agree that both Intel and AMD has decided that TDP means pretty much nothing nowadays which is quite unnerving. If I'm not mistaken the 8086K was the first CPU to break all limits.

Intel has had several years to retool Ice Lake to release it on their refined 14nm node. Why haven't they done that yet and instead they are going to release an Nth iteration of Sky Lake? It's not even funny Intel. Get your act together. Scrap 10nm if it doesn't work. Go to EUV 7nm for God's sake.

 

[IntelUser2000]

Intel has had several years to retool Ice Lake to release it on their refined 14nm node. Why haven't they done that yet and instead they are going to release an Nth iteration of Sky Lake?

It takes 4-5 years to go from inception to production. They simply didn't plan for it. Weren't you there when all the disasters were happening with the previous CEO of Intel, Brian Kraznich?

Scrap 10nm if it doesn't work. Go to EUV 7nm for God's sake.

Skipping a process isn't like jumping over some rock when trying to cross a river.

Technology advances are progressive, not a quantum leap that comes out of nowhere. The newer ones build upon the foundations of the older one. It's especially true in bleeding edge technology where advances are something not previously done before.

 

[birdie]

Intel started developing the 7nm process at the very least three years ago as all wafer producers always codevelop two nodes simultaneously. And their 10nm node hasn't yielded any decent results after five years of refinements. Anyways, they must know better and my speculations are worthless. It just sucks they we've been stuck with the Sky Lake uArch for four years already. I don't remember Intel dragging any of their previous uArchs for so long.

 

[IntelUser2000]

Intel started developing the 7nm process at the very least three years ago as all wafer producers always codevelop two nodes simultaneously.

It doesn't matter. Do you really think they develop from zero every time they make a new uarch or a new process?

They overlap development to reduce time to market, but newer processes very much build on older ones.

I don't remember Intel dragging any of their previous uArchs for so long.

Because of bad management.

 

[birdie]

Because of bad management.

Brian Krzanich couldn't be single-handedly responsible for this fiasco. And again they have all the money in the world to be on the edge. Why have they failed so hard?

 

[IntelUser2000]

Brian Krzanich couldn't be single-handedly responsible for this fiasco. And again they have all the money in the world to be on the edge. Why have they failed so hard?

Yes he is. Because when a new CEO comes in, he also brings his own team of advisors. He fired a bunch of people when he became a CEO. There are countless articles about this.

CanardPC wrote a great piece about what was happening inside Intel when he was in charge. It was chaos to put it simply. It said he was cancelling projects left and right, firing those that disagreed with him and had no patience. He took over divisions by himself. When he became CEO, allegedly him and Renee James had some "secret" ideas to penetrate the mobile market and was a reason why he got picked.

That secret idea became famously known as the Contra Revenue program. What did he do after losing billions of dollars for 2 years? He threw it away in 2015 when Broxton was canned.

For example, Intel Labs didn't have a head for years because it was BK that took over. Justin Rattner was head of Intel Labs and he was pretty great. Intel Labs are responsible for innovative ideas that might come in the future, so it was a big loss.

 

[birdie]

Yes he is. Because when a new CEO comes in, he also brings his own team of advisors. He fired a bunch of people when he became a CEO. There are countless articles about this.

CanardPC wrote a great piece about what was happening inside Intel when he was in charge. It was chaos to put it simply. It said he was cancelling projects left and right, firing those that disagreed with him and had no patience.

I've seen it and read it but these were speculations to put in mildly. Never seen anything coming directly from the horse's mouth. Not a single post on medium.com from a current/past Intel employee.

 

[IntelUser2000]

I've seen it and read it but these were speculations to put in mildly. Never seen anything coming directly from the horse's mouth. Not a single post on medium.com from a current/past Intel employee.

So do you believe he got fired because he was having sex with an employee?

Then you are being naive! They do far worse things in the corporate world. Broadcom CEO had lavish, private parties for it. It got revealed to the public when the wife found his secret dungeon. S|A even talked about how Intel threw parties for post-IDF celebrations.

Board of Directors ignore it likely because they participate it in themselves, or its so common its not shocking anymore. But when you screw up your job(like failing to deliver on products), then they'd use the knowledge they have to fire you.

It also serves the purpose of saving face.

 

[birdie]

So do you believe he got fired because he was having sex with an employee?

Absolutely not. I just cannot wrap my head around the fact that no one among Intel management had the guts to stand up against the company being actively destroyed/mismanaged for so long. Intel perfectly knew about Zen/Ryzen. Intel knew about TSMC/Samsung 7nm nodes. They still kept dragging their 14nm node when it was obvious they didn't even had the capacity to mass-produce many-cores CPUs not to mention that 14nm doesn't quite work for such a scenario 'cause you end up with CPUs which have an insane TDP.

 

[IntelUser2000]

Absolutely not. I just cannot wrap my head around the fact that no one among Intel management had the guts to stand up against the company being actively destroyed/mismanaged for so long..

I don't see how that's something hard to believe.

When Leo Apotheker was in charge of HP, he announced that they were going to abandon selling PCs.

Do you know how crazy you have to be as a CEO of a #1 manufacturer of computers to announce you'd flat out abandon it?

 

[birdie]

I don't see how that's something hard to believe.

When Leo Apotheker was in charge of HP, he announced that they were going to abandon selling PCs.

Do you know how crazy you have to be as a CEO of a #1 manufacturer of computers to announce you'd flat out abandon it?

What's the board of directors for then? Or CEO in Western countries is free to do whatever he pleases?

 

[IntelUser2000]

(I'm going to take this private because its off topic. Whether you want to continue is up to you)

I'm curious if Comet Lake core or graphics has any changes. Last time they changed the iGPU name was with Kabylake with the new media. It went from HD 530 to HD 630.

 

[Bouowmx]

Anon-chan board as source gg

Typography is pretty good though

 

[IntelUser2000]

@ApTeM I'm starting to have serious doubts on whether it's real.

Compare it to this: https://www.techpowerup.com/254838/intel-expands-9th-gen-core-desktop-and-mobile-processor-families

-7MB and 9MB L3 cache for a 4 core chip?
-14+++, when Intel refers to everything as just 14 now.
-No superscripts
-Processor Graphics goes from a checkmark to a name
-Most striking, is how RCP is shown as X$, when Intel presentations show it as $X. The latter is the accepted convention when denoting dollar amounts.

 

[mikk]

There are some conflicts in this slide. The slide says LGA1159 whereas a known source said LGA1200. Also the TDP differs, 105W to 125W TDP. Also it's strange for me having a price list for CPUs in 6-12 months.

 

[birdie]

@ApTeM I'm starting to have serious doubts on whether it's real.

Compare it to this: https://www.techpowerup.com/254838/intel-expands-9th-gen-core-desktop-and-mobile-processor-families

-i3-10300 is at 62W TDP. Why?
-7MB and 9MB L3 cache for a 4 core chip?
-14+++, when Intel refers to everything as just 14 now.
-No superscripts
-Processor Graphics goes from a checkmark to a name
-Most striking, is how RCP is shown as X$, when Intel presentations show it as $X. The latter is the accepted convention when denoting dollar amounts.

Might be a fake for all I know. Still it looks very convincing ;-)

Anon-chan board as source gg

Typography is pretty good though

Could you post something a little bit less cryptic? Google doesn't find anything at source.gg

 

[IntelUser2000]

The dollar sign is the biggest giveaway for me.

Must not live in US/Canada or something. We say "XX dollars" but write $X.

Update: If its fake, then its very convincing and done by someone knowledgeable about the field. So a 62W part does exist.

 

[JoeRambo]

And 7MB of L3 make very little sense as well. For example 6C dies have 6 partitions of 2MB of L3 cache. Can't really reduce that to 7MB?

 

[mikk]

I believe it's a fake, in this case it's not a bad fake, there is good knowledge required, even though the guy who did this made some errors.

 

[IntelUser2000]

Because of the much faster LPDDR4, the subscores are better for Comet Lake-U.

LPDDR4 isn't supposed to be in the initial CML-U chips.

And 7MB of L3 make very little sense as well. For example 6C dies have 6 partitions of 2MB of L3 cache. Can't really reduce that to 7MB?

If they really wanted to sure, there's no technical limitation. But what, they want to use 6 core dies which are larger, but disable some cores and caches a little bit more rather than using 4 core dies? It just seems strange.

 

[ikjadoon]

The score is bad anyways. The Icelake version running at 3.5GHz performs identically.

That's the same thought I've had. But, at the same time, I have doubts 10nm Ice Lake laptops will be priced reasonably in 2019, so hopefully price pressure brings down 14nm ++(+) pricing.

Project Athena can't come a moment sooner.

 

[Bouowmx]

Could you post something a little bit less cryptic? Google doesn't find anything at source.gg

Sorry. I opened the full-size version of your attached image, right-click, and Search on Google. The primary result is: https://2ch.hk/hw/res/2810449.html

Being that the source is anonymous 4chan-style board, I dismiss the info as fake (although the ideas within may be true)