Discussion Intel current and future Lakes & Rapids thread

[lobz]

I was talking to a friend who left Intel 5 years ago. He just keeps pointing to the stock prices, and says he is happy. I told him about my EPYC processors, and he says "sounds good", but denies anything is wrong at Intel, because the stock price is good.

This friend of yours is either a happy stock holder who's planning on pulling out cash very soon, or someone who's in for a very dire surprise 1-2 years in from now.

 

[lobz]

Still needs a lot of work; not even half of GeForce GTX 1050 Ti (average score: 33000).

Yes, but still, this is intel integrated graphics wihtout eDRAM. I'm actually impressed that they pulled at least this off

 

[lobz]

Also, I just realized, that Intel's CEO has the perfect name to use as an abbreviation when reacting to anything he says publicly.


(Sorry, I never had a great sense of humor 🤣)

 

[swilli89]

I think it was in this thread about 2 months ago that someone said they've never been so sure about shorting a stock as INTC at that time. Well, the stock is up 19% since then

the entire market is so circus/bozo hyperinflated its impossible to actually valuate anything right now. AMD is currently at a LTM P:E of 268x. that's two-hundred and sixty eight trading price to earnings ratio.

AMD will probably meet or exceed on earnings and but will miss on revenue. expectations are too high and its already extremely overvalued for its growth prospects.

anyhow I digress..

 

[lobz]

the entire market is so circus/bozo hyperinflated its impossible to actually valuate anything right now. AMD is currently at a LTM P:E of 268x. that's two-hundred and sixty eight trading price to earnings ratio.

AMD will probably meet or exceed on earnings and but will miss on revenue. expectations are too high and its already extremely overvalued for its growth prospects.

anyhow I digress..

This.

AMD set their expectances not because they were pessimistic, but because they knew exactly, how many wafers they could reserve 6-9 months before they released the forecast for Q4 '19. I think they forecast in a way that calculates with every single piece of product they can produce as being sold, and give that number a - 0,5-1%, just to be sure.
That's also one of the reasons their stock is very high too - they are in a very good position regarding product strength, but still not setting delusional goals. I'd say AMD's stock price is more grounded and healthier than Intel's, despite the huge difference between their current and actual financial horsepower.

 

[Markfw]

This friend of yours is either a happy stock holder who's planning on pulling out cash very soon, or someone who's in for a very dire surprise 1-2 years in from now.

He is not technical, so I guess he is in for a surprise. I guess he is more of an aquaintance, we goes to the same gym as me for the last 10 years. We have never done anything outside of the gym, so....

 

[lobz]

We have never done anything outside of the gym, so....

Let me just say this: I would hope so

 

[krumme]

Very well put, and the followup post is also on point.

One thing to add though: while in the short term supply constrained equals higher revenue for Intel, it does not necessarily mean the same for OEMs. So while Intel would only feel the pain after another node flop, system integrators might already be under pressure to insure this type of situation doesn't happen again. Intel borderline lying about 10nm progress in the past won't help ease their fear either.

This may have long term consequences even if Intel turns the ship around with late 10nm / 7nm.

I am not so sure the long term effect is so huge. Ironically AMD helped their oem sales by covering the enthusiast/direct retail market and Intel certainly prioritized the all important oem that needs supply stability.

 

[jpiniero]

I am not so sure the long term effect is so huge. Ironically AMD helped their oem sales by covering the enthusiast/direct retail market and Intel certainly prioritized the all important oem that needs supply stability.

And yet their #1 customer felt compelled to say that the shortage was hurting their business to the point where they were going to push AMD more.

 

[Ajay]

This friend of yours is either a happy stock holder who's planning on pulling out cash very soon, or someone who's in for a very dire surprise 1-2 years in from now.

Eh, if it's working, use sell triggers. I used to just notch them up from time to time. Then went a stock took a dive, it automatically sold off.

 

[Ajay]

From what I understand after CML-S and RKL-S on LGA 1200, ADL-S follows on LGA 1700, and it's gonna be fabbed at 10+++nm.
Then he speculates the reason for the increase in the pin count is next gen I/O such as DDR5 and PCIE gen 5.

DDR5 is reasonable, PCIE gen 5 would be a very pleasant surprise though.

Thanks!

Hmm, ADL-S should be on 10++ (???). DDR5 does make real sense. So long as LGA-1700 works for 10nm and 7nm, could be a good socket.

 

[OriAr]

Thanks!

Hmm, ADL-S should be on 10++ (???). DDR5 does make real sense. So long as LGA-1700 works for 10nm and 7nm, could be a good socket.

The whole 10+ thing is confusing. The way I see it is:
10 - Cannon Lake, never made it to commercial product
10+ - Ice Lake, released only to mobile is somewhat low volume (At least relatively to Intel)
10++ - Tiger Lake, should be released in higher volume than ICL to mobile, small (But not inexistent) chance it makes it to desktop with low volume parts
10+++ - Alder Lake, should have a full product line up released to clients, without any volume limitations.

Now others see it as:
10 - ICL
10+ - TGL
10++ - ADL

So it gets pretty confusing, and it doesn't help Intel does both.

And I agree, LGA 1700 has the potential to be a great socket, historically Intel brought support to new generations of PCIE at roughly the same time for consumer and servers, and they don't waste time for new DDR standards either, we know that Sapphire Rapids will have support for both DDR5 and PCIE 5 at late 2021, which is when ADL-S is supposed apparently to be released. This certainly aligns to the possibility of having both PCIE 5 and DDR5 on ADL-S and LGA 1700, which would be very nice indeed.

Now, PCIE 5 is apparently VERY hard to implement, so it makes sense why Intel might want to wait with it a bit, but if Intel's trends are anything to go by, there should be PCIE 5 support on ADL-S.

About DDR5, again, Intel usually implement new DDR support as early as they can (They started to support DDR3 with Penryn back in 2007, DDR4 support started with Haswell E in 2014, both very shortly after the standard got finalized). So I will be very surprised if ADL-S doesn't support DDR5. (It might even support both DDR5 and DDR4 like Penryn supported both DDR3 and DDR2).

 

[mikk]

Marketing process should be 10nm++ for ADL and real process should be 10+++. The marketing process is bound to the generation and because Cannonlake has been ignored Icelake is their first 10nm generation, Tigerlake their second 10nm generation and Alder Lake probably their third generation on 10nm.

 

[DrMrLordX]

So not when Tigerlake ramps? Really interested to see what the real core count is for ICL SP.

Intel announced last year that Ice Lake-SP would reach up to 38c (after initially announcing 26c or 28c, I forget which). There have been some leaks of 6c and 14c Ice Lake-SP parts in the past (I think?). Sadly none of those products sound very interesting given their projected timeframe for release.

 

[beginner99]

he entire market is so circus/bozo hyperinflated its impossible to actually valuate anything right now.

Exactly. Tesla probably being the most extreme case. It's obvious their car business will fail once all the big goes go all-in. Been in a tesla model s and the quality is maybe that of a 30k car at most plus the lack of button controls.
There value is entirely in the battery tech and factories but even with that it's way overvalued. The stock market is full of emotional investment, anyone remember theranos?

 

[Intel Inside]

The whole 10+ thing is confusing. The way I see it is:
10 - Cannon Lake, never made it to commercial product
10+ - Ice Lake, released only to mobile is somewhat low volume (At least relatively to Intel)
10++ - Tiger Lake, should be released in higher volume than ICL to mobile, small (But not inexistent) chance it makes it to desktop with low volume parts
10+++ - Alder Lake, should have a full product line up released to clients, without any volume limitations.

Now others see it as:
10 - ICL
10+ - TGL
10++ - ADL

So it gets pretty confusing, and it doesn't help Intel does both.

And I agree, LGA 1700 has the potential to be a great socket, historically Intel brought support to new generations of PCIE at roughly the same time for consumer and servers, and they don't waste time for new DDR standards either, we know that Sapphire Rapids will have support for both DDR5 and PCIE 5 at late 2021, which is when ADL-S is supposed apparently to be released. This certainly aligns to the possibility of having both PCIE 5 and DDR5 on ADL-S and LGA 1700, which would be very nice indeed.

Now, PCIE 5 is apparently VERY hard to implement, so it makes sense why Intel might want to wait with it a bit, but if Intel's trends are anything to go by, there should be PCIE 5 support on ADL-S.

About DDR5, again, Intel usually implement new DDR support as early as they can (They started to support DDR3 with Penryn back in 2007, DDR4 support started with Haswell E in 2014, both very shortly after the standard got finalized). So I will be very surprised if ADL-S doesn't support DDR5. (It might even support both DDR5 and DDR4 like Penryn supported both DDR3 and DDR2).

Can this be a 11900k i9 ?

https://twitter.com/x/status/1200240746239942658

 

[Intel Inside]

Can this be a 11900k i9 ?

https://twitter.com/x/status/1200240746239942658

Damn 180 second bug.

 

[RetroZombie]

And I agree, LGA 1700 has the potential to be a great socket

Interesting what use for so many extra 540 pins?

- Simultaneous DDR4/DDR5 support​

- Cpu mcm (like socket 478 to 775)​

- More memory channels​

- More PCIe lanes​

- External co processor connector for on-package/off-package​

- Dedicated bus for on-package/off-package gpu​

- Other?​

 

[Dayman1225]

Intel announced last year that Ice Lake-SP would reach up to 38c (after initially announcing 26c or 28c, I forget which). There have been some leaks of 6c and 14c Ice Lake-SP parts in the past (I think?). Sadly none of those products sound very interesting given their projected timeframe for release.

I’m pretty sure intel never officially announced any specs for Icelake Sp

 

[DrMrLordX]

I’m pretty sure intel never officially announced any specs for Icelake Sp

No clockspeeds, but:

Intel 10nm Ice Lake-SP Xeons Arriving Q3 2020 With Up To 38 Cores And 76 Threads

Intel is planning to make Ice Lake-SP available in up to 38-core/76-thread configurations (per socket) with a TDP (Thermal Design Power) of 270 watts.

hothardware.com

38c, 270W. Almost guaranteed to be Sunny Cove. Not looking very good.

 

[Markfw]

No clockspeeds, but:

Intel 10nm Ice Lake-SP Xeons Arriving Q3 2020 With Up To 38 Cores And 76 Threads

Intel is planning to make Ice Lake-SP available in up to 38-core/76-thread configurations (per socket) with a TDP (Thermal Design Power) of 270 watts.

hothardware.com

38c, 270W. Almost guaranteed to be Sunny Cove. Not looking very good.

My 64 core EPYC only uses 200 watt, and when I get it fired up, I bet it blows the doors off these 38 core chips. And thats last years news...

I still don't see how Intel will beat the Rome chips any time soon. Anybody with a brain would never buy any of these Intel server chips.

 

[DrMrLordX]

My 64 core EPYC only uses 200 watt, and when I get it fired up, I bet it blows the doors off these 38 core chips. And thats last years news...

I still don't see how Intel will beat the Rome chips any time soon. Anybody with a brain would never buy any of these Intel server chips.

38c Ice Lake-SP isn't competing with Rome due to delays. It's competing with Milan.

 

[Intel Inside]

No clockspeeds, but:

Intel 10nm Ice Lake-SP Xeons Arriving Q3 2020 With Up To 38 Cores And 76 Threads

Intel is planning to make Ice Lake-SP available in up to 38-core/76-thread configurations (per socket) with a TDP (Thermal Design Power) of 270 watts.

hothardware.com

38c, 270W. Almost guaranteed to be Sunny Cove. Not looking very good.

38c, 270w. 10nm + meh why didn't they manufacture in 10nm ++ ?

 

[Dayman1225]

No clockspeeds, but:

Intel 10nm Ice Lake-SP Xeons Arriving Q3 2020 With Up To 38 Cores And 76 Threads

Intel is planning to make Ice Lake-SP available in up to 38-core/76-thread configurations (per socket) with a TDP (Thermal Design Power) of 270 watts.

hothardware.com

38c, 270W. Almost guaranteed to be Sunny Cove. Not looking very good.

Yeah that slide wasn’t meant for the public, nor was it directly from Intel but ASUS . I don’t doubt that it’s upto 38c but intel still hasn’t actually announced those specs

 

[Markfw]

38c Ice Lake-SP isn't competing with Rome due to delays. It's competing with Milan.

I guess I didn;t state my point well enough. If a current 64 core Rome will easily beat their 38 core, then Milan might even beat it with only 32 cores. And the power usage difference will be incredible.

So how long can Intel still brainwash IT managers into buying Xeons ?????? Over Rome/Milan