Discussion Intel current and future Lakes & Rapids thread

[coercitiv]

"coercion was so the twenties"

you called?!

 

[coercitiv]

Intel Reports Q2 2021 Earnings: Client Computing Leads the Way

For Q2, Intel’s Client Computing Group was once again the biggest winner; that division pulled in $10.1B in revenue, and is up 6% YoY. According to Intel, both laptop and desktop revenue is up, as Intel has surpassed shipping 50 million Tiger Lake processors. That said, these revenue gains are largely volume-driven; ASPs for both desktop and mobile are down, due in part to what Intel is noting to be increased sales of low core count processors.

Intel’s famed gross margin has also recovered on both a quarterly and yearly basis. At 57.1% it’s up almost 2 percentage points higher than Q1, and almost 4 percentage points higher than Q2’20. Intel’s gross margin has been subject to greater than usual fluctuations as of late – typically dropping whenever a major new product is ramping – but at least for Q2 it is on the rise as Intel enjoys a very profitable quarter.

Meanwhile Intel’s Data Center Group is really starting to feel the impact of the increasingly competitive server environment. While revenue was up 16% versus Intel’s tough Q1, it’s still down 6% on a yearly basis.

Overall, Q2’2021 marks a better quarter for Intel on both a quarterly basis and a yearly basis. Along with recovering from their tough Q1 three months ago, Intel this year isn’t facing another fab delay, as Intel’s big 7nm delay was first announced as part of their Q2’20 earnings. To be sure, the company is essentially flat on revenue and profit on a yearly basis – while its competition has generally grown – but as Intel gets further into CEO Pat Gelsinger’s tenure, there is increasing confidence that Intel will hit its goals (and a decreasing window to miss them).

 

[uzzi38]

Not that I am aware of. Source? In CB20, Rocket Lake landed exactly where one would expect it to. Rocket Lake’s performance was never the issue, power consumption is. Rocket lake and Zen 3 were pretty close when it comes to performance, tge issue is that Rocket Lake needed 2-3 times the power to get there thanks primarily to an old node.

People were expecting a straight 18% boost to gaming perf thanks to SNC IPC uplift, and were disappointed when it ended up on par with CML (some wins, some losses) and on average ever-so-slightly behind Zen 3. That's really it

Intel’s 10nm node is comparable to TSMC 7nm, and that is what a lot of random people here don’t get, AMD is not launching a 5nm chip until well into next year, over a year from now if some rumors are to be believed. Intel is launching their 7nm node (equivalent to TSMC3/4nm, ahead of 5nm unless Intel gimps it) next year, but, if current rumors and/or announcements are to be believed, they won’t launch a 7nm (TSMC 3/4nm product) x86 part until 2023 if announcements/rumors are accurate. Intel will likely leapfrog AMD every other year if this trend continues.

Intel's 7nm is not equivalent to N3. I'm genuinely not sure what gave you that idea?

Anyway, the way I see it both Intel and AMD will be leapfrogging each other, but we'll see.

Meanwhile, Golden Cove itself is rumored to offer a significant bump over Rocket Lake and Zen 3. Intel’s first gen desktop 10nm. part beats AMD’s equivalent last-gen part by several percentage points. Both Zen 2 and Zen 3 were on 7nm, that is 2 years of parts Intel is going to blow past.

Let's not pretend ICL-S and TGL-S didn't exist at one point. Neither would have faired perfectly well vs Zen 2 nor Zen 3, being up to 8c only. The whole "first 10nm product from Intel" rhetoric isn't a genuine one.

Ignoring the eye-cancer paragraph for the sake of my eyes.

Before you cheer on AMD, ask them where their 16 core laptop chips are. AMD is more than capable of producing such a chip. You should also look at my post history here and elsewhere if you think I am an Intel shill. I am not afraid to call out either company for the shenanigans they pull. I have purchased AMD/Cyrix when they were beating up Intel, and Intel when Intel nearly bankrupted AMD. I have owned 3DFX, AMD, and NVIDIA GPUs, among others. My only Agenda with tech is I love tech and love being able to do my stuff (development, video/graphics production and editing and gaming).

Capable of designing one? Yes. Capable of spitting out reasonable volumes for one? Ha ha.

Die size is a major concern here. No point designing a product you can't produce, much less one you can't sell at a reasonable price.

I have a decent track record for past predictions (people thought I was nuts about RDNA2…funny nobody talks about THAT now) and I am making another one: Intel is going to do just fine for Alder Lake. AMD will keep Intel from getting too far ahead thanks to 3D cache, but both Intel and AMD will be within 10% of each other for the foreseeable future, with one leapfrogging the other as a node advantage appears.

You made RDNA2 predictions?

Anyway though, that expectation is in line with mine as well. I'm also expecting both companies to leapfrog each other for the foreseeable future, but the one thing that needs to take place for that to really happen is AMD needs a heck of a lot more wafer capacity. Their biggest problem right now is not being able to supply both the high and low ends of the market.

 

[DrMrLordX]

JEFF K."S BRAND NEW HOEMPAGE !!! NO HAX0RING!!!

You were warned...

I wonder how old he would be now? Does he work at Intel?

That aside, it's amusing that people seem to think Intel will be competing against Vermeer with Alder Lake-S. It would require their competition to just sit on their hands and do nothing for, I don't know . . . two years?

 

[Asterox]

That should be Alder Lake-P Mobile.

Intel's 14-Core Alder Lake-P CPU Falls to 8-Core AMD Ryzen 7 5800H in Early Benchmarks

Where's the hybrid power?

www.tomshardware.com

 

[DrMrLordX]

Isn't Alder Lake-P coming out later than Alder Lake-S?

 

[jpiniero]

Isn't Alder Lake-P coming out later than Alder Lake-S?

Yeah.

 

[mikk]

Alder Lake-P should launch at CES early 2022.

 

[ondma]

That should be Alder Lake-P Mobile.

Intel's 14-Core Alder Lake-P CPU Falls to 8-Core AMD Ryzen 7 5800H in Early Benchmarks

Where's the hybrid power?

www.tomshardware.com

Those numbers were very disappointing. Seems like AL is really designed for mobile, and this is where it should shine. Six big plus 8 small should easily beat 8 Zen cores. Hell, if the IPC gains hold up, six AL big alone should be very close to 5800x. Maybe this AL is an early sample and the numbers will improve.

 

[mikk]

It's a stepping 0. All pre QS ADL benchmarks are meaningless because of bugs, non-final clock speeds and other things.

 

[dmens]

If the scores are real (which I believe) he turned out right. Not only it's the biggest architecture movement with two interesting new architectures, it's also the biggest performance leap since Conroe. The thing is he works for Intel now and that means for most people he must be automatically wrong.

It's a stepping 0. All pre QS ADL benchmarks are meaningless because of bugs, non-final clock speeds and other things.

Seems like your cautionary tale on early steppings is really just whatever fulfills your bias. Heh.

So how do you know that is a A0 part? I assume that is what you meant by "stepping 0", because there is no such thing. The parts are always identified by an alphabet for base layers tapeouts followed by a number for metal-only changes.

 

[IntelUser2000]

Those numbers were very disappointing. Seems like AL is really designed for mobile, and this is where it should shine. Six big plus 8 small should easily beat 8 Zen cores. Hell, if the IPC gains hold up, six AL big alone should be very close to 5800x. Maybe this AL is an early sample and the numbers will improve.

Well, according to the most optimistic projections about Alderlake, 1250 ST and 6800 MT would mean it's entirely running on Gracemont cores.

Tremont 10W: 725/~1950

Assuming Gracemont gets 1250.
-1250/725 x 2 x 1950 = 6724

Also, user-submitted benchmarks are unreliable. 5800H gets 1450 in ST and 8000 in MT for the top result, which is more representative of the system using the CPU, while TH is saying it's 1330 and 7000.

 

[yuri69]

How could they compare that six core ALD-P with 5800H? Relevant ALD-P SKUs span TDPs from 20 up to 45W.

 

[Exist50]

Well, according to the most optimistic projections about Alderlake, 1250 ST and 6800 MT would mean it's entirely running on Gracemont cores.

Tremont 10W: 725/~1950

Assuming Gracemont gets 1250.
-1250/725 x 2 x 1950 = 6724

Also, user-submitted benchmarks are unreliable. 5800H gets 1450 in ST and 8000 in MT for the top result, which is more representative of the system using the CPU, while TH is saying it's 1330 and 7000.

Regardless of what cores its using, the scores are lower than TGL H. Can very safely be ignored, at the very least as an iso-power comparison.

 

[mikk]

Seems like your cautionary tale on early steppings is really just whatever fulfills your bias. Heh.

So how do you know that is a A0 part? I assume that is what you meant by "stepping 0", because there is no such thing. The parts are always identified by an alphabet for base layers tapeouts followed by a number for metal-only changes.

Because it's a stepping 0, it's the first stepping. It translates into A0 which we know from github this is the first ADL stepping. I haven't searched for ADL-P but on ADL-S there are 5 steppings listed:

ADLS_REVID_A0 0x0
ADLS_REVID_A2 0x1
ADLS_REVID_B0 0x4
ADLS_REVID_G0 0x8
ADLS_REVID_C0 0xC

And in general every Intel ES with the label 0000 isn't final. You can see this also on the ST score itself, it's by far slower than TGL-U. Do you really think it reflects the final performance of ADL-P, really? If you believe in this it's just another proof that you have a huge bias against Intel, everyone knows it here. The first quote from me is related to ADL-S QS and not even ADL-P related, you should see a difference there. All the ADL-S stepping 0 entries were poor as well. You are really confused.

 

[dmens]

Because it's a stepping 0, it's the first stepping. It translates into A0 which we know from github this is the first ADL stepping. I haven't searched for ADL-P but on ADL-S there are 5 steppings listed:

ADLS_REVID_A0 0x0
ADLS_REVID_A2 0x1
ADLS_REVID_B0 0x4
ADLS_REVID_G0 0x8
ADLS_REVID_C0 0xC

And in general every Intel ES with the label 0000 isn't final. You can see this also on the ST score itself, it's by far slower than TGL-U. Do you really think it reflects the final performance of ADL-P, really? If you believe in this it's just another proof that you have a huge bias against Intel, everyone knows it here. The first quote from me is related to ADL-S QS and not even ADL-P related, you should see a difference there. All the ADL-S stepping 0 entries were poor as well. You are really confused.

0000 is just a part without the ID fused in. It has absolutely no relation to whatever you are referring to. And just FYI... it's the same core for mobile and desktop. LOL.

And sorry if my bias offends your bias.

 

[mikk]

0000 is just a part without the ID fused in. It has absolutely no relation to whatever you are referring to. And just FYI... it's the same core for mobile and desktop. LOL.

And sorry if my bias offends your bias.

It has of course, this is well known from all the Intel CPUs in the past. A newer stepping inreases the stepping number on Geekbench as well. Just one example: Tigerlake-U refresh is based on C0 instead of B0 and the number in Geekbench increased from 1 (B0) to 2. It was exactly the same for TGL-U.

A0= stepping 0 Geekbench
B0= stepping 1 Geekbench
C0= stepping 2 Gekbench

Claiming it has absolutely no relation is just wrong.

 

[dmens]

It has of course, this is well known from all the Intel CPUs in the past. A newer stepping inreases the stepping number on Geekbench as well. Just one example: Tigerlake-U refresh is based on C0 instead of B0 and the number in Geekbench increased from 1 (B0) to 2. It was exactly the same for TGL-U.

A0= stepping 0 Geekbench
B0= stepping 1 Geekbench
C0= stepping 2 Gekbench

Claiming it has absolutely no relation is just wrong.

Oh I see, you're going to use an arbitrary assignment from Geekbench instead of following actual industry standard stepping terminology, then insist there is correlation between different CPU families using said arbitrary assignment. Sounds totally legit.

You do realize the stepping is also potentially left unfused on those "0000" parts, right?

 

[mikk]

Oh I see, you're going to use an arbitrary assignment from Geekbench instead of following actual industry standard stepping terminology, then insist there is correlation between different CPU families using said arbitrary assignment. Sounds totally legit.

You do realize the stepping is also potentially left unfused on those "0000" parts, right?

The 0000 has nothing to do with the stepping, it's the CPUID which is generic for pre QS models. If they are just 0000 labeled it means they are ES1/ES2/ES3 based. The real CPU ID comes into play in QS stage. Regardless of the CPUID, early ES models go up with the stepping number scheme during development. ES can be A0, B0, C0 etc. Once again, this is well known.

 

[dmens]

The 0000 has nothing to do with the stepping, it's the CPUID which is generic for pre QS models. If they are just 0000 labeled it means they are ES1/ES2/ES3 based. The real CPU ID comes into play in QS stage. Regardless of the CPUID, early ES models go up with the stepping number scheme during development. ES can be A0, B0, C0 etc. Once again, this is well known.

So glad we have you to tell everyone well-known things like how engineers actually use a "stepping number scheme" from Geekbench during CPU silicon debug to identify parts, as opposed to industry standard stepping names. Thank you!

Can you tell me what a "real CPU ID" is? It is not just a string you can fuse in?

 

[mikk]

So glad we have you to tell everyone well-known things like how engineers actually use a "stepping number scheme" from Geekbench during CPU silicon debug to identify parts, as opposed to industry standard stepping names. Thank you!

Can you tell me what a "real CPU ID" is? It is not just a string you can fuse in?

This is how it works in Geekbench and it is accurate. A CPU ID which is non generic is a real CPU ID. I hope this helps.

 

[dmens]

This is how it works in Geekbench and it is accurate. A CPU ID which is non generic is a real CPU ID. I hope this helps.

Man, get your sarcasm meter checked.

 

[Curmudgeon666]

Recently, someone on reddit by the username of mooreslawisnotdead posted an Intel roadmap all the way through 2025. This roadmap contains some products/code names that haven't been mentioned here before. The roadmap could be completely genuine or a total work of fan fiction. I guess that's for you to decide. The user has since deleted his account, but not before I copied this information. What follows is the roadmap itself, along with commentary by the reddit user that posted it:

---------------------------------------------------------------------------------------------
Alder Lake (Golden Cove/Gracemont) Q4'21 / Q1'22 - predicted to be competitively weak vs AMD/Apple offerings that time.
Raptor Lake (Raptor Cove / Gracemont) Q3'22 / Q4'22 - 10% CPU perf boost and 8/16 configuration puts intel back on par but expect AMD/Apple to refresh their products as well.
Meteor Lake (Redwood Cove / Crestmont) Q2'23 - Intel's first true chiplet or tile based design. Different dies built on TSMC / Intel processes. More of a node shrink with single digit performance improvements. AMD will again extend lead with Zen 4+ / 5.
Arrow Lake (Lion Cove / Skymont) Q4'23 - Will feature an updated compute tile with 8/32 config for the high end enthusiast products. Might achieve parity with AMD offerings at the time but loses out to Apple in power efficiency.
Lunar Lake (Lion Cove / Skymont) Q4'24 - This is the product that will use TSMC 3nm as reported by Nikkei. Big performance jump expected and designed to achieve parity or beat AMD and Apple in both performance and power efficiency.
Nova Lake (Panther Cove [tentative]/ Darkmont) 2025 - This will mark the biggest architectural change in cpu architecture since the Core architecture is introduced in 2006. Intel is working to build an entirely new architecture from the ground up much like Ryzen with up to 50% cpu performance improvement from lunar lake. This is also the reason why Glenn Hinton returned.
---------------------------------------------------------------------------------------------

 

[Hulk]

Those numbers were very disappointing. Seems like AL is really designed for mobile, and this is where it should shine. Six big plus 8 small should easily beat 8 Zen cores. Hell, if the IPC gains hold up, six AL big alone should be very close to 5800x. Maybe this AL is an early sample and the numbers will improve.

I agree with these thoughts. First we see ADL 8+8 seemingly being faster than 5950X, then 6+8 slower than 5800? Doesn't make sense. But then again these early leaks really never do (make senss). It's like searching for lost treasure that never pans out. Just one more clue and I'll find the treasure...

 

[DrMrLordX]

I agree with these thoughts. First we see ADL 8+8 seemingly being faster than 5950X, then 6+8 slower than 5800? Doesn't make sense.

Power constraints, possibly. Alder Lake-S might have been unconstrained on power usage while the Alder Lake-P sample might have been stuck within a particular power envelope.