• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Discussion Intel current and future Lakes & Rapids thread

Page 273 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
IntelUser2000 said:
If 2+8+2 is really the successor to Tigerlake, and considering it needed 6 cores, 2+8+2 seems like a serious downgrade.

Also unlike Lakefield, at the minimum the big cores have to work simultaneously with the small cores. This seems to be the key, and a big change for Lakefield, both in performance and in the technical front.

2+8+2 configuration with Golden Cove being 20% faster per clock and Gracemont being an additional 30% faster per clock over Tremont makes it possible in theory for 2+8+2 to beat 4+2 of Tigerlake, and by 20% or so.

We need more than that though. The same is true in desktops for Alderlake.

Is Intel going to permanently cede multi-threading performance leadership to AMD, or does it have a secret sauce?
Click to expand...

M almost has to be the Lakefield replacement, maybe replace Y as well. P will presumably be what the Tiger Lake replacement is, except the base TDP range would be like 12-45.
 
mikk said:
EXCLUSIVE: Jasper Lake lineup: https://www.fanlesstech.com/2020/09/exclusive-jasper-lake-lineup.html

The fastest Pentium Silver N6005 can go boost up to 3.3 Ghz with a 10W TDP. There is no SuperFin for Tremont und TDP is low, I think they can go over 4 Ghz with Gracemont on ADL-S based on enhanced SuperFin.
Click to expand...

Hmm. Why only 4c? Also we're less than a year out from Gracemont being available (presumably), so I have to wonder why they're bothering with Tremont now . . .
 
4C only because it's a low end and low power product line and the Jasper/Elkhart Lake successor won't be available next year, ADL-P is not a successor of this.
 
mikk said:
ADL-P is not a successor of this.
Click to expand...

That's not what I meant. The release window is so late that they've got Gracemont ready or nearly ready-to-go on the same process. They could have skipped over Jasper Lake and gone to a Gracemont-based successor (NOT Alder Lake).

IntelUser2000 said:
You don't need more than 4 cores, not for this lineup.
Click to expand...

I guess? The cores are tiny, yields should be (relatively) good, and they could tamp down TDP by dropping boost clocks a little. Anything they did with 4c on 14nm should be doable with 8c on 10nm.
 
DrMrLordX said:
That's not what I meant. The release window is so late that they've got Gracemont ready or nearly ready-to-go on the same process. They could have skipped over Jasper Lake and gone to a Gracemont-based successor (NOT Alder Lake).
Click to expand...

Really not being nitpicky, because 6-8 months difference is a lot. And low end parts are never released first. The same could be said for Celeron/Pentium Tigerlake. Why not wait for Alderlake?

As for core counts, you get something better on Jasperlake. Such as the enhanced GPU and Media with Gen 11.

The clock seems a decent increase from Goldmont Plus. Icelake is only 18% higher. Now Sunny Cove being 4-5x large for core size seems even more questionable.
 
Last edited:
IntelUser2000 said:
Why not wait for Alderlake?
Click to expand...

. . . I guess? Different market segments, and the consumer Atom devices have been long-neglected, while the mobile devices served by Alder Lake-P next year get TigerLake this year.

As for core counts, you get something better on Jasperlake. Such as the enhanced GPU and Media with Gen 11.
Click to expand...

I can see that being of benefit. It may also be that Intel will eventually can Atom-only consumer devices in favor of Lakefield-R/Alder Lake-M.
 
DrMrLordX said:
I can see that being of benefit. It may also be that Intel will eventually can Atom-only consumer devices in favor of Lakefield-R/Alder Lake-M.
Click to expand...

I'm not sure if Jasper and Elkhart are actually physically different, but I think industrial IoT is the main focus anyway. Consumer Atom can just be where they send the duds, or if IoT demand falls (like it did last quarter, which they blame the virus). That it has a full 32 EU GPU, maybe the IoT devices use the compute power.
 
DrMrLordX said:
I guess? The cores are tiny, yields should be (relatively) good, and they could tamp down TDP by dropping boost clocks a little. Anything they did with 4c on 14nm should be doable with 8c on 10nm.
Click to expand...

This only holds if we are talking same cores - not necessarily when you increasing per core performance (e.g. power and area adder) and increase core count on top of it (another power and area adder). Add to this, that everything we know about Tremont, aside from Intel's polished slides, is the implementation in Lakefield - which points to an awful power efficiency (either performance is lower than expected or power is higher than expected or both)
 
Last edited:
I have read, watched at few places that tigerlake mobile will not be here till november??
Again Intel??
Also performance charts presented look to be total bs.
I don't even...
Intel has major production problem, mindshare problem and this does not help at all.
 
randomhero said:
I have read, watched at few places that tigerlake mobile will not be here till november??
Again Intel??
Also performance charts presented look to be total bs.
I don't even...
Intel has major production problem, mindshare problem and this does not help at all.
Click to expand...
Several specific release dates are given earlier in the thread. I agree the benchmarks so far are..... suspect, but you seem determined that it will be a bad product no matter what.
 
Thala said:
This only holds if we are talking same cores - not necessarily when you increasing per core performance (e.g. power and area adder) and increase core count on top of it (another power and area adder).
Click to expand...

Tremont is very, very small on Intel 10nm+ (look at Lakefield). If they can't yield 8c dice out of those tiny cores reliably then I don't know what to say. It's probably more that Intel doesn't want to bother with it - too many other projects hoovering up 10nm+ wafers. Yet another reason why I think they should have just canned Jasper Lake and moved to a Gracemont successor on 10nm SuperFETs. Unless they're recovering Jasper Lake from the Ridge CPUs, but somehow I doubt that.

Add to this, that everything we know about Tremont, aside from Intel's polished slides, is the implementation in Lakefield - which points to an awful power efficiency (either performance is lower than expected or power is higher than expected or both)
Click to expand...

I had thought the awful power efficiency had more to do with the Sunny Cove core?
 
DrMrLordX said:
Tremont is very, very small on Intel 10nm+ (look at Lakefield). If they can't yield 8c dice out of those tiny cores reliably then I don't know what to say.
Click to expand...

My statement was more of general nature - less particularly pointing at Tremont - and one of the problems of Intel in general. The core area is increasing drastically - which collides with ability to double the core counts with a shrink on a similar die size.
I think thats the main reason why they try to combine cores with higher area efficiency with cores with lower area efficiency - at this point i am everything but convinced that this strategy will work against AMD.

I had thought the awful power efficiency had more to do with the Sunny Cove core?
Click to expand...

Nope, we have benchmarks, where we already know, that Sunny Cove essentially is not participating (i guess this is even the normal case) - like a Cinebench R15 multi score of 250, 7zip multi 7200MIPS - very low values performance wise. But still this thing is drawing power like no tomorrow - compared to the performance anyway.
 
DrMrLordX said:
Yeah, I know, which is a bit odd. You'd think Elkhart Lake would keep to 4c and lower, but not necessarily Jasper. Oh well!
Click to expand...

That was sort of what I was getting at - if there's any actual physical difference between the two, it's due to Elkhart having some additional fixed function blocks aimed at IoT.
 
Thala said:
My statement was more of general nature - less particularly pointing at Tremont - and one of the problems of Intel in general. The core area is increasing drastically - which collides with ability to double the core counts with a shrink on a similar die size.
I think thats the main reason why they try to combine cores with higher area efficiency with cores with lower area efficiency - at this point i am everything but convinced that this strategy will work against AMD.



Nope, we have benchmarks, where we already know, that Sunny Cove essentially is not participating (i guess this is even the normal case) - like a Cinebench R15 multi score of 250, 7zip multi 7200MIPS - very low values performance wise. But still this thing is drawing power like no tomorrow - compared to the performance anyway.
Click to expand...

Doesn’t seem like it is drawing a ton of power to me, especially given the performance. People keep looking at that 50W peak and saying it is a power hog, but it only peaked at 50W briefly and it spent most of it’s time consuming significantly less power. IMO there is nothing wrong with that. Quite frankly, I am intrigued by the platform. It is the first *real* new thing Intel has put out in a long time.

The things that will determine how great it is include heat, battery life, and overall performance. OEMs are already reporting 15-20 hour battery life. We will see what the reviews say.

I am also curious as to what the yields on Intel’s 10SF process are.

An 8 core version of this with a 65W TDP would likely beat all of the current gen chips out there with the possible exception of the 3950X. Given how much power Intel’s current chips use, that is progress.

EDIT: I thought you were referring to Tiger Lake.
 
ondma said:
Several specific release dates are given earlier in the thread. I agree the benchmarks so far are..... suspect, but you seem determined that it will be a bad product no matter what.
Click to expand...
To be honest, I am not expecting much from tigerlake. Maybe I will be pleasantly suprised.
But when competition has 8 core product for months on market, and excellent one, releaseing 4 core so late is bit meh.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top