Discussion Intel Nova Lake in H2-2026: Discussion Threads

[branch_suggestion]

EMR was like 5M@stop and purportedly decent.
Fingers crossed their L3 stops sucking donkey balls, especially now that they're having shared L2 with less net capacity per core.

Ah yeah the RETVRN of Conroe, well the L1.5d has a purpose after all.
Might lead to some funky results.

 

[adroc_thurston]

Might lead to some funky results.

I wonder if they have a static partitioning mode for NVL shared L2.

I am curious as to what clocks they’re getting (or not) out of a quality node jump

Well >1.1X 1t means the freq jump barely exists if at all.
Their cores are fat, just without the perf.

 

[Io Magnesso]

N2 is wasted on Intel. IMO, desktop is okay but mobile looks grim.

What are you talking about? Both are no good
If I had to say it, it's still better for laptops.

 

[Io Magnesso]

Don't make a fool of a person who works hard, but... A product called NOVA LAKE has been dead since before it was born, so it's okay to be a fool.
However, the Arctic Wolf with the E Core architecture used in Nova Lake is excellent in terms of architecture. Don't lick it

 

[511]

I wonder if they have a static partitioning mode for NVL shared L2.

Well >1.1X 1t means the freq jump barely exists if at all.
Their cores are fat, just without the perf.

Yeah which is a shame considering N2.

EMR was like 5M@stop and purportedly decent.
Fingers crossed their L3 stops sucking donkey balls, especially now that they're having shared L2 with less net capacity per core.

Well they went as far as to reduce L3 stops from 12 to 8 and if they can't make L3 good it's hopeless

 

[adroc_thurston]

Yeah which is a shame considering N2.

It's also on 18A-P so there's that.

Well they went as far as to reduce L3 stops from 12 to 8 and if they can't make L3 good it's hopeless

Stop count doesn't really matter.

 

[poke01]

Well >1.1X 1t means the freq jump barely exists if at all.

It's also on 18A-P so there's that.

So higher clocks on 18A-P?

 

[adroc_thurston]

So higher clocks on 18A-P?

Oh no, N2 is a much, much faster node.

 

[Io Magnesso]

Oh no, N2 is a much, much faster node.

But... N2 and 18A have nothing to do with each other. Because nova lake is garbage

 

[adroc_thurston]

Because nova lake is garbage

It's not garbage, but not leadership either.

 

[Io Magnesso]

However, the process after 3 nm There is no way to get a high clock of 6GHz in a physical phenomenon
N2 is certainly fast
In terms of transistor density, it is the highest speed.

 

[Io Magnesso]

It's not garbage, but not leadership either.

If you don't call NOVA LAKE trash, what would you say?
How do you say zen6? What about Xelite2?

 

[511]

So higher clocks on 18A-P?

Without blowing off the BSPDN I doubt 18AP Could be faster but the hotspots will be difficult to cool.

 

[poke01]

If you don't call NOVA LAKE trash, what would you say?
How do you say zen6? What about Xelite2?

X Elite 2 isnt N2. Qualcomm cores are also not leadership.

 

[adroc_thurston]

Qualcomm cores are also not leadership.

They're certainly trying.

 

[511]

They're certainly trying.

Yes they are trying the only leadership core is Apple right now in terms of ST and PnP characteristics

 

[511]

 

[Fjodor2001]

Couldn't the problem for the P team be that they have a lot of old legacy stuff from previous core generations to carry over to the next generation, but the E team has started more from scratch not so long ago so they have a bit more freedom when doing the design?

 

[poke01]

View attachment 126376

Question is will e core in nova lake surpass the p core?

 

[511]

Couldn't the problem for the P team be that they have a lot of old legacy stuff from previous core generations to carry over to the next generation, but the E team has started more from scratch not so long ago so they have a bit more freedom when doing the design?

They had the time to sort out this stuff with RWC

Question is will e core in nova lake surpass the p core?

Depends on metric maybe it will match the P core in IPC but not in absolute Peak performance.

 

[Darkmont]

https://www.anandtech.com/show/2493/4 The concept and design of Atoms go back to 2004 so it's difficult to wave the "less bloat" magic wand. The truth is that the designers at Haifa IDC are having serious skill issues with the area and power usage to performance they're getting. Also most tocks a team will probably look at fat to trim where they see fit and I doubt legacy bits take up any significant area vs. some fat L2 macro or a big BPU.

 

[511]

https://www.anandtech.com/show/2493/4 The concept and design of Atoms go back to 2004 so it's difficult to wave the "less bloat" magic wand. The truth is that the designers at Haifa IDC are having serious skill issues with the area and power usage to performance they're getting. Also most tocks a team will probably look at fat to trim where they see fit and I doubt legacy bits take up any significant area to vs. some fat L2 macro or a big BPU.

Well most of the designer left Haifa IDC even the Golden Cove's Chief architect Adil Yaoz left for Huawei.
They bled lots of talent in Pat's August Layoff.

 

[adroc_thurston]

https://www.anandtech.com/show/2493/4 The concept and design of Atoms go back to 2004 so it's difficult to wave the "less bloat" magic wand. The truth is that the designers at Haifa IDC are having serious skill issues with the area and power usage to performance they're getting. Also most tocks a team will probably look at fat to trim where they see fit and I doubt legacy bits take up any significant area vs. some fat L2 macro or a big BPU.

Eh modern Atoms have jack to do with Bonnell the OG.
Not a single gimmick survived.

 

[Darkmont]

Eh modern Atoms have jack to do with Bonnell the OG.
Not a single gimmick survived.

Even so, I'm not privy to arguments that there's a mythical glob of shart that if only the designers could rip out unlock peak PPC. Now Royal OTOH, if it's true that x86S was a gimmick specifically for RYC, stripped legacy bits that would've made people unhappy

 

[Io Magnesso]

Even so, I'm not privy to arguments that there's a mythical glob of shart that if only the designers could rip out unlock peak PPC. Now Royal OTOH, if it's true that x86S was a gimmick specifically for RYC, stripped legacy bits that would've made people unhappy

No, the X86S is not like that
Just removing the legacy doesn't increase the PPC