Discussion Intel current and future Lakes & Rapids thread

[A///]

Because in its whole history AMD never made such derogatory statements, Intel on the other hand routinely practice such baseless statements, among others they once made a slide to spread FUD and saying that AMD was using defective dies for the Phenom X3...

AMD as far back as I can remember prior to their x86 lawsuit never disparaged Intel by name. Only referring to them as the competition or our competitors. It's kinda classy.

 

[A///]

Is Gregory Bryant's leaving Intel related to that? I remember he infamously called AMD a copycat during the Tiger Lake launch presentation without mentioned AMD. He kept his job after that and but left soon after when the Alder Lake live overclocking failed during its launch presentation.

He may have been on thin ice after that. he left in january 2022 and joined analog devices a couple months later. amd dumped a few marketing people a few years ago, guy and woman team forget their names, intel hired them and then let them go and then they made a third party and then hired that third party. the bald weirdo from amd is now at intel. the semiconductor industries non engineering people are like in one big incestuous relationship.

 

[A///]

hassan's crap rag has published a new rumor.

LGA 1851 Socket Longevity Planned Uptill 2026
DDR5 Only Compatibility, No DDR4 Support
Kicks off With 800-Series Motherboards
Support For Up To DDR5-6400 Memory (Native JEDEC)
Increased PCIe Gen 5.0 Lanes Through CPU & PCH
Arrow Lake-S First Desktop Family Supported
Arrow Lake-S CPUs feature 3 MB L2 Cache Per P-Core
Arrow Lake-S CPUs feature refreshed Alchemist iGPUs
Arrow Lake-S CPUs feature integrated LLC "Adamantine" for GPU Tile
Arrow Lake-S CPUs feature 8+16, 8+0, 6+8 CPU SKUs
Launching In 2H 2024

 

[A///]

the socket longevity is interesting in that both hassan's crap rag or theo ther crap rag are erring on the side of caution of there being a refresh unless intel has designed 1851 in such a way that 3 unique generations of processors can be outfitted to the same mobo. this may be in response to amd's ability to do as much. expect intel board prices to climb and mirror that of am5 prices.

 

[A///]

Socket 1851 or Z890 seem to be Intel's first real attempt at offering board functionality and longevity in unison or parity with amd's. If I'm even correct the age old argument of amd offering value tdue to board life goes out the window. what will remain instead is raw performance and whetgher or not your need to take out additional fire insurance on your home or flat if you choose intel.

 

[igor_kavinski]

It could also mean three generations of lame Intel CPUs on Z890, followed by their Royal Core based supposed-AMD-butt-kicker on Z990 in Q4 2026 or Q1 2027.

 

[H433x0n]

It could also mean three generations of lame Intel CPUs on Z890, followed by their Royal Core based supposed-AMD-butt-kicker on Z990 in Q4 2026 or Q1 2027.

If ARL-S ends up performing per the projections from IgorsLab, it’d be solved in the first refresh.

By the 2nd and 3rd generation of LGA1851 it’ll reach >=5.2ghz boost clocks and it’ll have the expected uplifts.

 

[adroc_thurston]

If ARL-S ends up performing per the projections from IgorsLab, it’d be solved in the first refresh.

By the 2nd and 3rd generation of LGA1851 it’ll reach >=5.2ghz boost clocks and it’ll have the expected uplifts.

How would refresh make LNC better?
They're running out of things to bloat and it's a synthetisable core so you ain't making it go faster as is.

 

[H433x0n]

How would refresh make LNC better?
They're running out of things to bloat and it's a synthetisable core so you ain't making it go faster as is.

A different node with further DTCO. Looking at the history of skylake and golden cove it’s hardly a stretch to believe they’ll get more frequency out of a core arch over time.

Similar to how Zen 4 achieved much greater frequency over Zen 3 (developed with either the same or similar design toolset).

 

[adroc_thurston]

A different node with further DTCO

Not happening, 18A DT is like Panther Lake.

Looking at the history of skylake and golden cove

The old ways.

Similar to how Zen 4 achieved much greater frequency over Zen 3

That's a shrink to a node with SiGe fins and it had to add some L2/3 cycles to keep the speed.
No bueno.

 

[H433x0n]

Not happening, 18A DT is like Panther Lake.

So you’re saying a year after ARL-S, they won’t have 20A refined or 18A available? If some of the ARL-S SKUs are indeed 20A, I really doubt that a full year later they won’t be able to produce the entire product line on an updated version of 20A.

I’m not that pessimistic, if there’s one thing Intel has proven they are capable of it’s adding more frequency and performance to existing product lines. A variant of LNC will reach >=5.0ghz, I’m pretty confident with that statement.

They will create a 20A/18A UHP library and make the changes necessary to get it to work. They’re not going to spend 3 years on a core arch that’s incapable of hitting 5.0ghz. I’m not totally convinced that ARL-S itself won’t get close to 5.0ghz (it’ll hit at least 4.7ghz boost clocks).

 

[Saylick]

A variant of LNC will reach >=5.0ghz, I’m pretty confident with that statement.

*Ahem*

 

[Henry swagger]

Don't forget Pat Gelsinger once said that AMD / Ryzen was in the rear view mirror.



I will say that the people who favor AMD are extremely sensitive about that comment. It gets mentioned all of the time in numerous discords and subreddits I frequent.

Pat was right intel has 80% market share.. arrow lake will deliver unquestioned leadership

 

[adroc_thurston]

A variant of LNC will reach >=5.0ghz, I’m pretty confident with that statement.

Clocks aren't really down versus MTL, it's just not a very good core.

I’m not totally convinced that ARL-S itself won’t get close to 5.0ghz (it’ll hit at least 4.7ghz boost clocks).

Clocks are the easy part; not really what they're struggling with at large.

 

[A///]

i wouldn't play emphasis on a mobile soc as the harbinger of success for either company

 

[H433x0n]

Clocks aren't really down versus MTL, it's just not a very good core.

Clocks are the easy part; not really what they're struggling with at large.

Then what are they struggling with?

As of now it doesn’t appear they’ll hit the same clocks they are achieving with MTL (certain MTL SKUs will reach 5.2ghz).

 

[adroc_thurston]

Then what are they struggling with?

IDC is no good at making big cores.
You will see what I mean in <3 quarters.

Atom guys are by far more capable even if client Skymont seems middling (but CWF looks promising at least).

they’ll hit the same clocks they are achieving with MTL

Clocks aren't the issue (it clocks less relative to RPL but that's the price you pay for portable designs with higher IPC).

 

[H433x0n]

IDC is no good at making big cores.
You will see what I mean in <3 quarters.

Atom guys are by far more capable even if client Skymont seems middling (but CWF looks promising at least).

Clocks aren't the issue (it clocks less relative to RPL but that's the price you pay for portable designs with higher IPC).

afaik, ARL won't be released within that time window.

 

[Joe NYC]

IDC is no good at making big cores.
You will see what I mean in <3 quarters.

Is that the timeframe for Intel Redwood Cove? In Granite Rapids?

I am bad at keeping up with Intel roadmaps, but isn't that the same core that will also be in Meteor Lake?

 

[adroc_thurston]

ARL won't be released within that time window.

Wasn't talking ARL at all.
There are far, far more important and interesting products launching in 2024.

Is that the timeframe for Intel Redwood Cove?

MTL is a wee bit earlier but yes.
GNR vs Turin will be a fight to ugh, remember.
~same core count and wattage and and all.
Very fair! (Not really and you'll see why).

 

[H433x0n]

Wasn't talking ARL at all.
There are far, far more important and interesting products launching in 2024.

What does Zen 5’s release have to do with LNC performance? Or as you said Intel’s design team being bad?

They’re not mutually exclusive. Zen 5 performing well doesn’t have any effect on the performance of ARL.

 

[SpudLobby]

At least I was spot on about this a while back...

I did say ARL is gonna get a bigger cache. News leaks suggest that ARL is getting a 50% bump in L2. i.e, up to 3MB of L2.

Looks like, instead of doing something great with the massive transistor budget, Intel has chosen the easiest & safest way instead. Yuck.

I wonder what the cache bump is about. But either way that's good for energy consumption.

 

[adroc_thurston]

Or as you said Intel’s design team being bad?

Yes.
Let's just say IDC underdelivered.

What does Zen 5’s release

Not just it really, there's another bloatcore in 2024 too.

Zen 5 performing well doesn’t have any effect on the performance of ARL

It does because they're all chongus cores now.

 

[H433x0n]

GNR vs Turin will be a fight to ugh, remember.
~same core count and wattage and and all.
Very fair! (Not really and you'll see why).

Neither company has any interest in a fair fight. It’s a blood sport with billions of dollars on the line.

An academic exercise comparing Specint & Specfp scores with the same specs isn’t the goal.

It does because they're all chongus cores now.

That was always inevitable, as you should know.

 

[adroc_thurston]

An academic exercise comparing Specint & Specfp scores with the same specs isn’t the goal

Oh but this gen is the closest we'll ever get to a semi-fair fight.

That was always inevitable, as you should know.

Some do it better than the others.
As I've said, only a few quarters of waiting left!