Discussion Intel current and future Lakes & Rapids thread

[SiliconFly]

One need only look at Intel’s launch cadence to know Arrow Lake is launching Q4.

Arrow Lake on this Q4 sounds very exciting! But those are just rumors. Not even a hint from any official sources. And Q4 is way too optimistic.

If I'm suddenly expected to believe that Intel can do magic and can release a brand new µarch on a revolutionary new node in such a short span of time, I'm gonna take it with a tiny grain of salt!

 

[SiliconFly]

Unless Intel botches everything (likely), Meteor Lake appears to be targeting efficient, low power designs. No 5.5ghz to be found. The evidence I’ve seen indicates these chips will be in premium ultrabooks and such. Think 14 cores (6+8) with a 10-20 hour battery life in a sub 2lb laptop.

Ya. Looks like Intel 4 didn't clock as high as they expected. Thats probably the reason they stopped with i5. Or maybe they've down-clocked it a bit to improve efficiency. Now that it's in volume ramp, I think they'll be releasing some info very soon.

 

[SiliconFly]

amen, that boy ain't right in the head or the tumor he calls his jaw.

He kinda keeps worshiping AMD all the time. And tries to spreads too much FUD & hatred about nvidia & intel. What a low life. Looks like he's one of those AMD's paid leakers.

 

[jpiniero]

Ya. Looks like Intel 4 didn't clock as high as they expected.

It's a yield/supply thing. Presumably Meteor Lake-S reuses the 6+8 P tile but there's no tile that's bigger. The locked desktop i5's currently top out at 5 Ghz for the P-Cores so as long as it's somewhere in the mid 4's it's not going to be a problem if Meteor Lake doesn't clock that high.

 

[Hitman928]

Yes and no. while watts are heat energy in the end your watt to heat ratio isn't always 1 to 1 when dealing with surface area and enclosure. smaller package higher heat while larger packager at same wattage easier to dissipate heat as an example. otherwise you are correct, but this concept has some ifs and butts when you look at external factors. i should stress a bigger surface area is one of many factors.

Wattage is turned 99.99% into heat, it has nothing to do with surface area or package size. Temperature and method of cooling are dependent on those things but heat does not equal temperature.

 

[Abwx]

According to MLID Raptor Lake-R will come with DLVR enabled for a 10-20% power consumption reduction which they will use for higher all core clock speeds.

DLVR increase efficency at lower powers but at full tilt efficency decrease in respect of the same chip without regulators, that s a feature that is usefull for laptops dedicated SKUs.

 

[Geddagod]

Arrow Lake on this Q4 sounds very exciting! But those are just rumors. Not even a hint from any official sources. And Q4 is way too optimistic.

If I'm suddenly expected to believe that Intel can do magic and can release a brand new µarch on a revolutionary new node in such a short span of time, I'm gonna take it with a tiny grain of salt!

No, ARL is officially confirmed for 2024. So is Lunar Lake. Q4 is pessimistic, I think ARL desktop might be out by end of Q3.

 

[A///]

He kinda keeps worshiping AMD all the time. And tries to spreads too much FUD & hatred about nvidia & intel. What a low life. Looks like he's one of those AMD's paid leakers.

he is what we call a moron and a strong enough reason to bring back corporal punishment in public schools.

 

[MarkPost]

The only test I see SPR bring faster in is unreal engine. Threadripper is faster in the rest of them. . .

In Unreal TR is faster too. Lower is better there (seconds).

Spoiler

 

[FangBLade]

Intel 14th Gen Raptor Lake Refresh Desktop CPUs Rumored To Clock Up To 6.2 GHz

Intel's 14th Gen Raptor Lake Refresh Desktop CPUs are expected to further boost the clock speeds compared to 13th Gen Raptor Lake chips.

wccftech.com

Intel 7 ultra node is superb 6.5ghz coming 😃

I like your trolling style, maximum hyping of Intel to the skies, where every move they make is magical and you are thrilled with everything.

 

[eek2121]

Arrow Lake on this Q4 sounds very exciting! But those are just rumors. Not even a hint from any official sources. And Q4 is way too optimistic.

If I'm suddenly expected to believe that Intel can do magic and can release a brand new µarch on a revolutionary new node in such a short span of time, I'm gonna take it with a tiny grain of salt!

Erm, that would be Q4 2024. The actual time frame will likely be late q3/early q4 2024.

Ya. Looks like Intel 4 didn't clock as high as they expected. Thats probably the reason they stopped with i5. Or maybe they've down-clocked it a bit to improve efficiency. Now that it's in volume ramp, I think they'll be releasing some info very soon.

It's a yield/supply thing. Presumably Meteor Lake-S reuses the 6+8 P tile but there's no tile that's bigger. The locked desktop i5's currently top out at 5 Ghz for the P-Cores so as long as it's somewhere in the mid 4's it's not going to be a problem if Meteor Lake doesn't clock that high.

They don’t have the capacity. Nothing wrong with Intel 4. They just can’t make enough chips for the full stack refresh. Frequency has nothing to do with it.

Remember that Intel has to manufacture and support these chips for years.

 

[SiliconFly]

It's a yield/supply thing. Presumably Meteor Lake-S reuses the 6+8 P tile but there's no tile that's bigger. The locked desktop i5's currently top out at 5 Ghz for the P-Cores so as long as it's somewhere in the mid 4's it's not going to be a problem if Meteor Lake doesn't clock that high.

You're right. Looks more like a yield issue rather than a clock speed regression. Or maybe both.

 

[SiliconFly]

I like your trolling style, maximum hyping of Intel to the skies, where every move they make is magical and you are thrilled with everything.

Actually, Intel 7+ (or 10 ESF++) is a extremely mature, performant and overall a terrific node that is a lot better than TSMC N7.

 

[SiliconFly]

Erm, that would be Q4 2024. The actual time frame will likely be late q3/early q4 2024.




They don’t have the capacity. Nothing wrong with Intel 4. They just can’t make enough chips for the full stack refresh. Frequency has nothing to do with it.

Remember that Intel has to manufacture and support these chips for years.

After all these years, Intel very badly needs a win. And to win, it need top-end high-performance parts like i9 & i7. No i5s and less.

Instead of making i9 & i7, meteor lake is restricted to low-end parts like i5 & i3. So, we have to conclude Intel 4 has yield issues (and/or maybe a bit of a clock regression).

 

[SiliconFly]

No, ARL is officially confirmed for 2024. So is Lunar Lake. Q4 is pessimistic, I think ARL desktop might be out by end of Q3.

ARL should be out in 2024 Q3 (both mobile & desktop). Hopefully. But only if everything goes according to Pat's plans. And that sounds a bit unlikely.

Consider this:
Intel 4 is a relatively mature node compared to 20A. Intel 4 has been in the works for the past couple of years (four i think). And it looks like Intel 4 still has yield issues! Considering Intel couldn't make top-end i9 & i7 parts in Intel 4 and had to regress to Intel 7 with RPL+ (older architecture on a older node) says a lot about Intel 4. In other words, they switched to their backup plan since their primary plan failed.

If Intel has issues with an almost well-established node like Intel 4, how on earth are they going to deliver a cutting edge product like ARL based on a brand new architecture on a revolutionary new node in just 12 months? Remember 20A is a relatively new node and is barely one year old. It hasn't even come of age yet!!!!!!!

ARL is the one thats supposed to put Intel back in the spotlight. And I want to get my hands on it asap. But my confidence is a bit shaken at the moment after the Meteor Lake debacle!!! (that is, i7 & i9 being RPL)

 

[Geddagod]

ARL should be out in 2024 Q3 (both mobile & desktop). Hopefully. But only if everything goes according to Pat's plans. And that sounds a bit unlikely.

Consider this:
Intel 4 is a relatively mature node compared to 20A. Intel 4 has been in the works for the past couple of years (four i think). And it looks like Intel 4 still has yield issues! Considering Intel couldn't make top-end i9 & i7 parts in Intel 4 and had to regress to Intel 7 with RPL+ (older architecture on a older node) says a lot about Intel 4. In other words, they switched to their backup plan since their primary plan failed.

If Intel has issues with an almost well-established node like Intel 4, how on earth are they going to deliver a cutting edge product like ARL based on a brand new architecture on a revolutionary new node in just 12 months? Remember 20A is a relatively new node and is barely one year old. It hasn't even come of age yet!!!!!!!

ARL is the one thats supposed to put Intel back in the spotlight. And I want to get my hands on it asap. But my confidence is a bit shaken at the moment after the Meteor Lake debacle!!! (that is, i7 & i9 being RPL)

The i7 ad i9 being RPL+ isn't a bad sign for Intel 4 imo. Because RWC doesn't increase IPC/clocks high itself by itself to compete with an extremely mature Intel 7/10mm process hitting 6GHz. With how old (and refined) 10nm and 14nm became, newer nodes are going to face an issue reaching equivalent fmax clocks. However that prob isn't going to be an issue for ARL, considering LNC is supposed to bring a hefty IPC increase, meaning it could bear slightly lower fmax and still get higher perf, and also because Intel 4 won't be nearly as mature as the 10 and 14nm desktop products.
But more importantly about the lack of i7 and i9s, Intel 4 should be capacity limited, it being their first EUV node, and them not having nearly as many EUV tools yet. It would make sense to focus on the premium market first, that being high premium thin and lights. And they will have i7 and i9 mobile parts there, just not for desktop. And considering that mobile is a much more profitable and important segment for Intel, it makes sense to focus there.
Edit: Oh and I forgot to add, ARL desktop is supposed to be on TSMC. So the node problem shouldn't be an issue there.

 

[DrMrLordX]

Edit: Oh and I forgot to add, ARL desktop is supposed to be on TSMC. So the node problem shouldn't be an issue there.

N3-family nodes should be mature enough for Intel to launch Arrow Lake on something like N3e by q4 2024. That assumes that Intel has planned out such a launch years in advance. It's hard to believe that Intel 20a would be ready by then, though. Guess we'll find out when we get there.

 

[SiliconFly]

The i7 ad i9 being RPL+ isn't a bad sign for Intel 4 imo. Because RWC doesn't increase IPC/clocks high itself by itself to compete with an extremely mature Intel 7/10mm process hitting 6GHz. With how old (and refined) 10nm and 14nm became, newer nodes are going to face an issue reaching equivalent fmax clocks. However that prob isn't going to be an issue for ARL, considering LNC is supposed to bring a hefty IPC increase, meaning it could bear slightly lower fmax and still get higher perf, and also because Intel 4 won't be nearly as mature as the 10 and 14nm desktop products.
But more importantly about the lack of i7 and i9s, Intel 4 should be capacity limited, it being their first EUV node, and them not having nearly as many EUV tools yet. It would make sense to focus on the premium market first, that being high premium thin and lights. And they will have i7 and i9 mobile parts there, just not for desktop. And considering that mobile is a much more profitable and important segment for Intel, it makes sense to focus there.
Edit: Oh and I forgot to add, ARL desktop is supposed to be on TSMC. So the node problem shouldn't be an issue there.

Intel 20A is expected to have clock speed regression which directly negates LNC's IPC increase. For ARL i3 & i5 its perfectly fine because they won't be clocked to fmax. It'll be an awesome node for ARL i3 & i5 actually. But a lower fmax will hit the perfomance hard on ARL i7 & i 9. What will they do then? Also, the same capacity issue affects Intel 20A like Intel 4.

Do they do the same like they're doing with MTL? Regress back to Intel 7 or Intel 4 for i7 & i9? Thats sounds tragic. Has intel written all over it.

I dream of a day when they release a ARL i9 in 20A (in 2024 Q3). But sounds very unrealistic. It's May already and hightime for LNC tapeout, but still hasn't happened yet. It's another bad sign

 

[mikk]

Intel 20A is expected to have clock speed regression which directly negates LNC's IPC increase.

Why is this expected? And regression to what? Intel 4? Did we get any informations regarding 20A performance? It's a bit early to talk about, isn't it.

 

[Geddagod]

Intel 20A is expected to have clock speed regression which directly negates LNC's IPC increase. For ARL i3 & i5 its perfectly fine because they won't be clocked to fmax. It'll be an awesome node for ARL i3 & i5 actually. But a lower fmax will hit the perfomance hard on ARL i7 & i 9. What will they do then? Also, the same capacity issue affects Intel 20A like Intel 4.

Do they do the same like they're doing with MTL? Regress back to Intel 7 or Intel 4 for i7 & i9? Thats sounds tragic. Has intel written all over it.

I dream of a day when they release a ARL i9 in 20A (in 2024 Q3). But sounds very unrealistic. It's May already and hightime for LNC tapeout, but still hasn't happened yet. It's another bad sign

No one should expect anything about Intel 20A yet. It's way to early for any leaks about how it will perform.
Historically, products on new (working) nodes for Intel have stagnated clock fmax clock speeds. Even the troubled, but still working Intel 14nm, hit 85% the fmax clocks of the previous top sku, and a new arch's ipc increase of 15-20% should be enough to at least equalize the performance. Plus, since IPC scales worse at higher clocks, and for MT clocks never hit max anyway, even worst case scenario ARL would be an improvement over MTL.
Regardless, the idea of an ARL i9 desktop launching in 2024 shouldn't sound unrealistic because it's on TSMC 3nm.
And LNL did tape out... taped out at the start of the year. Puts it right on track for a q4 2024 launch.

 

[SiliconFly]

No one should expect anything about Intel 20A yet. It's way to early for any leaks about how it will perform.
Historically, products on new (working) nodes for Intel have stagnated clock fmax clock speeds. Even the troubled, but still working Intel 14nm, hit 85% the fmax clocks of the previous top sku, and a new arch's ipc increase of 15-20% should be enough to at least equalize the performance. Plus, since IPC scales worse at higher clocks, and for MT clocks never hit max anyway, even worst case scenario ARL would be an improvement over MTL.
Regardless, the idea of an ARL i9 desktop launching in 2024 shouldn't sound unrealistic because it's on TSMC 3nm.
And LNL did tape out... taped out at the start of the year. Puts it right on track for a q4 2024 launch.

Thats fantastic news. How did i miss LNL tapeout.

https://twitter.com/x/status/1618734128337240066

And how did this news come about that ARL i9 is on TSMC N3? I didn't see it anywhere until now. Or has it been mistaken for the ARL tGPU thats also on TSMC N3? The later sounds more likely as according to Intel itself ARL tCPU is purely a 20A product.

 

[Geddagod]

Thats fantastic news. How did i miss LNL tapeout.

https://twitter.com/x/status/1618734128337240066

And how did this news come about that ARL i9 is on TSMC N3? I didn't see it anywhere until now. Or has it been mistaken for the ARL tGPU thats also on TSMC N3? The later sounds more likely as according to Intel itself ARL tCPU is purely a 20A product.

Intel never claimed Intel cpu tile was 20A only.
ARL having 3nm compute tile variants has been rumored for a while. I think Jim from Adored was the first one though, leaking this document:

 

[Exist50]

according to mlid, both n3e and n3b
To me, n3e seems like the clear choice for Intel for ARL, but who knows

He again demonstrates he has no idea what he's talking about N3B and N3E are not design rule compatible, and no chance Intel's going to port it to both. I'd bet on N3B right now, but we'll see.

DLVR increase efficency at lower powers but at full tilt efficency decrease in respect of the same chip without regulators, that s a feature that is usefull for laptops dedicated SKUs.

Per-core power regulation is also helpful for sparsely threaded loads. The low-frequency/idle cores don't have to be penalized by sharing the same voltage as the bursty ones.

 

[Abwx]

Per-core power regulation is also helpful for sparsely threaded loads. The low-frequency/idle cores don't have to be penalized by sharing the same voltage as the bursty ones.

Read again the point i was answering to.

 

[Geddagod]

He again demonstrates he has no idea what he's talking about N3B and N3E are not design rule compatible, and no chance Intel's going to port it to both. I'd bet on N3B right now, but we'll see

If I were Intel, I would use N3E considering the density hit is slight, but you get better perf and efficiency characteristics from it (as well as cost too most likely).
However since Intel were aiming for ARL to come out end of 2023 originally, using N3E would have been cutting it pretty close. Plus I'm not even sure of the timeline of when TSMC realized N3B was not yielding very well and decided to switch tracks to N3E, if N3E wasn't ever part of the original roadmap even.
ARL development also got an extra years worth of time though due to RPL, so I'm wondering if Intel decided to port the N3B cores over to N3E. Doesn't seem like there is enough time for it though.