Discussion Intel current and future Lakes & Rapids thread

[Geddagod]

8 wide skymont is wild lol, even if it's 2x4

 

[Henry swagger]

View attachment 87283
8 wide skymont is wild lol, even if it's 2x4

Would be amzing if skymont has rocket lake ipc 😁

 

[Henry swagger]

https://twitter.com/x/status/1714114078086688825

Skymont aims for alder lake ipc 💻😄😄 .. revenge of the e-cores is here lol

 

[S'renne]

View attachment 87283
8 wide skymont is wild lol, even if it's 2x4

Are we expecting a notable MT gains but less so for ST on Arrow Lake with way more performant E cores

 

[Kepler_L2]

Are we expecting a notable MT gains but less so for ST on Arrow Lake with way more performant E cores

AFAIK both cores have large IPC gains but LNC fmax is low hence the mediocre ST bump.

 

[igor_kavinski]

It's funny how Intel gets assisted for their part in global warming: https://submer.com/press/intel-coll...-the-path-to-immersion-cooling-for-1000w-tdp/

Absolutely no reason for them to dial back on their CPU core power consumption. In fact, this allows them to start making even more inefficient CPUs!

Coming soon to a water heater near you, the Intel Core iLovetoHeatyouUp 1000W! Making homes warm and cozy without letting waste heat go to waste!

 

[Ranulf]

It's funny how Intel gets assisted for their part in global warming: https://submer.com/press/intel-coll...-the-path-to-immersion-cooling-for-1000w-tdp/

Absolutely no reason for them to dial back on their CPU core power consumption. In fact, this allows them to start making even more inefficient CPUs!

Coming soon to a water heater near you, the Intel Core iLovetoHeatyouUp 1000W! Making homes warm and cozy without letting waste heat go to waste!

Haha, a water heater that mines crypto on the side. Brilliant.

 

[SiliconFly]

AFAIK both cores have large IPC gains but LNC fmax is low hence the mediocre ST bump.

I think it's a bit too soon to speculate on LNC's Fmax as we still dunno how performant 20A is. A clock regression may not be there at all.

 

[Dayman1225]

I think it's a bit too soon to speculate on LNC's Fmax as we still dunno how performant 20A is. A clock regression may not be there at all.

It’s not all about node, sometimes new cores with higher IPC just can’t clock as high

 

[DAPUNISHER]

Without a serious performance per watt improvement it isn't going to move the needle for the DIY community. A margin of error percentage level of gaming advantage isn't going to do anything for sales to our crowd either. Of course there is the point that unless things change, it is really no big deal how 14th gen desktop performs. The desktop market is in a historic slump. OEMs love the new shiny to help sell their stuff, so that's its best chance to have any impact.

Don't think most people care about that tbh, esp if prices are alright. Though that's my "US-based" perspective, I think electricity elsewhere is way more expensive

This launch is an object example of how erroneous the cheap power taking point can be. The terrible performance per watt for gaming is the death knell in most reviews.

I have been reading some try to put lipstick on this pig; good luck with that.

It's a damned shame. I think they were better just producing these primarily for OEMs and S.I.s. With none of the usual media blitz. Certainly the tech press would have still bought and tested 14th gen, but it would have greatly muted the effect on the DIY audience. As we all understand that the prebuilt makers want new shiny stuff to market their products with.

Instead it turned into this -

 

[Geddagod]

Would be amzing if skymont has rocket lake ipc

Tbh, given that GRC is ~SKL IPC, it wouldn't be too out of left field to expect skymont to have CYPC IPC...

Skymont aims for alder lake ipc 💻😄😄 .. revenge of the e-cores is here lol

Tbh I think that's a bit too hopeful. That's like 2 P-core generations worth of IPC increases, which would mean (for the P-cores to not look outright stupid in comparison to the E-cores) that the P-cores (LNC) should be ~2 P-core generations worth of IPC increases as well, which doesn't seem like it's happening....

AFAIK both cores have large IPC gains but LNC fmax is low hence the mediocre ST bump.

This is prob the best case scenario for Intel. Intel has had this recent trend where they release new archs with mid ST clocks and then boost them really high in the subsequent generation.
The only problem is that it also required a new die/node update, which doesn't sound like it's happening with ARL-R (with the exclusion of the 8+32 die, but doubt this one die includes core level changes).

This launch is an object example of how erroneous the cheap power taking point can be. The terrible performance per watt for gaming is the death knell in most reviews.

Tbh, didn't expect the power diff to be that high. Haven't looked at gaming power consumption figures in a while. That's mb.
The cheap power talking point is a legitimate talking point though. There's a reason the 12900k still sold pretty well in comparison to vanilla Zen 3, despite being disproportionately power hungry. The death knells in reviews aren't coming from the outright massive power consumption diff, but the fact that the 7800x3d is often beating the 14900k while consuming so much less power.

I think they were better just producing these primarily for OEMs and S.I.s. With none of the usual media blitz.

Has Intel ever done that before?

but it would have greatly muted the effect on the DIY audience.

Eh. I think the DIY community would have bullied Intel harder for trying to hide it lol. And it would prob be justified tbh.

 

[msj10]

The only problem is that it also required a new die/node update, which doesn't sound like it's happening with ARL-R (with the exclusion of the 8+32 die, but doubt this one die includes core level changes).

It would be a massive disappointment if the 8+32 die is still on N3B in late 2025. they should at least port it to N3P.

 

[Geddagod]

It would be a massive disappointment if the 8+32 die is still on N3B in late 2025. they should at least port it to N3P.

Don't think Intel is going to go through all the effort of porting it to N3P, which has different design rules than N3B IIRC, when it would be much easier for them to just expand the 8+16 die, IMO.
The naming of "ARL-R" really makes it seem like it's the same dies across the stack with one new die being added which is just tweaking the old die with a longer ringstop and more cores lol.

 

[H433x0n]

This is prob the best case scenario for Intel. Intel has had this recent trend where they release new archs with mid ST clocks and then boost them really high in the subsequent generation.
The only problem is that it also required a new die/node update, which doesn't sound like it's happening with ARL-R (with the exclusion of the 8+32 die, but doubt this one die includes core level changes).

What makes you think ARL-R will still be on N3B? I don’t think there’s been any information on this since it’s still 2 years away. If they’re sticking with TSMC I’d imagine they’d at least move to another flavor of N3. If they’re doing an internal node it’d likely be 18A.

It’s probably going to play out similar to ADL -> RPL if I had to guess.

 

[H433x0n]

Don't think Intel is going to go through all the effort of porting it to N3P, which has different design rules than N3B IIRC, when it would be much easier for them to just expand the 8+16 die, IMO.
The naming of "ARL-R" really makes it seem like it's the same dies across the stack with one new die being added which is just tweaking the old die with a longer ringstop and more cores lol.

It seems like you’re going off of MLID’s data but his accuracy 2 years out is pretty bad. I wouldn’t read too much into it.

 

[Geddagod]

It seems like you’re going off of MLID’s data but his accuracy 2 years out is pretty bad. I wouldn’t read too much into it.

Nah, Raichu was also talking about the 8+32 ARL die launching a year after the OG ARL lineup.

 

[Geddagod]

What makes you think ARL-R will still be on N3B? I don’t think there’s been any information on this since it’s still 2 years away. If they’re sticking with TSMC I’d imagine they’d at least move to another flavor of N3. If they’re doing an internal node it’d likely be 18A.

It’s probably going to play out similar to ADL -> RPL if I had to guess.

The word "refresh".
If it doesn't have a new codename, then I doubt it's going to change much. Even RPL had a new codename lol.
And Intel 18A "ARL" is Panther Lake... which doesn't appear to be a full lineup.

 

[DAPUNISHER]

Tbh, didn't expect the power diff to be that high. Haven't looked at gaming power consumption figures in a while. That's mb.
The cheap power talking point is a legitimate talking point though.

I agree, just not in this instance.

There's a reason the 12900k still sold pretty well in comparison to vanilla Zen 3, despite being disproportionately power hungry.

Citation needed. AMD, and in particular Zen 3 has consistently dominated retail since release from all indications. Be it weekly Mind Factory stats, top 10 on Amazon and Newegg, or the recent Techpowerup poll.

The death knells in reviews aren't coming from the outright massive power consumption diff, but the fact that the 7800x3d is often beating the 14900k while consuming so much less power.

This amounts to splitting hairs. Your original reply to me was over my comment concerning performance per watt needing to be a massive improvement to move the needle. All we have done here is circle back.

Has Intel ever done that before?

Don't recall. It would however, simply have numbered among the things Intel has been forced into doing, that they have never done before.

Eh. I think the DIY community would have bullied Intel harder for trying to hide it lol. And it would prob be justified tbh.

Only the detractors that always rake them over the coals regardless, would have tried to make it a big issue. The tech press would have approached it differently IMO. That would have helped control the narrative. Even keeping the 13th gen designation would have helped. They laid off over 10K in marketing and advertising; are the chickens coming home to roost?

 

[H433x0n]

The word "refresh".
If it doesn't have a new codename, then I doubt it's going to change much. Even RPL had a new codename lol.
And Intel 18A "ARL" is Panther Lake... which doesn't appear to be a full lineup.

Suppose we’ll see. If it remains on N3B there’s literally nothing to refresh for the lower core count SKUs. What’s going to be improved? The Skylake & Alder Lake refreshes were only possible because they controlled the node and even in those instances it didn’t really work.

Why would they purposely clock ARL-S less than is possible to make headroom available for a refresh when it’s already barely keeping its head above water in ST perf?

 

[Geddagod]

Be it weekly Mind Factory stats,

Mindfactory has AMD dominating Intel even if AMD released a pile of poop. Just like the Japanese market favors Intel, at least from what I hear.

top 10 on Amazon and Newegg,

Right now the 12900kf is 38th and the 5950x is... 35th.
In newegg the 12900k is 9th, and the 5950x is 37th...
(US)

or the recent Techpowerup poll.

No idea what that is
Overall though, the 12900k was reviewed very favorably despite the power consumption. One would think it translated well to sales, and it looked like it did, even despite Zen 3's numerous sales advantages (mobo upgrades, memory).

This amounts to splitting hairs.

I don't find the distinction of power consumption vs perf splitting hairs, especially when with 12th gen it's shown that higher power consumption is often excused in regards to higher perf.

Your original reply to me was over my comment concerning performance per watt needing to be a massive improvement to move the needle. All we have done here is circle back.

I've already said I misremembered RPL's original power consumption in gaming difference, and that's my bad. I don't know what else you want me to do, get on my knees? 🤣

Don't recall. It would however, simply have numbered among the things Intel has been forced into doing, that they have never done before.

Wouldn't be necessary

Only the detractors that always rake them over the coals regardless, would have tried to make it a big issue. The tech press would have approached it differently IMO. That would have helped control the narrative. Even keeping the 13th gen designation would have helped. They laid off over 10K in marketing and advertising; are the chickens coming home to roost?

I expect the overall decision to come from a select few higherups, not from the numerous lower level employees they laid off
The tech press wouldn't have approached it much differently IMO. Maybe the amount of noise it made would be lower, but the intensity of the ridicule (even if the total amount was lower) would almost certainly be higher, since it would appear as if Intel was trying to hide the new gen.
Lastly, keeping the 13th gen designation wouldn't have helped sales for OEMs.
They could have kept the 13th gen designation for DIY and 14th gen for OEMs (to boost sales for them), but that would just be very confusing, and honestly give Intel an excuse to increase 14th gen pricing and not decrease 13th gen.

 

[Geddagod]

Suppose we’ll see. If it remains on N3B there’s literally nothing to refresh for the lower core count SKUs. What’s going to be improved? The Skylake & Alder Lake refreshes were only possible because they controlled the node and even in those instances it didn’t really work.

Why would they purposely clock ARL-S less than is possible to make headroom available for a refresh when it’s already barely keeping its head above water in ST perf?

There was no alder lake refresh
Even Skylake's refreshes got new codenames lol
As for what's going to be improved? Well obviously, not much.
And who said they would purposefully clock ARL-S less than possible? I'm sure they will clock it as far as possible.

 

[DAPUNISHER]

I surmised you would be overly argumentative for your own entertainment purposes. You do you bruh.

1. Since you failed to provide citation. 2. Are relying on hearsay as a debating tactic. And 3. Employed the moving the goalpost fallacy, there is no reason for my continuing to belabor the point. I am good with what I wrote. I as always, instead of fruitlessly debating ad nauseum, leave it to the reader to decide. I yield the floor, you can have the last word.

BTW, here is their brief article about that TPU poll showing Intel getting merked in retail among their readers - https://www.techpowerup.com/314665/...with-tpu-readers-23-market-share-poll-results

Aussie Steve made the same observation/ drew the same conclusions both myself (in the Zen 3 builders thread) and TPU did. Video starts at correct time -

My conclusion: This launch was a disaster. It did nothing to move the needle, and it further hurts Intel's street cred with the DIY/retail crowd. Those trying to put lipstick on this pig are wasting their time and energy IMhumbleO.

 

[Geddagod]

I surmised you would be overly argumentative for your own entertainment purposes. You do you bruh.

Yesn't

1. Since you failed to provide citation.

Jeezus christ, you want me to SS it? Is it so deep that you couldn't take my word on it?

2. Are relying on hearsay as a debating tactic.

Errr?

And 3. Employed the moving the goalpost fallacy

Hold up, I admitted I was wrong, and explained what I thought originally. You disagreed with that as well (by claiming ADL did not sell nearly as well, or wasn't received very well, bcuz of higher power consumption) and are now trying to say I'm moving goal posts? Blud, don't respond to that point at all if you want to say now I'm moving goal posts, not after you already started responding to the other point I was talking about...

I as always, instead of fruitlessly debating ad nauseum, leave it to the reader to decide.

Can't wait for the readers polls to start.

I yield the floor, you can have the last word.

How generous of you.

BTW, here is their brief article about that TPU poll showing Intel getting merked in retail among their readers -

... You mean exactly what I said here :

The death knells in reviews aren't coming from the outright massive power consumption diff, but the fact that the 7800x3d is often beating the 14900k while consuming so much less power.

?

Those trying to put lipstick on this pig are wasting their time and energy IMhumbleO.

No one here is trying to put lipstick on this "pig" other than perhaps for the 14700k, which reviews were much more positive about. This is pretty much making mountains out of molehills.

and it further hurts Intel's street cred with the DIY/retail crowd.

Literally all your other suggestions would have likely hurt Intel's street cred just as much

 

[Markfw]

Yesn't

Jeezus christ, you want me to SS it? Is it so deep that you couldn't take my word on it?
View attachment 87490
View attachment 87491

Errr?

Hold up, I admitted I was wrong, and explained what I thought originally. You disagreed with that as well (by claiming ADL did not sell nearly as well, or wasn't received very well, bcuz of higher power consumption) and are now trying to say I'm moving goal posts? Blud, don't respond to that point at all if you want to say now I'm moving goal posts, not after you already started responding to the other point I was talking about...


Can't wait for the readers polls to start.

How generous of you.

... You mean exactly what I said here :

?

No one here is trying to put lipstick on this "pig" other than perhaps for the 14700k, which reviews were much more positive about. This is pretty much making mountains out of molehills.

Literally all your other suggestions would have likely hurt Intel's street cred just as much

Just going to add this one time in support of Dapunisher.

First, as old as they are the 5950x is selling faster on ebay and the like. Mine are going for $360. (with a motherboard for a little more)
Second, its well known that the 7950x3d is the fastest gaming CPU, so I don't know why you included that quote that the Intel 12900k is the fastest. Even the 13900k and the 14900k are faster than the 12900k.

I love the "Those trying to put lipstick on this pig are wasting their time and energy IMhumbleO."


As for the rest, I will let Dapunisher respond if he wants to.

 

[Geddagod]

First, as old as they are the 5950x is selling faster on ebay and the like. Mine are going for $360. (with a motherboard for a little more)

I'm sure Intel has no second hand market what so ever as well.

Second, its well known that the 7950x3d is the fastest gaming CPU, so I don't know why you included that quote that the Intel 12900k is the fastest.

Wha...
It was in response to how the "cheap power talking point" claim. I maintain it has a place still, even if it's not very useful in the 14900k vs 7950x3d discussion, (due to there not being any perf advantage for the cost of higher power consumption for those specific skus).
Bringing up ADL vs Zen 3 is an example of the "cheap power talking point" claim being valid at times, due to ADL being well received despite having higher power consumption, because there is a perf advantage.

Just going to add this one time in support of