Discussion Intel current and future Lakes & Rapids thread

[SiliconFly]

Lunar lake is the big one from insider sources will beat apple in all stacks... beating apple in power efficiency is the goal for intel.. apple are undisputed in performance per watt on laptop

Apple is in a different league. Even Lunar Lake might have trouble competing with the mature Apple Soc. But since LNL is a long way ahead & due to lack of information, all we can do now is just speculate.

 

[SiliconFly]

(Intel 14th gen laptop lineup (MTL) is chiplet based, but sadly the 14th gen desktop lineup (RPL+) is still monolithic on Intel 7.)

But something very interesting is happening with Intel starting 2024...

(1) Intel's going 100% chiplet based.
(2) Getting rid of all outdated nodes like Intel 7 & Intel 4.
(3) All client processes are moving to cutting-edge node with GAAFET & BPD (RibbonFET & PowerVia).
(4) Getting rid of all old architectures like RWC (very likely).
(5) Introduction of ADM cache.

That is, brand new p-core u-arch, advanced new nodes, 100% tile-based (both laptop & desktop) with more focus on power-efficiency.

In other words, Intel is actually turning a new leaf next year!

 

[igor_kavinski]

In other words, Intel is actually turning a new leaf next year!

Dial back on the caffeine

 

[SiliconFly]

Yep. I'm just a bit excited about the brand new 20A. Whenever they say 10, 7, 4, etc, it's just infuriating cos they're all seriously outdated nodes. Glad they're getting rid of all the old & outdated baggage next year.

It's a fresh start!

 

[soresu]

(2) Getting rid of all outdated nodes like Intel 7 & Intel 4.

No foundry just gets rid of outdated production nodes.

There are countless fabless companies that work with older nodes which become much cheaper as the newer, cutting edge nodes take prominence.

Case in point Raspberry Pi, who waited until 2019 to pull the trigger on their 28nm design RPi4 SoC, when 7nm was already in production for sometime.

Some customers simply cannot afford state of the art fab node prices.

 

[FangBLade]

Well, it looks like 2023 is the first time in many years Intel is going to have a comfortable lead both in desktop cpu performance and laptop power-efficiency. Next 6 months is gonna be very interesting for Intel...

8 more days to go...

Can you share a link with us to the results?

 

[SiliconFly]

No foundry just gets rid of outdated production nodes.

There are countless fabless companies that work with older nodes which become much cheaper as the newer, cutting edge nodes take prominence.

Case in point Raspberry Pi, who waited until 2019 to pull the trigger on their 28nm design RPi4 SoC, when 7nm was already in production for sometime.

Some customers simply cannot afford state of the art fab node prices.

Oops. I meant for Intel client cpus

 

[FangBLade]

Oops. I meant for Intel client cpus

As I expected, there's no response from you , I've noticed that you're using a lot of words like Intel is the best, most efficient, etc., and when I asked you for testing results for the second time, you ignored it.

 

[SiliconFly]

As I expected, there's no response from you , I've noticed that you're using a lot of words like Intel is the best, most efficient, etc., and when I asked you for testing results for the second time, you ignored it, so it's quite apparent that this is trolling in the style of another user who's using the same pattern.

Not at all. This has been discussed at length already. 14th gen is expected to be roughly up to 9% - 10% faster. MTL is expected to be very power-efficient. It's all in this thread and/or MTL/ARL/LNL thread. And also, many other sites too. So, pls don't expect any reply from me. With all due respect, I think you're the one who's trolling.

 

[lightisgood]

Thunderbolt 5 Debuts, 120 Gbps Speed is 3x Faster Than Previous Gen

240W charging cables are universal, too.

www.tomshardware.com

Thunderbolt 5 will come in 2024 !!

 

[FangBLade]

Not at all. This has been discussed at length already. 14th gen is expected to be roughly up to 9% - 10% faster. MTL is expected to be very power-efficient. It's all in this thread and/or MTL/ARL/LNL thread. And also, many other sites too. So, pls don't expect any reply from me. With all due respect, I think you're the one who's trolling.

You use the word 'expected' a lot , and before that, you jumped to conclusions like it's definitely, done deal, etc. .

 

[SiliconFly]

You use the word 'expected' a lot , and before that, you jumped to conclusions like it's definitely, done deal, etc. .

A gentleman's bet. Just the two of us. Lets see who gets it more accurately. Just a 9 day wait...

 

[FangBLade]

A gentleman's bet. Just the two of us. Lets see who gets it more accurately. Just a 9 day wait...

Betting on what? Computer components? Hmm, no thanks, just go ahead and play with your little buddies. Seeing how excited computer components make you, I don't want to bother you.

 

[dullard]

You use the word 'expected' a lot , and before that, you jumped to conclusions like it's definitely, done deal, etc. .

The word 'Expected' means "to consider probable or likely". Probable is not a done deal. Neither is "likely". But you think he is saying a done deal. SiliconFly also uses words such as "looks like". Still not a definitive claim. https://www.oed.com/search/dictionary/?scope=Entries&q=expected

There aren't benchmarks yet of finished products on finished BIOS. We probably won't even get many benchmarks on the Sept launch. We'll have to wait for mid-October at the earliest for anything definitive. But you ask for links to benchmark results. Why ask for something that doesn't exist yet?

Hmm, no thanks, just go ahead and play with your little buddies.

He is excited by a big change coming soon. Why ridicule him (which is against our rules)?

 

[FangBLade]

The word 'Expected' means "to consider probable or likely". Probable is not a done deal. Neither is "likely". But you think he is saying a done deal. SiliconFly also uses words such as "looks like". Still not a definitive claim. https://www.oed.com/search/dictionary/?scope=Entries&q=expected

There aren't benchmarks yet. We probably won't even get many benchmarks on the Sept launch. We'll have to wait for mid October at the earliest. But you ask for links to benchmark results. Why ask for something that doesn't exist yet?

Exactly, he has repeatedly stated that non-existent hardware will beat competition, so he didn't write 'maybe,' 'probably,' etc., but that it will be like that. That's why I asked him for testing results; I thought he might have some information we don't.

 

[dullard]

I know him and another person from other forums and Wccftech, I'll refrain from engaging in a discussion with trolls.

That is your second rule infraction in two posts.

 

[FangBLade]

That is your second rule infraction in two posts.

I've edited the comments, putting him on ignore; we won't poison the topic.

 

[ondma]

Well, it looks like 2023 is the first time in many years Intel is going to have a comfortable lead both in desktop cpu performance and laptop power-efficiency. Next 6 months is gonna be very interesting for Intel...

8 more days to go...

Not sure they even have a lead in desktop performance now, much less "comfortable" one. I think it comes down to workload. They better enjoy whatever advantage they have now, though, while they can, because if Zen 5 even comes close to expectations, and Arrow Lake is as underwhelming as initial leaks show, they are going to be hurting on the desktop for at least a year or two. I take no glee in saying this either, as I would definitely like to see Intel at least competitive or superior to AMD. I was hoping Arrow Lake would set the stage for a series of releases from Intel that kept parity with AMD in performance and made them much more competitive or superior in power consumption as well. Sadly, ARL (no HT, seriously?) looks like the exact opposite.

 

[H433x0n]

But that s not how things work.

When perf increase by 10% perf/watt decrease by at least 10%, so that s not a good comparison since the better performing chip is at a disadvantage in such a comparison, best is to compare at equal perf to see what is the real efficency difference.

Here’s another good example: The 7950X is infinitely more efficient than the 5950X! I don’t think you could increase power enough to get the 5950X to match the performance of 7950X.

That s surely due to the fact that there s surely some short turbo at work even when the TDP is fixed at 142W, we can see that the 13700K@253W perform only 9% better than when set at 142W, wich is not logical, the difference should be more.

No, that’s the point. The VF curve once pushed out of its efficiency range requires exponentially more power for diminishing returns. You’re using this to skew the results in your favor.

You’re trying to compare the 7950X in it’s peak efficiency range to the 13900K outside of it’s peak efficiency range and saying that’s the real results.

 

[lightisgood]

Here’s another good example: The 7950X is infinitely more efficient than the 5950X! I don’t think you could increase power enough to get the 5950X to match the performance of 7950X.


No, that’s the point. The VF curve once pushed out of its efficiency range requires exponentially more power for diminishing returns. You’re using this to skew the results in your favor.

You’re trying to compare the 7950X in it’s peak efficiency range to the 13900K outside of it’s peak efficiency range and saying that’s the real results.

I remember this.

 

[Abwx]

Here’s another good example: The 7950X is infinitely more efficient than the 5950X! I don’t think you could increase power enough to get the 5950X to match the performance of 7950X.

7nm vs 5nm, it s like wanting RKL to match a 8C RPL..

No, that’s the point. The VF curve once pushed out of its efficiency range requires exponentially more power for diminishing returns. You’re using this to skew the results in your favor.

I skew nothing, comparing two CPUs at different perfs for each is to disadvantage the higher performing one since the better perf will translate by lower efficency, why should a CPU that score say 30 000 in CB be put against a CPU that score say 20 000 at same power..?.

If you downclock the former down to 20 000 pts it will have 2.5x the perf/Watt of the latter instead of 1.5x if one is required to perform 1.5x better.

You’re trying to compare the 7950X in it’s peak efficiency range to the 13900K outside of it’s peak efficiency range and saying that’s the real results.

At stock it run between 180W and 205W, and to match it the 13900K must be pushed in the 250-300W region, that s what computerbase measurement show explicitely.

Get used to it, if Intel CPU was that good they wouldnt have pushed power that high starting with the 12900K, at the time AMD s AM4 plateform was tailored for 142W max, so they couldnt even release a higher clocked part since the 5950X was already at this power level, they increased the TDP on AM5 because Intel did it as a mean to perform better at the expense of efficency.

 

[H433x0n]

7nm vs 5nm, it s like wanting RKL to match a 8C RPL..

You don’t see the irony in that statement?

 

[soresu]

Thunderbolt 5 Debuts, 120 Gbps Speed is 3x Faster Than Previous Gen

240W charging cables are universal, too.

www.tomshardware.com

Thunderbolt 5 will come in 2024 !!

240W charging........



😂🤣😆

 

[igor_kavinski]

Sadly, ARL (no HT, seriously?) looks like the exact opposite.

It should have the rentable cores feature which could be a superior alternative if Intel engineers have done their homework.

 

[Abwx]

You don’t see the irony in that statement?

The irony will be when Intel launch ARL, you ll see that they ll advertise huge numbers in comparison with RPL perf/watt wise, and yet it wont be more efficient than the 7950X...