Alder Lake-S leaks

[Grim5]

Heterogenous CPU cores then. It makes a certain amount of sense. I wouldn’t expect all threads to be the same so you could economise with smaller cores; darkbahamut’s point about maximising performance for a die area with some giant, some little (or efficiently) makes sense.

This could just be the first step of many to come, as Moore’s law slows and we seek other ways to optimise: many different core sizes, specialised cores and so on.

Its not about optimizing performance in the way that most people care about, they are optimizing performance per watt for a task, not maximum performance - desktop users usually care about maximum performance, Big.Little is not for maximum performance.

Big.Little is not new, it's been used in processors since 2013 and ever since it's existence has been for one reason and one reason only - to maximize battery life.

Intel is moving to Big.Little because it believes the Desktop market is dying and mobile is the future - it's as simple as that, people can feel how they want to feel about it, it's a simple fact.

 

[muziqaz]

Heterogenous CPU cores then. It makes a certain amount of sense. I wouldn’t expect all threads to be the same so you could economise with smaller cores; darkbahamut’s point about maximising performance for a die area with some giant, some little (or efficiently) makes sense.

This could just be the first step of many to come, as Moore’s law slows and we seek other ways to optimise: many different core sizes, specialised cores and so on.

On higher end desktop it makes zero sense, especially the way Intel is going about it. We are looking at 24 thread CPU, which clearly will be marketed to dummies looking at core counts or thread counts. Unless those 8 big cores are earthshattering (Ryzen shattering) fast, I don't see those making any dent into AMDs high end.
These kind of CPU make complete sense on low end, laptops, if done like ARM does it, not like intel does it.

 

[LePhuronn]

These kind of CPU make complete sense on low end, laptops

But do they? Ryzen 4800U: 8 Zen 2 cores, 8 Vega CUs, 4.2GHz boost clock, 15W TDP. I don't see how a big.little design could help that, just having the fabric ticking over or having the small cores handle dedicated tasks whilst waiting for the big cores to kick in would probably use near enough the same power as that APU in a low-power state.

 

[philo-sofa]

Its not about optimizing performance in the way that most people care about, they are optimizing performance per watt for a task, not maximum performance - desktop users usually care about maximum performance, Big.Little is not for maximum performance.

I agree but as darkbahamut points out, it can be turned towards optimising performance per mm^2 of silicon.

Big.Little is not new, it's been used in processors since 2013 and ever since it's existence has been for one reason and one reason only - to maximize battery life.

Intel is moving to Big.Little because it believes the Desktop market is dying and mobile is the future - it's as simple as that, people can feel how they want to feel about it, it's a simple fact.

I'm familiar with ARM's big.LITTLE or at least the broad strokes of it. Remember reading about it in the register way back when and then seeing the philosophy filter through to actual chips, even my iPhone's SoC owes a lot to it for all that it's an in-house design. It is however new in the x86 space.

Have you considered the possibility that Intel are moving to a similar philosophy because process node upgrades are slowing towards death? Moving towards heterogeneous cores like small and large, as well as dedicated compute, AI and other optimised chips (as cellphones already are) is a way of optimising overall performance from a chip. I doubt Intel quite sees desktop as dead because a) they view desktop as the unimportant cousin of server, which absolutely isn't disappearing and b) it has some utility for power users with Moore's law heading the way it is.

 

[philo-sofa]

On higher end desktop it makes zero sense, especially the way Intel is going about it. We are looking at 24 thread CPU, which clearly will be marketed to dummies looking at core counts or thread counts. Unless those 8 big cores are earthshattering (Ryzen shattering) fast, I don't see those making any dent into AMDs high end.
These kind of CPU make complete sense on low end, laptops, if done like ARM does it, not like intel does it.

Can't say I'm hugely keen, but can see why it might be a valid approach to optimising x86 silicon (certainly for laptops, possibly for desktops). It's also a step along the road to moving x86 towards truly heterogeneous cores, with dedicated cores for tasks (think the cell processor) which is a valid (if difficult to code for) way of optimising performance. We may be forced along that route if Moore's law continues to peter out without any real changes.

 

[muziqaz]

But do they? Ryzen 4800U: 8 Zen 2 cores, 8 Vega CUs, 4.2GHz boost clock, 15W TDP. I don't see how a big.little design could help that, just having the fabric ticking over or having the small cores handle dedicated tasks whilst waiting for the big cores to kick in would probably use near enough the same power as that APU in a low-power state.

Intel feels that it might give them a chance of high performance in under 15W, or higher performance in 15W envelope. I totally get this approach where the power is an issue. This makes zero sense in high performance sector, unless, as said before, those 8 big cores is a la Apple M1 nut crunchers

Can't say I'm hugely keen, but can see why it might be a valid approach to optimising x86 silicon (certainly for laptops, possibly for desktops). It's also a step along the road to moving x86 towards truly heterogeneous cores, with dedicated cores for tasks (think the cell processor) which is a valid (if difficult to code for) way of optimising performance. We may be forced along that route if Moore's law continues to peter out without any real changes.

yeah, MS needs to get their fingers out of their arses too. And If different feature set support rumours are true with Intel, they are going about it the wrong way.

 

[rustyk21]

They're physically smaller of course, it's the entire point. Arm themselves (to who big.LITTLE is a marketing term) call then high efficiency CPUs. The term 'LITTLE' is used in quote or brackets, in full capitals because it's a trademark. Intel themselves (who we are discussing here) have never used the term little at any point. Unless you can point out where they have?



Did I say anything about AMD, or are you making stuff up? It's comparing designs of what would happen if Intel shipped 8+8 or 16 cores. There are already 64 core CPUs on the market if we want to core count circle jerk.

By the way, I own a 5950X. It's a great CPU with amazing performance, but when run like for like with how Intel CPUs are run it definitely doesn't use 125 W!

Unfortunately I think you're wasting your time mate. I think it's an interesting idea and will see where it goes, as from an architectural point of view you could end up with cores that are optimised for certain tasks or scenarios.
It's possible that using a combination of cores you could achieve the same performance as a large number of homogenous cores, at lower cost in silicon, which costs everyone money.
Of course it's also possible that it will completely backfire, but CPU archictures have to evolve so it's best to keep an open mind.

The problem you've got on this forum, is that it's an Intel product and therefore all the cpu design and fabrication (and software) experts here will just jump all over it and crush it into the ground. The mentality is entrenched.

 

[smilingcrow]

The problem you've got on this forum, is that it's an Intel product and therefore all the cpu design and fabrication (and software) experts here will just jump all over it and crush it into the ground. The mentality is entrenched.

As well as that, many on here seem to think if a desktop architecture can't compete in the £450 - £800 range then it's pointless.
This seemingly because that is their area of interest even though overall it's a small segment of the market and is meaningless to the majority.

 

[muziqaz]

As well as that, many on here seem to think if a desktop architecture can't compete in the £450 - £800 range then it's pointless.
This seemingly because that is their area of interest even though overall it's a small segment of the market and is meaningless to the majority.

So you are saying that 8+8 (24 thread) Alder lake will sell below your given price range? Or is it above? If it's above, then let's hope that 8 big cores are Apple like performance

<...>
The problem you've got on this forum, is that it's an Intel product and therefore all the cpu design and fabrication (and software) experts here will just jump all over it and crush it into the ground. The mentality is entrenched.

You don't need to be an expert to see how idiotic 11 gen is, or how stupidly confusing the whole Intel CPU line up is. Don't think for a second, that if it is Intel, they automatically know better. Usually they don't. They might have brilliant engineers, but they are run by short sighted managing morons. The amount of failed products (even the ones which were released) come out in past several years is a testament to that. Intel managing to concede performance crown, and be absolutely slaughtered by 10x smaller competitor, who they themselves were beating soundly just few years back, tells you that, us, keyboard warriors have a point and a reason to doubt the company.

 

[LePhuronn]

The problem you've got on this forum, is that it's an Intel product and therefore all the cpu design and fabrication (and software) experts here will just jump all over it and crush it into the ground. The mentality is entrenched.

Stop baiting, that's just untrue and you know it. If AMD went down a big.little concept for desktop then the same "experts" here would decry it as pointless just as they are doing now. It has zero to do with it being Intel.

Hell, Bulldozer in its concept was very forward thinking, but it was a utter travesty in implementation. I don't see any AMD fanboys defending it just because it's AMD...

 

[Simon Belmont]

I'm hearing Alder Lake coming in September? If true why even bring rocket lake? DOA before it's even out???

 

[LePhuronn]

I'm hearing Alder Lake coming in September? If true why even bring rocket lake? DOA before it's even out???

Apparently so. Funny how we all thought Comet Lake was a pointless, DOA release because of Rocket Lake's impending launch, but if any of the leaked benches are true, Comet Lake is going to be superior!

TBH roll on Alder Lake. Ignoring the big.little issue for now, Golden Cove was designed for 10nm and built on 10nm and by all accounts is an impressive design. Let's see what those cores can do. I'm hoping though they're not doing an Nvidia and rushing Alder Lake out in a panic because they can't have Zen 3 kicking their ass for an entire year and then still can't beat Zen 4.

 

[4K8KW10]

I'm hearing Alder Lake coming in September? If true why even bring rocket lake? DOA before it's even out???

Let's first clarify what Alder Lake actually is. It is a hybrid, 3D-stacked die with the name Foveros https://en.wikichip.org/wiki/intel/foveros
10nm for the cores and 14nm for the uncore parts...

It depends on the volumes that Intel will be able to maker and ship, and if those are low, Rocket Lake would serve as an important supplementary backup.

 

[rustyk21]

So you are saying that 8+8 (24 thread) Alder lake will sell below your given price range? Or is it above? If it's above, then let's hope that 8 big cores are Apple like performance

You don't need to be an expert to see how idiotic 11 gen is, or how stupidly confusing the whole Intel CPU line up is. Don't think for a second, that if it is Intel, they automatically know better. Usually they don't. They might have brilliant engineers, but they are run by short sighted managing morons. The amount of failed products (even the ones which were released) come out in past several years is a testament to that. Intel managing to concede performance crown, and be absolutely slaughtered by 10x smaller competitor, who they themselves were beating soundly just few years back, tells you that, us, keyboard warriors have a point and a reason to doubt the company.

This is crazy. You're just looking at it from your point of view. No-one is assuming they know better and anyway they are still making tons of cash, so who's right and who's wrong? Not sure about the failed products either?
You seem to be preoccupied with performance metrics of the stuff you're interested in, when Intel's only motivation in this particular arena is to make money out of selling silicon, same as AMD.

IMO Intel are just guilty of sitting on their laurels, but that's what all companies do when they get to the point where they become too big and comfortable. They become a cash cow, money is floating around internally and shareholders get used to solid returns, while the actual investment in engineering is eroded. It's like when accountants took over Mercedes and boost short term profit while churning out cars with dodgy electronics and eroding the value of the brand.
Putting all the emotive language to one side, you're right, that we all have reason to doubt them, but I'm sure I'm not the only one who's worked in a corporate environment and seen how the mentality and business drivers become prevalent and diminish the long term value of a company. It's got nothing to do with morons, it's to do with risk versus reward, at a personal and a company level.

Some people don't obsess over future proofing (which is a false construct) and getting x years of life out of their product. They also don't care about what's coming up next year. They will just buy whatever does the job, get the utility value from that product and move on with their life. That's why Intel (and AMD) will always sell cpus and motherboards, because there is a constant level of new demand.

I don't care either way, it could be AMD, Intel, TI, Cyrix, Motorolla, Zilog, Generic Chinese Company etc. etc. It's all just history repeating itself.

 

[LePhuronn]

Let's first clarify what Alder Lake actually is. It is a hybrid, 3D-stacked die with the name Foveros

No it's not, at least nothing has been confirmed if Alder Lake uses Foveros stacking. Besides, Foveros is just the 3D stacking tech, which that Wikichips page clearly says. And Alder Lake is not 14nm, it's 10nm. Even Lakefield, the first and only Foveros product, was 10nm.

 

[4K8KW10]

No:

Alder Lake
10nm Big And Small Cores With 14nm GPU
https://wccftech.com/intel-alder-la...eaked-10nm-big-and-small-cores-with-14nm-gpu/

Lakefield:
22FFL + 10nm
https://en.wikichip.org/wiki/intel/foveros

 

[LePhuronn]

It literally says 10nm Big and Small cores in that title! Who gives a toss about the GPU?!?!? Even the Wikichip stub says 10nm
https://en.wikichip.org/wiki/intel/microarchitectures/alder_lake

And the Foveros substrate on Lakefield is 22nm but the actual cores are 10nm!

Do you not read or (once again) arguing the toss purely for the sake of it?

 

[g67575]

I don't think people will be able to buy an alder lake / ddr5 system until January/Feb 2022, but it will be a worthwhile upgrade.

I'll be advising my dad to go with AM5 however, as I think the 5nm EUV process will provide better performance and AM5 will have a longer upgrade lifespan.

Also, I'm happier supporting the competition this time around (if prices are reasonable), I feel they deserve it for catching up with Intel.

 

[pc-guy]

I for one look forward to the big little design tbh. I am not sure if their top range sku should sport this design. 16core part should be full fat not some trimmed down skinny latte version.

but the lower SKU especially 6c 8c 10c ones definitely have some low powered ones. I am happy if the CPU can save me a few pennies while I am browsing and watching YouTube or typing up reports. No one can be gaming all the time on their PC while it is on right.

but for people who are after 16c CPU they would want all those cores to be doing stuff at the same time for a period of time. I feel having 8big 8little is too much of a compromise. 12big 4little or 14big 2little would be a better solution.

intel will loose market cap at the top end where they just can’t compete with AMDs ever March towards more cores per package.

 

[Grim5]

I got one look forward to the big little design tbh. I am not sure if their top range sku should sport this design. 16core part should be full fat not some trimmed down skinny latte version.

but the lower SKU especially 6c 8c 10c ones definitely have some low powered ones. I am happy if the CPU can save me a few pennies while I am browsing and watching YouTube or typing up reports. No one can be gaming all the time on their PC while it is on right.

Current top of the range models from amd and intel sit at around 40-50w while you web browse, YouTube or even just idle. I assume big.little will bring that down to like 10-15w, but do you care?