First pictutres of Intels Hybrid Design Alder Lake CPUs (LGA1700) emerge with a rectangular package

[Master Disaster]

The first images of Intels new big little hybrid CPUs have emerged thanks to a leaker and its quite a big change. Based on Intels 10nm SuperFin process the new CPUs are expected to launch late in 2021 after the launch of Rocket Lake at the beginning of next year.

Quote

Intel’s first desktop processors based on 10nm SuperFin architecture will officially launch in the second half of 2021. Our sources believe that the launch of Alder Lake will be closer to the fourth quarter, rather than the third. Intel already confirmed that its Rocket Lake series (a predecessor to Alder Lake) launches in the first quarter of 2021, possibly March, meaning Intel is in no rush to launch another series just a few months apart.

Alder Lake is going to be a big change in X86 CPU design adopting a BIG little approach similar to how mobile SOCs are made. The CPU is comprised of BIG cores which are high power but low efficiency and little cores which are low power high efficiency.

Quote

Intel Alder Lake will feature a hybrid technology with high-performance (big) and high-efficiency (small) cores. This is a completely new approach to desktop x86 computing, and we only got a glimpse of how it may work with Lakefield processors (one big and four small cores).

Alder Lake will also come with a new scheduler to allow the CPU to choose between saving power or offering high performance based on workload unlike Lakefield which used a similar design but was focused purely on saving power.

Quote

Back in August Intel confirmed Alder Lake will feature Golden Cove and Gracemont core architectures. Unlike Lakefield, which was focusing on battery life, Alder Lake will focus on performance. The hybrid design will involve a next-generation hardware scheduler, providing seamless operation to applications.

Alder Lake is also expected to support DDR5 memory and maybe even PCIe 5.0.

Quote

Another set of features coming to Alder Lake is DDR5 support and possibly even PCI Express 5.0 (this has not yet been confirmed). The transition to DDR5, PCIe 5.0 and 10nm SuperFin architecture will require a new socket – LGA1700.

Obviously the new design means a radically different package and a new socket, LGA1700. Rumours suggest Intel intends to keep LGA1700 for 3 generations though which could make this a compelling upgrade.

Quote

According to recent rumors, Intel intends to keep the LGA1700 socket for at least three generations. The socket itself has already been a topic of rumors before and it was revealed that the package dimensions for the mainstream desktop series will change to 37.5×45 mm (7.5mm taller than LGA1200). The photo of the CPU that we have appears to confirm those dimensions.

 

The CPU pictured below features exactly 1700 contact pads (we counted). This is actually the first LGA1700 processor codenamed Alder Lake to be photographed. Our source has not yet confirmed the specifications, so might provide an update later. Please note, that is an engineering sample that might look different than retail product.

 

Source - https://videocardz.com/newz/exclusive-intel-alder-lake-s-cpu-pictured

 

Very interesting stuff but you gotta wonder how they plan to maintain current levels of performance by sacrificing half of the cores in the package. Does this mean Intel are willing to give up the clock speed and gaming performance crowns to try and get ahead in efficiency?

[Deli]

Does it mean CPU coolers require a new mount?

[Arika]

2 minutes ago, Deli said:

Does it mean CPU coolers require a new mount?

i would say no given it's the same width as the older sockets. you might just be restricted on the orientation of the cooler. but i've never seen a cooler mounted top to bottom instead of side to side.

[straight_stewie]

15 minutes ago, Master Disaster said:

Very interesting stuff but you gotta wonder how they plan to maintain current levels of performance by sacrificing half of the cores in the package. Does this mean Intel are willing to give up the clock speed and gaming performance crowns to try and get ahead in efficiency?

Big-little can bring performance improvements.

While it's true that it commonly means "purposeful performance hits", that's only because the only devices that use them are battery powered.

There is a problem in processor design. When things get too far apart, you have to slow signalling down. That's because the signals can get "skewed", resulting in timing problems.This is actually a pretty large part of the reason why processors with more cores also generally have lower clocks. 

However, with a big-little architecture, you can design smaller "cores" focused on more specific tasks, but with their own clocks. So the bigger, fully-featured cores, run on slower clock speeds, while the smaller, more specialized cores, can run at higher clock speeds. This could actually bring noticeable performance gains: most Non-OS-call stuff is still just branching and integer math.
 

[Ankh Tech]

Honestly, Intel is too late anyways. AMD just won the game. By the time Intel reaches 10nm AMD would be 5nm

[porina]

25 minutes ago, Master Disaster said:

Very interesting stuff but you gotta wonder how they plan to maintain current levels of performance by sacrificing half of the cores in the package. Does this mean Intel are willing to give up the clock speed and gaming performance crowns to try and get ahead in efficiency?

What core sacrifice? The possible core configurations have been leaked long ago, and the expected offerings are up to 8 cores each of big and little cores. Of course, this wont be as fast as 16 big cores, but to date they haven't offered more than 10 big cores on mainstream desktop. The last HEDT offering was cost per core competitive with Ryzen, and they could continue on that path for users genuinely needing more big cores, where a HEDT platform makes much more sense.

 

If you look at die area of the big vs little cores using our only example in Lakefield for now, 4 small cores are comparable in area to 1 big core. So an expected maximum 8+8 configuration would take approximately the area of 10 big cores.

 

We don't know how Alder Lake will perform, but Tiger Lake has shown us that 4.8 GHz is now possible on 10nm, although exact peak performance when not power limited by mobile platform remains to be seen. Alder Lake is a generation beyond Tiger Lake, which already has major IPC boosts compared to its predecessor Ice Lake, which in turn was improved over Skylake. Even if they don't manage to catch up to Zen 3/4 they're not sitting still. The "cove" series cores could be likened to AMD's "Ryzen". They will improve on them generation over generation.

 

1 minute ago, Ankh tech tips said:

Honestly, Intel is too late anyways. AMD just won the game. By the time Intel reaches 10nm AMD would be 5nm

What should Intel do? Give up, pick up the ball and go home? AMD in 'dozer era were much further behind Intel than Intel are behind AMD today, and they continued fighting and came back. Intel will try to do the same. They may or may not succeed, but doing nothing is not an option.

[Ankh Tech]

I am not saying they'll give up I'm saying by the time they catch up amd would be too good to be true

[LAwLz]

40 minutes ago, Master Disaster said:

Very interesting stuff but you gotta wonder how they plan to maintain current levels of performance by sacrificing half of the cores in the package. Does this mean Intel are willing to give up the clock speed and gaming performance crowns to try and get ahead in efficiency?

Do we even know if that's the case?

It might be that the current 6 core will be replaced with a 6+2 (6 big, 2 little) core config. I seriously doubt Intel will just replace big cores with small ones and call it a day.

 

 

18 minutes ago, Ankh tech tips said:

Honestly, Intel is too late anyways. AMD just won the game. By the time Intel reaches 10nm AMD would be 5nm

Those numbers are fairly meaningless.

What Intel calls "10nm", TSMC calls "7nm".

 

Intel is already producing chips on their 10nm node, and (if it doesn't get delayed) they will ship desktop chips on their 10nm node in Q1 2021.

AMD is ahead of Intel a lot right now, but that's only because AMD just launched their new products and Intel will launch theirs in 3-5 months. I don't think AMD will release the 6000 series of CPUs in ~3-5 months.

[Ankh Tech]

Well rocket lake might be 14nm we still don't know

[Ankh Tech]

And alder lake might even be q4 2020 or even 2021 and I think by then amd would release the amd 6000

[Arika]

3 minutes ago, Ankh tech tips said:

I am not saying they'll give up I'm saying by the time they catch up amd would be too good to be true

But you're basically saying that what ever they do is pointless. same same but different.

 

you're also assuming a lot, unless you have a crystal ball, no one can say what will happen in the future. There is already indications that Intel is going to use TSMC for their next generation of CPUs while they work on their own fabs to get them up to scratch.

 

[Ankh Tech]

ok maybe Im biased since I am an amd fanboy but I honestly think, all bias aside, is that amd won

[Arika]

3 minutes ago, Ankh tech tips said:

ok maybe Im biased since I am an amd fanboy but I honestly think, all bias aside, is that amd won

....no...they are winning. There is no "winner" and nor should you want one,The only winner should be the consumer and that comes from constant competition

[Ankh Tech]

sorry i misphrased I mean what you mean but I have a point

[porina]

16 minutes ago, LAwLz said:

It might be that the current 6 core will be replaced with a 6+2 (6 big, 2 little) core config. I seriously doubt Intel will just replace big cores with small ones and call it a day.

It remains to be seen how these CPUs will be marketed. From a performance standpoint, leading with the big cores, with the small cores as a kinda bonus would be my personal preference. But there is great opportunity for simplification to the total and that could be misleading, intentional or not. If they offer all the previously leaked configurations, it could get very confusing, so I'd hope they'll be more sensible in that.

[yian88]

Does anyone even care anymore? Ive been waiting for better laptop CPU's for over a decade and still nothing. Where are the massive improvents in laptop mobile SoC  CPU+GPU, why is a mid- top end phone or tablet so much faster and feature packed than a laptop? where are the 6-8 core laptops with a decent midrange iGPU for 500$? except Ryzen 5 4000 mobile there is nothing to buy and even such an laptop is 6-700$ without a good GPU, add in an 1650/1660 Ti mobile and you are well over 1000$, thats just unreasonable and unnafordable, i can game/work easily on a midrange tablet or phone for <3-400$ but on any laptop <500$ is just cancer and unusable, i feel like laptops and windows are stuck in 2010 performance era, i cant use a desktop due to space and travel constraints so im forced to use my phone mostly due to shitty laptops, a sluggish i3 or ryzen 3 2-4 core 4gb ram with an useless iGPU, is 4-500$ what the actual  fuck?

[LAwLz]

12 minutes ago, Ankh tech tips said:

Well rocket lake might be 14nm we still don't know

That's very true.

Rocket Lake might be 14nm.

 

If that's the case then we won't get 10nm until Alder Lake, which is expected to be released in late 2021.

That would be a shame (and certainly make Rocket Lake a bad purchase) but last time I heard anything, Zen4 was scheduled for 2022. That would still give Intel roughly half a year where they are competitive on the node with AMD, before AMD jumps one generation ahead again.

 

What I am trying to say is that I don't think Intel will be in any worse situation in the future than they are today, and blindly looking at the nanometer number is not a great idea.

 

 

15 minutes ago, Arika S said:

But you're basically saying that what ever they do is pointless. same same but different.

 

you're also assuming a lot, unless you have a crystal ball, no one can say what will happen in the future. There is already indications that Intel is going to use TSMC for their next generation of CPUs while they work on their own fabs to get them up to scratch.

Yeah, Intel was essentially curb stomping AMD between Sandy Bridge (2011) and Zen1 (2017) and AMD still managed to make a comeback.

Hell, Intel has been able to keep up fairly well with AMD up until fairly recently as well, so AMD have only really been beating Intel hands down for like a year or so. If the situation hasn't changed in 5 years then I will start being worried about Intel. But until then I see it as the typical leapfrogging of technology that sometimes happens in tech. One company is ahead for a couple of years, then the other company is ahead, and then it goes back and forth.

[Ankh Tech]

5 minutes ago, LAwLz said:

That's very true.

Rocket Lake might be 14nm.

 

If that's the case then we won't get 10nm until Alder Lake, which is expected to be released in late 2021.

That would be a shame (and certainly make Rocket Lake a bad purchase) but last time I heard anything, Zen4 was scheduled for 2022. That would still give Intel roughly half a year where they are competitive on the node with AMD, before AMD jumps one generation ahead again.

 

What I am trying to say is that I don't think Intel will be in any worse situation in the future than they are today, and blindly looking at the nanometer number is not a great idea.

 

 

Yeah, Intel was essentially curb stomping AMD between Sandy Bridge (2011) and Zen1 (2017) and AMD still managed to make a comeback.

Hell, Intel has been able to keep up fairly well with AMD up until fairly recently as well, so AMD have only really been beating Intel hands down for like a year or so. If the situation hasn't changed in 5 years then I will start being worried about Intel. But until then I see it as the typical leapfrogging of technology that sometimes happens in tech. One company is ahead for a couple of years, then the other company is ahead, and then it goes back and forth.

yeah nanometer number isn't everything but that isn't the only thing that amd will improve other things too. Another proof that amd might be ahead for a long time is them taking control of the hedt and server markets and the next one might be the consumer market

[Elisis]

9 minutes ago, Ankh tech tips said:

yeah nanometer number isn't everything but that isn't the only thing that amd will improve other things too. Another proof that amd might be ahead for a long time is them taking control of the hedt and server markets and the next one might be the consumer market

Sir, this is a Wendy's.

[TetraSky]

One day, Intel will be on LGA 6900 and the CPU will be as big as a nano ITX board.

 

All jokes aside, really makes you wonder though, where will it stop?

 

 

That said, I'm assuming this extra size is needed to have big and small cores on the same package? If they can implement DDR5 and PCI-e 5.0 into it, right from the "first generation" of these in Q1 2021, it might give AMD a run for their money (coupled with the usual x% IPC improvement)

[Moonzy]

51 minutes ago, Ankh tech tips said:

Well rocket lake might be 14nm we still don't know

and node size matters because...????

i dont give a rat's patoot as long as the performance is good

 

that said, i dont see why desktop chips needs big little design, since efficiency isnt of much concern?

or im missing something here?

[porina]

28 minutes ago, LAwLz said:

Rocket Lake might be 14nm.

That is what current rumours indicate.

 

3 minutes ago, TetraSky said:

That said, I'm assuming this extra size is needed to have big and small cores on the same package?

I'd guess a little differently here. I don't think core counts and sizes are the direct contributor to this. As mentioned earlier, using Lakefield as an indicator, an 8+8 configuration would take ball park same area as 10+0. This is within their current capabilities. What I think might require more space is Intel's own move to chiplets. We still don't have detail on how they're going to do that, other than they are going to move in that direction also. I think a logical split for Intel might be a CPU die and a GPU die. The silicon area might not grow (much) but the connectivity space might.

[Ankh Tech]

2 minutes ago, Moonzy said:

and node size matters because...????

i dont give a rat's patoot as long as the performance is good

 

that said, i dont see why desktop chips needs big little design, since efficiency isnt of much concern?

or im missing something here?

Some people do car about efficiency more than performance, e.g. and node size matters as the smaller the node size the snappier performance

[Ankh Tech]

One thing Intel could do to be potentially ahead is drop the whole igpu that would increase performance.

[Moonzy]

Just now, Ankh tech tips said:

and node size matters as the smaller the node size the snappier performance

...? no

 

1 minute ago, Ankh tech tips said:

Some people do car about efficiency more than performance

true, but it's probably minority of the market

 

i can see it being a potential issue if the heat is not located at the centre of the IHS, like zen 2 chips