Skip to main content

Report: Intel Alder Lake-S CPUs to Drop Into New, Taller Socket

We edited this photo to make the CPU look taller.

We edited this photo to make the CPU look taller. (Image credit: Tom's Hardware/Intel)

We've hardly had a moment to catch our breath since Intel formally announced the Comet Lake-S desktop CPUs, but new rumors are already streaming in about future Intel products. The latest claims Alder Lake-S chips will drop into a new socket known as LGA1700. 

The information comes courtesy of the Taiwanese lit-tech, where the company listed a product entry for an LGA1700 interposer and noted that it is for the Alder Lake-S platform. 

Intel's 9th Generation of Core processors were formerly called Coffee Lake, and the new 10th Gen chips announced last week were known as Comet Lake-S. Following Comet Lake-S, Intel Rocket Lake-S CPUs are expected to arrive and drop into the same LGA1200 socket that Intel just released for Comet Lake-S. But now we're hearing that with Rocket Lake-S' successor, Alder Lake-S, will debut with yet another new socket. 

However, Alder Lake-S featuring a new socket wouldn't be that surprising. Alder Lake-S chips are expected to feature a whole new design, with rumors pointing to a big.Little architecture. This is a hybrid architecture featuring small, efficient Gracemont cores paired with an equal number of big, punchy and power-hungry Golden Cove cores. In a way, the design is similar to Intel's Lakefield platform, which will be the first to feature Intel's Foveros 3D stacking technology on a CPU.

Because of its increased pin count, the LGA1700 socket is also expected to grow in size, as was tweeted by hardware leaker @Komachi_ensaka in January. Whereas the LGA1151 and LGA1200 sockets measure 37.5 x 37.5mm, LGA1700 is expected to maintain the same width but grow slightly taller to 45mm. Unfortunately, this radically new socket design  would also mean that existing CPU coolers likely won't be compatible without tweaks. 

We previously heard rumors that Alder Lake-S would debut this year, but with Rocket Lake-S still needing to land first, we're not positive on a release date. Plus, at this time Intel hasn't confirmed either upcoming CPU generations, so take information on the rumored chips with a pinch of salt.

  • bit_user
    You may be replacing your motherboard again in two generations.
    This should surprise exactly no one.

    In spite of all the recent turmoil in their product stack, one thing Intel has been incredibly consistent about is changing their socket every two generations of desktop CPUs (i.e. with generation defined by the "thousands" digits of their numbering scheme).

    Alder Lake-S chips are expected to feature a whole new design, with rumors pointing to a big.Little architecture.
    This makes no sense, for desktop CPUs (much less workstation or server, for that matter). I'll believe it when I see it.
    Reply
  • Deicidium369
    Note even a little bit surprised about the new socket - I never upgrade a system I build - I hold it for no more than 2 years and build a whole new system - new case, new power supply, mother board CPU, RAM, GPU and Windows license. So I don't really care about what socket it is.

    Not sure about the big little thing - i cannot see a single use case for that config - and how would Windows even handle that? And what application would benefit from that?
    Reply
  • spongiemaster
    The even more bizarre part is the 6 core version that has no little cores. So, the little cores can't be there to enable a low power mode. It could be that Golden Cove cores are massive, and better suited for 7nm. Intel would want to avoid getting stuck holding back architectures again if something goes wrong with 7nm so this is sort of a stop gap by releasing an 8 core version on 10nm++(+++++?).

    I would expect 8 cores to still be the sweet spot for mainstream software in 2021 and 22. So Intel would have 8 massive cores to dominate the mainstream software benchmarks and then tack on 8 smaller cores to improve highly multi threaded workloads and combat AMD's 16 core CPU's. I can't see AMD releasing anything higher than 16 cores on the mainstream platforms in the next couple of years, it just wouldn't make any sense from a software perspective.

    Intel’s next generation of microarchitecture will be ‘significantly bigger’ than Sunny Cove
    Reply
  • watzupken
    I am actually unsure if this will land in the retail desktop space. Now, we see big little configs on phones with the intent of reducing power under light load. On a desktop space, the power requirement is not such a big problem (since their current 9th and 10th gen processors are consuming a lot of power), so less reason for people to want to have this sort of big little CPU configuration. On the laptop space, perhaps this may be a way to squeeze more battery life, but at the expense of die space that could have been used to increase the high performance core count.
    Reply
  • bit_user
    Deicidium369 said:
    Not sure about the big little thing - i cannot see a single use case for that config - and how would Windows even handle that? And what application would benefit from that?
    IMO, it's clearly aimed at the laptop market. Lakefield already stands as an example, and I believe it's even used in a current MS Surface model.

    Thread scheduling between big.little cores isn't exactly trivial, but it's also not rocket science. Windows will need to add support for it, if it's not there already.
    Reply
  • bit_user
    spongiemaster said:
    So Intel would have 8 massive cores to dominate the mainstream software benchmarks and then tack on 8 smaller cores to improve highly multi threaded workloads and combat AMD's 16 core CPU's.
    That's an interesting concept. It lets them call their CPU a 16-core, even though it'd be 8.8 and it'd probably give them the equivalent about about 13 big cores' worth of overall throughput, for the all-important Cinebench-type benchmarks.

    I was just thinking about the little cores in terms of power-savings, not as an area-efficient way to add a little more horsepower.

    spongiemaster said:
    I can't see AMD releasing anything higher than 16 cores on the mainstream platforms in the next couple of years, it just wouldn't make any sense from a software perspective.
    Yeah, I feel like that well is running dry. AMD needs to get more creative about how to compete in the next few iterations.

    spongiemaster said:
    Intel’s next generation of microarchitecture will be ‘significantly bigger’ than Sunny Cove
    Thanks for the link. I do wonder how much of the supposed IPC improvements are actually accounted for by simply widening AVX out to 512, though. Otherwise, these massive gains are seriously stretching my credulity.
    Reply
  • travsb1984
    Deicidium369 said:
    Note even a little bit surprised about the new socket - I never upgrade a system I build - I hold it for no more than 2 years and build a whole new system - new case, new power supply, mother board CPU, RAM, GPU and Windows license. So I don't really care about what socket it is.

    Not sure about the big little thing - i cannot see a single use case for that config - and how would Windows even handle that? And what application would benefit from that?

    Okay fair enough, I've actually never upgraded a CPU without a MB either now that I think about it even though I always liked the idea... Do you sell you whole computer after you buy a new one? It makes no sense to me to replace a power supply, case, or windows(?) unless you just like new things. I'm sporting the same case and power supply I've had for almost 8 years and I've been rockin the same windows retail key for almost 12 years since Microsoft constantly has free upgrade programs. One time OEM keys are a ripoff if you're upgrading every two years.
    Reply
  • bit_user
    travsb1984 said:
    It makes no sense to me to replace a power supply, case, or windows(?) unless you just like new things. I'm sporting the same case and power supply I've had for almost 8 years
    I'm no expert on the subject, but I believe PSUs do wear out.

    I'd probably replace it after total usage hours exceeds the warranty term. Maybe sooner, if frequently subjected to high load, or later if load and temps are kept low.

    The risk isn't simply that the PSU dies, but rather that a bad PSU can damage other parts or result in system instability.
    Reply