Intel Alder Lake-S leak hints at a big.LITTLE future for Desktop CPUs
big.LITTLE on desktop could be a disaster for Intel
Published: 9th March 2020 | Source: PTT via Ghost Motley |
Intel Alder Lake-S leak hints at big.LITTLE future for Desktop CPUs
Now, it looks like Intel is planning to bring this same approach to desktop systems, with new Alder Lake leaks suggesting that Intel plans to release a 16-core processor with eight big cores and eight little cores.
While this information comes via a Chinese forum (via Ghost Motley), we already know that Intel is working on a big.LITTLE architecture for its future processors. Intel's Lakefield processor combines a single Sunny Cove CPU core with four power-efficient Tremont (Atom-series) cores to deliver high levels of single-threaded performance and incredible levels of power efficiency. Moving the concept behind Lakefield forward leads to a simple question, would it work on desktop platforms.
Alder Lake is rumoured to be a successor to Rocket Lake, which itself is a successor to Intel's soon-to-be-released Comet Lake series of desktop processors. After Comet Lake, Intel's high-end product offerings are said to regress from a 10-core high-end SKU back to an 8-core SKU with Rocket Lake. With Rocket Lake, Intel will reportedly make up for the core count difference with increased single-threaded performance, with Alder Lake innovating further with a 16-core big.LITTLE CPU design.
Alder Lake will likely release in 2022, and before then, Intel will need to mitigate the shortcomings of a big.LITTLE design. These issues include the need to update OS' to support big.LITTLE processor designs and the need for software makers to create their applications with such an architecture in mind. If Intel doesn't have stable application or OS-level support, big.LITTLE would be a liability for Intel, not a benefit.
If these rumours of a big.LITTLE desktop processor are correct, Intel will need to get Microsoft and other software giants to re-architect their OS' and software with x86 big.LITTLE processors in mind. This level of support is not guaranteed, and could spell disaster for Intel in the short term.
Software will need to be able to tell what features every core on a processor supports and if their workloads should be given to big or little cores for processing. Will these little cores support AVX instructions properly, will the big cores have core-specific features that applications need to be coded to exploit? These are complexities that the x86 software market hasn't needed to consider before, making it a minefield for Intel to navigate.
While the power consumption benefits of a big.LITTLE style designs are great for mobile devices, desktop systems are a lot less power restricted, making the primary benefit of the architecture practically irrelevant. If power isn't an issue, having 16 big cores will be preferable to having 8 big cores and 8 little ones.
While big.LITTLE could prove to be a major innovation for the x86 CPU market; it could also prove to be a disaster for Intel in the short term. Intel doesn't control the software scene, and it will take a lot of convincing to get companies to explicitly support such a major architectural change.
You can join the discussion on Intel's rumoured big.LITTLE Alder Lake processors on the OC3D Forums.