More CPU cores are always welcome in my system, but it truly is a shame to see so much of that processing power fall by the wayside in single-threaded workloads. These cries of woe from enthusiasts haven’t fallen on deaf ears, as Intel hopes to save the day through the introduction of ‘Super Cores’.
More specifically described as ‘Software Defined Supercores’ in a recent patent, this isn’t another addition to the Intel’s proverbial core library, á la Performance and Efficient. Instead, the company is exploring ways to virtually fuse two normal cores into a singular, more-powerful entity.
This essentially amounts to reverse hyper-threading. Intel would split a single-threaded application’s calls into separate blocks, divvying them up in parallel between two cores. As far as the program and operating system are aware, this is just one core.
This should improve single-thread performance, while also benefitting processor efficiency. With Super Cores in tow, there’s theoretically less need to push raw clock speeds and power ever-further.
This is a genuinely exciting concept in my eyes, but such potential doesn’t always translate perfectly off the page. My immediate concern is reliability, as much hinges on Intel’s ability to construct a bulletproof soft-marriage between cores with some input presumably necessary from Microsoft too.
Let’s not forget the problems Task Scheduler conjured following the launch of 12th Gen Intel Core processors. It’s easy to see how similar instances of incompatibility and more could arise from the introduction of Super Cores.
It’s unclear how quickly Intel could (or wants to) implement Super Cores. Since software defines this process, it’s entirely possible that we could see them pop up on existing architectures. It’s unlikely, but I’d love to see Arrow Lake thrown a bone.
For further analysis and more on everything processors, you’ll find plenty to read on the Club386 Google News feed.
