The post points to benchmarks on a Dell XPS 13 using Intel’s Core 5 320, a Wildcat Lake chip built around a small mix of performance and efficiency cores. In Jeff Geerling’s tests, it did unusually well on performance per watt in HPL, a double-precision floating point benchmark better known from supercomputing, and looked competitive with some Apple systems on idle draw and sustained multicore work. That got attention because Intel laptops have spent years chasing clocks and losing badly on battery life and heat.
The strongest read is that Intel has crossed an important threshold, but only in a narrow sense. People were convinced the chip is a real step forward because it is not just another hot x86 part brute-forcing benchmarks. They were not convinced it means Intel has broadly beaten
ARM or Apple. The benchmark at the center of the claim is heavily shaped by
FP64 behavior,
SIMD, libraries, and configuration choices. Several comments pointed out that this says more about one class of numerical workload than about web, office, developer, or general laptop use. Geerling’s own follow-up comments reinforced that the broader picture is mixed. The Intel part can hold up better in long multicore runs thanks to active cooling, while Apple’s A18-based MacBook Neo still wins comfortably in
GPU tests and remains ahead in single-core speed.
The discussion also pulled the focus away from
ISA tribalism and toward system design. Apple’s efficiency lead was framed less as a magic property of ARM and more as years of coordinated work across silicon, power management, OS behavior, displays, and thermal limits. A few experienced commenters argued Intel has actually been capable of competitive task-level efficiency for a while if clocks and boost behavior are tuned sanely, and that OEMs often ruin the result by shipping aggressive power profiles for tiny headline gains. That makes this Dell result feel less like a miracle and more like Intel and partners finally choosing not to waste power.
So the thread landed in a pragmatic place. Intel’s low-power x86 parts look meaningfully better than many people assumed, especially for sustained CPU work and Linux-friendly setups. But the headline overreaches. This is a promising data point about one carefully tuned laptop and one benchmark family, not a clean declaration that Intel has beaten ARM on laptop efficiency.
hackaday.com