Intel x86 Laptop Edges Out Apple Silicon on Efficiency, Undercutting ARM's ISA Myth

Jeff Geerling ran the HPL Linpack benchmark—the same test used to rank Top500 supercomputers—on a MacBook and a new Dell XPS 13, and the results upend a long-held assumption. The Dell’s Intel Core 5 320 hit 127.91 Gflops while drawing 20.6W, yielding 6.21 Gflops per watt, versus the MacBook’s 57.012 Gflops at 10.6W, or 5.38 Gflops per watt. The x86 machine also stayed competitive at idle and during web browsing, and it outperformed the M3 and M4 Mac Studios on efficiency, losing only to the M4 Mac Mini at 7.57 Gflops/W.

The significance is less about which laptop is better overall—the Mac still wins on integrated graphics and audio—and more about what the numbers imply for architecture. ARM’s reputation for doing more compute per watt is what pulled Apple toward the design and won ARM the mobile market. These benchmarks support the argument that the efficiency gap was never really inherent to the instruction set, but rather a product of specific chip implementations. When Intel invests in an efficient x86 design, it can match or beat Apple Silicon on the metric that supposedly favored ARM.

Geerling also notes that squeezing out efficiency gains doesn’t necessarily require dropping down to assembly or C, suggesting programming language choice matters less than commonly assumed. He has published the full benchmark data on GitHub for both machines, making the claims independently checkable.