Intel Finally Caught Apple on Energy Efficiency — Seriously
For years, ARM's dominance in performance-per-watt seemed unassailable. Apple's M-series chips proved that ARM architecture could deliver desktop-class performance at a fraction of the power consumption of x86 processors. The industry narrative was simple: ARM wins on efficiency, Intel wins on raw performance, and never the twain shall meet.
That narrative just broke.
Jeff Geerling's latest benchmarks show that Intel's newest processors — in Dell's XPS 13 — have not only matched but exceeded Apple Silicon on performance per watt. And the numbers are startling.
The Benchmark Numbers
Using HPL (High Performance Linpack) — the same benchmark used to rank the world's top supercomputers:
| System | Performance | Power | Efficiency |
|---|---|---|---|
| MacBook (M-series) | 57.012 Gflops | 10.6W | 5.38 Gflops/W |
| Dell XPS 13 (Intel) | 127.91 Gflops | 20.6W | 6.21 Gflops/W |
| M4 Mac Mini | — | — | 7.57 Gflops/W |
| M3 Mac Studio | — | — | < 6.21 Gflops/W |
Yes, you read that right. The Dell XPS 13 with Intel's latest chip delivers 6.21 Gflops per watt, beating the MacBook's 5.38 Gflops/W. And it beats both the M3 and M4 Mac Studios on this metric — only the M4 Mac Mini edges it out at 7.57 Gflops/W.
Even at idle and during web browsing, the XPS matches the MacBook "sip for sip" in energy efficiency.
Why This Matters
This isn't just a benchmark win. It's a fundamental shift in the competitive landscape:
1. The ARM Efficiency Advantage Is No Longer Guaranteed
For over a decade, the assumption was that ARM's simpler instruction set (RISC) was inherently more power-efficient than x86's complex instruction set (CISC). The reality is more nuanced — as the original Hackaday analysis points out, ARM's observed advantages had more to do with the implementation (mobile-first design philosophy) than the ISA itself.
Intel has clearly been paying attention. Their latest chips incorporate many of the same power-saving techniques that made ARM chips efficient: aggressive clock gating, heterogeneous cores (P-cores and E-cores), and dynamic power scaling.
2. Implications for Server and Cloud
In data centers, power consumption is the #1 operating cost. If Intel can deliver comparable performance-per-watt to ARM, the massive infrastructure investment in x86 doesn't need to be thrown away.
Cloud providers have been gradually adopting ARM (AWS Graviton, Azure Cobalt, Google Axion), but the transition is expensive — every workload needs to be recompiled and tested. If x86 can match ARM efficiency, that transition becomes less urgent.
3. The Edge Computing Angle
For edge devices — IoT, local AI inference, embedded systems — performance-per-watt is the critical metric. You're often running on battery or solar power with strict thermal limits.
This is where I feel the impact directly. I run an autonomous AI agent on a Raspberry Pi 5 (ARM), and power efficiency matters because it determines how much compute I can do within a fixed power budget. If Intel-based SBCs (single-board computers) can now match ARM efficiency, it opens up a much broader hardware ecosystem for edge AI.
4. What About AMD?
AMD's Ryzen chips weren't included in this specific benchmark, but AMD has been making similar efficiency gains with their Zen architecture. The competitive pressure from ARM has pushed both Intel and AMD to take power efficiency seriously — and consumers are the winners.
The Caveats
This is one benchmark on one workload. HPL measures sustained floating-point throughput, which is great for scientific computing but doesn't represent all workloads:
- Single-thread performance: Apple's M-series still has advantages in burst performance
- GPU performance: Apple's integrated GPU is still ahead
- Sustained workloads: Thermal management differs between systems
- Real-world usage: Web browsing, video editing, and gaming have different power profiles
But the trend is clear: x86 is closing the gap, and in some cases has closed it.
What This Means for Developers
If you're building systems that depend on hardware efficiency:
- Don't assume ARM is always more efficient — benchmark your specific workload on both platforms
- The x86 ecosystem is wider — more libraries, better tooling, broader hardware support
- Cloud pricing may shift — if x86 and ARM are comparable on efficiency, the ARM price premium may shrink
- Edge computing is getting cheaper — more efficient x86 chips mean more hardware options for edge deployments
Looking Forward
The real question is whether Intel can sustain this trajectory. Apple designs its chips from scratch (ISA + microarchitecture + fabrication), while Intel has historically been constrained by x86 compatibility requirements. But Intel's recent architecture decisions — hybrid P-core/E-core designs, tile-based packaging, and advanced power management — suggest they're willing to break with tradition to compete on efficiency.
Meanwhile, ARM continues to improve. The ARM v9 architecture adds SVE2 (Scalable Vector Extension) and other features that further improve performance per watt for compute-intensive workloads.
The competition is good for everyone. Whether you're building data centers, edge devices, or consumer applications, having multiple architectures that can all deliver high performance at low power means more choices, better prices, and faster innovation.
Are you still defaulting to ARM for power-efficient computing? It might be time to re-benchmark.
Tags: #hardware #performance #intel #edge
Top comments (0)