Apple M6 Pro Achieves the Highest Single-Core CPU Score in Geekbench 7

Original Article ↗ Hacker News Discussion ↗

An apparent Geekbench 7 result for an unannounced Apple M6 Pro chip shows record-breaking single‑core performance, surpassing current Intel, AMD and Qualcomm CPUs and prompting speculation about Apple’s next Mac hardware. Commenters debate how such results leak despite Apple’s tight secrecy, whether the entry is even genuine, and how much synthetic benchmarks like Geekbench reflect real‑world workloads such as video encoding or gaming. The conversation broadens into trade‑offs between Apple’s performance and ecosystem lock‑in versus the flexibility of Linux and Windows, and whether Apple’s hardware and software stack can justify its closed nature and premium price.

Leak authenticity and source

  • Initial confusion: Apple has not announced an M6 Pro, yet a Geekbench 7 entry appears with that name and record single‑core score.
  • Explanations offered:
    • Internal pre‑release machine where someone ran Geekbench and uploaded, as has allegedly happened with past M‑ and A‑series chips.
    • Skeptical view: Apple’s internal controls are too strict for accidental uploads; if it’s there, Apple likely allowed it (e.g., via marketing).
  • Others counter that leaks happen at Apple like any other company, despite monitoring and policies.
  • A late comment claims the submission is likely forged because the reported L1/L2 caches don’t match known M4/M5 Pro patterns, casting doubt on authenticity.

Chip details and benchmark context

  • The leaked chip is described as an 18‑core “M6 Pro”, compared to a 12‑core base M6.
  • Other Geekbench database entries are cited: M5 Ultra, base M6 CPU/GPU, with the base M6 reportedly averaging just over 4000 in single‑core.
  • Comparisons:
    • AMD Ryzen 9 9950X3D2 ~3160; Intel Core Ultra 7 270K Plus ~2940 single‑core.
    • Apple’s mobile A20 Pro and other Apple SoCs are also near the top of Geekbench single‑core charts.
  • Some users note that M‑series single‑core gains over M1 are less than 2x despite higher clocks.

Reliability of Geekbench and cross‑architecture comparisons

  • Multiple participants question how meaningful Geekbench is across architectures:
    • Real workloads (e.g., AV1 encoding with svt‑av1, Blender renders) sometimes show AMD/x86 machines outperforming Macs despite lower Geekbench scores.
    • Reasons suggested: AVX2/AVX‑512 advantages, heavy SIMD tuning for x86, incomplete NEON optimizations, different scaling of multi‑core tests, and Geekbench workload choices that may favor Apple‑style wide, scalar ILP.
  • Others argue that for “interactive user stuff,” Geekbench correlates reasonably well with perceived snappiness, while being less representative for long‑running, highly parallel or SIMD‑bound tasks.

Real‑world performance issues

  • One developer reports a 2004 Windows game running at ~4 FPS on an expensive MacBook Pro via Wine+Rosetta, while an ARM Linux setup runs it smoothly, suggesting something is badly off in the Mac stack for that case.
  • Responses emphasize profiling, synchronization bottlenecks, 32‑bit under Rosetta, and the complexity of Wine/Rosetta layers rather than raw CPU limits.

Ecosystem, lock‑in, and OS preferences

  • Strong split in attitudes:
    • Pro‑Apple: best overall devices, high performance per watt, no driver drama, smooth integration between devices; many don’t feel “locked in.”
    • Anti‑Apple: closed ecosystem, difficulty running alternative OSes, proprietary stacks (e.g., media, GPUs), and limited user control; some prefer Linux for “peace of mind” and control despite weaker polish.
  • There is debate over what counts as “lock‑in” (e.g., export of Notes, app distribution, patching vendor bugs, running alternative OSes easily).

Gaming, GPUs, and future directions

  • Some believe AAA gaming on macOS could be transformative; others note that while some big titles exist, support is thin and late, and Mac GPUs are weak for high‑end gaming and general‑purpose compute compared to discrete Nvidia/AMD GPUs.
  • Apple’s hardware media encoders are praised for speed in video workflows, but several commenters insist hardware encoders intrinsically trail top‑quality software encoders in efficiency and flexibility.