What has changed in CPU cores in M3 chips?

Original Article ↗ Hacker News Discussion ↗

Apple’s new M3-based Macs prompt mixed reactions: many users with M1/M2 machines see little practical need to upgrade, while others with heavy workloads like video editing, large Docker stacks, Swift/Xcode builds or AI models report substantial gains and are willing to pay for them. Much of the debate centers on memory and configuration—e.g., moving from 8–16 GB to 32–128 GB unified RAM—as well as corporate refresh cycles, resale value, and tax treatment that make frequent upgrades more rational for some. Alongside praise for performance-per-watt and silence under load, critics highlight high RAM pricing, some software and virtualization gaps on Apple Silicon, and environmental concerns around short hardware replacement cycles.

Upgrade motivations & user segments

  • Many participants keep Intel-era Macs (2012–2015) until hardware failure; they see little need to move from M1/M2 to M3 for typical dev/office work.
  • Others upgrade aggressively (every 1–3 years), especially when:
    • They regret low RAM or small SSD choices.
    • Trade‑in values make the incremental cost small.
    • Corporate refresh cycles (often 3 years, tied to AppleCare and depreciation) trigger replacement.

Workloads that benefit from M3

  • Heavy workloads often cited:
    • Video editing, 4K+ timelines, encoding, and GPU-heavy effects.
    • Music production with many plugins and VSTs.
    • Large Xcode/SwiftUI projects, Rust/C++ builds, Docker-heavy dev setups, k8s, and ML/LLMs.
  • Several report big jumps moving from Intel to any Apple Silicon, and from base M1 to higher-end M2/M3 Pro/Max (often halving build or encode times).
  • Some still find even M1s marginal for complex DAW sessions or multi-container dev.

RAM, storage, and pricing

  • Recurrent theme: 8 GB is widely considered inadequate beyond light use; 16 GB is “barely ok,” many recommend 32–64+ GB for dev/ML/Docker.
  • Apple’s RAM premiums are called “outrageous,” though some argue total tool cost is minor relative to high billable rates.
  • Unified memory equivalence claims (8 GB ≈ 16 GB elsewhere) are generally viewed as marketing, with only a small grain of truth.

Battery life, thermals, and hardware feel

  • Users praise Apple Silicon laptops as quiet, cool, and rigid compared to Intel MBPs; fans rarely spin up even under load.
  • Some anecdotal reports of M3 Pro draining faster than M1 Max; others cite tests suggesting improved battery life. Overall impact is unclear and may depend on browser, codecs, or background tasks.

P-cores vs E-cores and profiling

  • Discussion notes confusion about an article’s “Total Time” graphs; several think it must be summed CPU time, not wall time.
  • Tips shared for viewing per-core activity in Activity Monitor and Instruments to decide whether more P‑cores are warranted.

Software & ecosystem concerns

  • A minority highlight downsides of the ARM transition:
    • Weaker x86 virtualization/emulation.
    • Some specialized tools and workflows no longer working or not yet ported.
  • Others counter that Rosetta and strong Apple Silicon performance have made the shift mostly smooth for mainstream use.