First Impressions: Lenovo T14s with Qualcomm Snapdragon ARM64 CPU

Original Article ↗ Hacker News Discussion ↗

Early tests of Lenovo’s new T14s Gen 6 with Qualcomm’s Snapdragon X Elite show strong performance and battery life potential, but also highlight how immature ARM laptop support still is for Linux and BSD: core hardware like keyboards, touchpads, USB‑C/Thunderbolt, audio, and power management often need custom drivers or bleeding‑edge kernels. Commenters compare this to Apple’s far more polished ARM Macs, noting that many desktop apps and packaging formats still prioritize x86_64, leaving ARM users with fewer stable options. Some see Qualcomm’s active upstream work and improving OpenBSD/Ubuntu support as a promising path toward a viable ARM Linux laptop, while others prefer to run Linux in a VM on Macs until feature parity with x86 hardware is closer.

FreeBSD / BSD support on T14s Snapdragon

  • Initial FreeBSD experience: keyboard, trackpad, Thunderbolt/USB‑C, and thermal/frequency management don’t work; USB‑A does. Device is considered attractive mainly as a “driver‑writing playground,” not yet a practical laptop.
  • OpenBSD reportedly has more working, including the keyboard, and there are public patches for Snapdragon X Elite laptops.

Linux support and kernel status

  • Ubuntu provides an experimental image for this laptop: everything works except audio and brightness controls (brightness stuck on high). Battery life is already >12 hours; Windows gets even more.
  • For NixOS and other distros, users expect kernel 6.12+ (with vendor‑supplied patches) to be the tipping point for solid support.
  • Linaro and Qualcomm are said to be actively contributing Linux patches; some contrast this positively with Apple’s hands‑off approach.

Performance and battery life

  • One FreeBSD cross‑build comparison: T14s (Snapdragon) 3210s vs M3 Max MacBook ~791s; price differential ($1000 vs ~$3000) is seen as aligning with performance per dollar.
  • Users report T14s battery life much better than any x86 ThinkPad they’ve owned.
  • Broader debate: ARM’s “promise” (perf‑per‑watt vs raw performance) vs real‑world examples like Apple M‑series and AWS Graviton; synthetic benchmarks (e.g., Geekbench) are viewed skeptically.

Ports, peripherals, and ergonomics

  • Heavy discussion around USB‑A vs USB‑C: many keyboards, mice, and USB sticks remain USB‑A; USB‑C is costlier, more complex, and often over‑spec’d for simple peripherals.
  • HDMI is widely considered inferior to USB‑C, but TVs can be usable as monitors if “game mode”/post‑processing is disabled.
  • Thunderbolt support status on this Snapdragon machine is unclear beyond “not working on FreeBSD yet.”

ARM Linux desktop viability

  • Some report ARM Linux (e.g., Asahi on Apple Silicon) as a second‑class experience: many AppImages/Snaps/Flatpaks are x86‑only, ARM builds lag and have more bugs.
  • Others describe stable, daily‑driver ARM systems (e.g., NixOS via Asahi) with long battery life and only a few missing hardware features (e.g., internal mic, Thunderbolt).
  • Concern that ARM’s fragmented hardware ecosystem and lack of standardization make support much more work than x86.

Laptops, pricing, and hardware quality

  • Several users consider running Linux in a VM on Apple Silicon (Parallels, NixOS/Asahi) preferable to today’s ARM Linux laptops, accepting some VM overhead for better hardware and battery life.
  • Lenovo pricing is criticized in Europe vs the US; business and reseller discounts can change the picture substantially.
  • Soldered RAM on Snapdragon laptops is disliked, but some argue maxing RAM at purchase is now affordable; Lenovo’s ARM RAM upgrade pricing is contrasted favorably with Apple’s.
  • Opinions on modern ThinkPad quality are split: some report thermal, input, and reliability issues; others are happy with recent T/P/X1 models. Framework and Asus are mentioned as alternatives, especially for users who value repairability or better Linux support.

NPU / AI workloads

  • Expectations that the Snapdragon NPU would accelerate local LLMs are not yet met in mainstream tools like llama.cpp/LM Studio, which currently use only CPU on Snapdragon but Metal/GPU on Apple, making M1/M‑series much faster.
  • Experimental work shows small models (3B–8B) running on Snapdragon’s NPU/GPU, but this is not yet production‑ready.