Apple introduces M6 and M5 Ultra for a big leap in performance and AI compute
Apple unveils its M6 and M5 Ultra chips, touting significant performance gains with its first 2nm chip and a quad-die architecture, pushing the boundaries for AI compute. Hacker News is buzzing over the potential for running massive local LLMs and the eye-watering price tags for maxed-out configurations. The perennial debate rages: is this hardware a game-changer for on-device AI, or just another testament to Apple's premium pricing for specialized niches?
The Lowdown
Apple has announced its latest custom silicon, the M6 and M5 Ultra chips, promising a substantial leap in performance and AI capabilities. The M6, fabricated on a cutting-edge 2-nanometer process, features a 12-core CPU, a 12-core GPU with Neural Accelerators, and a Dual 16-core Neural Engine, designed to enhance everyday tasks, coding, and on-device AI workflows. The M5 Ultra is Apple's most powerful chip to date, utilizing a quad-die architecture through UltraFusion technology, boasting up to a 36-core CPU, an 80-core GPU with Neural Accelerators, and a massive 32-core Neural Engine, all backed by up to 512GB of unified memory with a staggering 1.2TB/s bandwidth. This powerhouse is aimed at professionals tackling demanding tasks like 3D rendering, scientific analysis, and large-scale, compute-intensive AI models.
- M6: Everyday Powerhouse: Positioned for a broad range of users, from students to developers, the M6 emphasizes a balance of performance, power efficiency, and on-device AI. Its 2nm process allows for greater transistor density, and the Dual 16-core Neural Engine significantly accelerates AI workflows, enabling faster model execution for applications.
- M5 Ultra: Professional AI Engine: Tailored for the most demanding professional and AI workloads, the M5 Ultra's quad-die design and immense unified memory capacity are engineered to handle complex simulations, high-resolution video editing, and frontier AI models directly on the device. Its next-generation GPU with Neural Accelerators and enhanced Media Engine provide unparalleled processing power for AI and graphics-intensive applications.
- AI-Centric Development: Both chips are deeply integrated with Apple's developer frameworks, including Core AI, Core ML, Metal, and Xcode, allowing developers to optimize and fine-tune large AI models locally. This on-device capability aims to offer secure, private, and energy-efficient AI operations.
These new chips underscore Apple's commitment to advancing its silicon leadership, particularly in the burgeoning field of on-device artificial intelligence, by delivering specialized hardware designed to meet both mainstream and high-end computational demands.
The Gossip
Premium Price Point Ponderings
Many commenters balk at the astronomical prices for maxed-out M5 Ultra configurations, comparing them to down payments on homes or even the historical cost of SGI workstations. There's a debate on whether such an investment is justified for individuals or small businesses, especially when considering the alternative of cloud-based AI services, though some acknowledge the value for privacy-sensitive professional workflows and for businesses aiming to cut cloud AI bills.
Local LLM Lamentations & Longings
A significant portion of the discussion revolves around the chips' suitability for running large language models locally. While the M5 Ultra's 512GB unified memory is lauded for its potential, the M6's 32GB is deemed insufficient for 'useful' AI. Users express both excitement for the promise of on-device AI and frustration over current limitations, questioning the actual token-per-second performance and debating whether it's more economical to subscribe to cloud AI services or wait for the rumored M7 chip which might offer even better AI capabilities.
Performance Prowess & Platform Parity
Commenters discuss Apple's continuous performance lead, drawing parallels to past eras of CPU competition (Intel vs. AMD) and how Apple consistently 'meep-meeps' past competitors. Some note that while other vendors are catching up to older M-series chips, Apple keeps innovating. The discussion touches on the efficiency advantage of ARM-based Apple silicon over x86, particularly regarding battery life and thermal management, though some claim Intel Windows laptops can now beat M5 in certain metrics.
Ecosystem Ennui & Linux Longing
Despite the powerful hardware, a segment of the HN community voices discontent with Apple's closed software ecosystem and lack of official Linux support. Some users are actively exploring alternatives like Omarchy (a Linux desktop environment) or running Linux in VMs, lamenting the inflexibility of macOS for advanced customization and debugging. There's a strong desire for Apple to open up its hardware to other operating systems.
Gaming's Peculiar Promotional Pick
Several comments express bewilderment and amusement at Apple's choice of 'Mixtape' as an example of a 'demanding game' that showcases the M6's GPU capabilities. Critics suggest the game is neither graphically intensive nor popular, leading to speculation about why Apple would highlight it.