Thiago Trevisan for iOS Central:
The iPad Pro hardware can certainly have you racing down the highway of productivity, but then you hit a bottleneck: iPadOS. iPadOS just isn’t designed for daily production on a larger scale, but Apple has carefully curated a suite of apps that are designed specifically with iPad Pro users in mind to make it a valuable tool in smaller instances.
I am biased towards favoring the iPad for everything. However, I think the iPad is at a critical disadvantage for current and future computing, namely because of it’s limited hardware performance specs when compared to Macs for what LLMs need to run efficiently (and quickly).
The fastest M4 iPad Pro you can buy is:
- 10-core CPU with 4 performance cores and 6 efficiency cores
- 10-core GPU
- Hardware-accelerated ray tracing
- 16-core Neural Engine
- 120GB/s memory bandwidth
- 16GB RAM
MacBook Pro M3:
- 16-core CPU with 12 performance cores and 4 efficiency cores
- 40-core GPU
- Hardware-accelerated ray tracing
- 16-core Neural Engine
- 400GB/s memory bandwidth
- Up to 128GB RAM
Yes, the prices are not even remotely on the same planet, but the current top-tier open source LLMs are hungry for GPU processing and RAM. Two areas where the iPad Pro just is not a beast on the specs. (There are companies, likely also Apple, working to offload more of these models to the CPU, but as of right now this is a GPU game.)
I get all the past complaints and why iPad or Mac — but right now, I would be buying based on what can run local LLMs on the fastest. Which is decidedly not the iPad Pro.