4 ms·
Wouldn’t a cluster of M4 minis cost less and provide more VRAM? There are posts about people getting decent performance for a lot less than 12k USD.
by NKosmatos 2y ago
Wouldn’t a cluster of M4 minis cost less and provide more VRAM? There are posts about people getting decent performance for a lot less than 12k USD.
- whimsicalism 2y agono, these chips are optimized for inference not training & frankly cuda is still table stakes. HN loves it some Apple
- lostmsu 2y agoThey are not optimized for inference vs RTX GPUs.
- sabareesh 2y agoYou get more vram but not enough cores
- angoragoats 2y agoIf you are willing and able to put together the type of system described in the OP (a workstation-class PC, with multiple discrete GPUs and often multiple power supplies), a Mac never makes sense. There are hardware options available at essentially every price point that beat (in some cases drastically) the performance and memory capacity of a Mac. And I say this at the risk of being called pedantic, but a cluster of Mac minis would have zero VRAM.
- lostmsu 2y agoIf you want to wait for over a year to get your model trained (vs 7 days).