11–20 of 37 posts
Earlier quoted context omitted.
Because bandwidth between those sticks and the GPU isn't fast enough for fast inference.
Even with an integrated gpu? Surely it'll be faster that using thunderbolt cables
Apple is not doing anything magical with hardware that other companies, including Nvidia, cannot do for roughly the same price/performance. The M series is amazing, but there are comparable products out there now. If you want to say eventually demand for Nvidia will fade because demand for training frontier models will decrease with time, fair, but that doesn’t suddenly make Apple the king of AI.
Also, for new better AI models to happen someone will need to train those new better models and that isn’t going to happen on 5000 Mac minis duct taped together.
Earlier quoted context omitted.
Even with an integrated gpu? Surely it'll be faster that using thunderbolt cables
both AMD and Apple hit the unified memory goal. but its the bandwith thats bottlenecking their performance and onboarc memory is even worse.
but of course author chose to use a.i to write this.
& bury very important points which are 1. Chinese Memory Manufacturers 2. Apple Unified Memory + their own chips which are a result of their 'car' effort.
Earlier quoted context omitted.
both AMD and Apple hit the unified memory goal. but its the bandwith thats bottlenecking their performance and onboarc memory is even worse.
Is Apple in a better position than Amd as this article seems to claim?
Earlier quoted context omitted.
both AMD and Apple hit the unified memory goal. but its the bandwith thats bottlenecking their performance and onboarc memory is even worse.
Is Apple in a better position than Amd as this article seems to claim?
Perhaps AI serves also as an attempt to reverse course on this: raise prices on general-purpose computing hardware (excluding phones, which are increasingly locked down) so that it becomes again inaccessible to consumers.
Earlier quoted context omitted.
Is Apple in a better position than Amd as this article seems to claim?
AMD's Strix Halo has a 256-bit memory bus. Apple's Max class chips have a 512-bit memory bus, Ultra class chips have a 1024-bit memory bus. Unless you want to go for datacenter hardware, AMD's offerings simply aren't in the same league in terms of memory bandwidth.