The absolute most important thing is the ability for end users to replace their batteries and displays. Storage expansion is somewhat moot by now thanks to cloud and NAS storage options coupled with 5G speeds.
Personally, I think there is absolutely no reason why in something like the iMacs, HD and RAM shouldn’t be user replaceable and upgradable.
They always used to be until Jonny I’ve got his thinness bug.
If Apple went for it and introduced a new aesthetic where there were small visible screws which became a symbol for cate about the environment, they could probably push the industry in that direction
Personally, I think there is absolutely no reason why in something like the iMacs, HD and RAM shouldn’t be user replaceable and upgradable
Simple: user-replaceable RAM is too slow. Apples M-series SoCs combine the CPU and GPU and both share the same memory. This has massive performance advantages, especially for GPU-compute tasks. Performance of GPU code is very dependent on memory bandwidth. You cannot have high-bandwidth memory on a user-replaceable module, you have to have the memory chips physically close to the processor. This is the reason there are no user-replaceable RAM modules on GPUs either.
With GPU compute becoming more and more important, I expect the PC world to get rid of user replaceable RAM and GPUs as well in the future.
The absolute most important thing is the ability for end users to replace their batteries and displays. Storage expansion is somewhat moot by now thanks to cloud and NAS storage options coupled with 5G speeds.
Personally, I think there is absolutely no reason why in something like the iMacs, HD and RAM shouldn’t be user replaceable and upgradable.
They always used to be until Jonny I’ve got his thinness bug.
If Apple went for it and introduced a new aesthetic where there were small visible screws which became a symbol for cate about the environment, they could probably push the industry in that direction
Simple: user-replaceable RAM is too slow. Apples M-series SoCs combine the CPU and GPU and both share the same memory. This has massive performance advantages, especially for GPU-compute tasks. Performance of GPU code is very dependent on memory bandwidth. You cannot have high-bandwidth memory on a user-replaceable module, you have to have the memory chips physically close to the processor. This is the reason there are no user-replaceable RAM modules on GPUs either.
With GPU compute becoming more and more important, I expect the PC world to get rid of user replaceable RAM and GPUs as well in the future.