3 ms·

This whole discussion seems to leave out a point that I've been thinking about... "New features" isn't the ONLY reason to keep making incremental improvements

This whole discussion seems to leave out a point that I've been thinking about... "New features" isn't the ONLY reason to keep making incremental improvements to software. We also need to make adjustments as hardware paradigms shift over time. Hardware improvements are meaningful, and we often SHOULD upgrade software to take advantage of that. My question is, will hardware ever stop improving? Will we reach a point where an M1-like full-desktop SOC includes enough processing, storage and graphics power that it satisfies any user's desktop/laptop needs? If/when that happens, can software stop evolving in parallel? And a related thought, when we finally start getting significant amounts of people off the planet (space travel), will all networking software need to undergo a major paradigm shift to manage the extremely-high-latency-low-bitrate situation as a normal case?

bocklund 6y ago
> Will we reach a point where an M1-like full-desktop SOC includes enough processing, storage and graphics power that it satisfies any user's desktop/laptop needs? I doubt it. It used to be that things requiring a cluster of computers 20 years ago now runs on desktops while clusters and HPC systems run even bigger calculations. With more power, memory, and storage, things will trickle down and new approaches will take their place. In my research, it would be great if I could use an algorithm scaling at N^5 which gives high accuracy, but it’s too expensive for most N, so pretty much everyone uses an N^3 scaling approach which is less accurate.
ReptileMan 6y ago
Every software expands until it utilizes all available resources