I think nobody has mentioned the (energy) cost of running an algorithm on these machines. For example when you are taking a photo on a phone, it is running a face recognition algorithm in real time. This consumes X watts of power and costs Y USD / second to run.

You'd need to calculate how many multi-million machines you'd need from a specific era to run the algorithm at same speed on those (assuming it was possible), and how many watts of power (CPU power + cooling) it would take.

Algorithms on the topic of linear algebra or databases could be better examples but most people aren't that familiar with those. Video compression is an other good example, you could calculate how many USD it costs to compress 1 hour of 1080p video on a modern laptop (most energy efficient?) and an older system. It may need like 100x the time and 10000x the power, making the whole process cost a million times more even without considering initial investment costs.

I think nobody has mentioned the (energy) cost of running an algorithm on these machines. For example when you are taking a photo on a phone, it is running a face recognition algorithm in real time. This consumes X watts of power and costs Y USD / second to run.

You'd need to calculate how many multi-million USD machines you'd need from a specific era to run the algorithm at same speed on those (assuming it was possible), and how many watts of power (CPU power + cooling) it would take.

Algorithms on the topic of linear algebra or databases could be better examples but most people aren't that familiar with those. Video compression is an other good example, you could calculate how many USD it costs to compress 1 hour of 1080p video on a modern laptop (most energy efficient?) and on an older system. It may need like 100x the time and 10000x the power, making the whole process cost a million times more even without considering initial investment costs.

I think nobody has mentioned the (energy) cost of running an algorithm on these machines. For example when you are taking a photo on a phone, it is running a face recognition algorithm in real time. This consumes X watts of power and costs Y USD / second to run.

You'd need to calculate how many multi-million machines you'd need from a specific era to run the algorithm at same speed on those (assuming it was possible), and how many watts of power (CPU power + cooling) it would take.

Algorithms on the topic of linear algebra or databases could be better examples but most people aren't that familiar with those. Video compression is an other good example, you could calculate how many USD it costs to compress 1 hour of 1080p video on a modern laptop (most energy efficient?) and an older system. It may need like 1000x the time and 1000x the power, making the whole process cost million times more.

I think nobody has mentioned the (energy) cost of running an algorithm on these machines. For example when you are taking a photo on a phone, it is running a face recognition algorithm in real time. This consumes X watts of power and costs Y USD / second to run.

You'd need to calculate how many multi-million machines you'd need from a specific era to run the algorithm at same speed on those (assuming it was possible), and how many watts of power (CPU power + cooling) it would take.

Algorithms on the topic of linear algebra or databases could be better examples but most people aren't that familiar with those. Video compression is an other good example, you could calculate how many USD it costs to compress 1 hour of 1080p video on a modern laptop (most energy efficient?) and an older system. It may need like 100x the time and 10000x the power, making the whole process cost a million times more even without considering initial investment costs.