The numerous references to hardware in your comment only serve to illustrate my point.
Of course there have been some new algos and improvements to existing ones since 1984. But nothing approaching a Moore's Law. The dramatic changes the blog post highlighted are due to hardware, not software.
Take away the hardware advances (hold the hardware as a constant), then look at software as the variable. Then we can have a meaningful discussion of software improvements over time. IOW, run new software on old hardware.
To measure advances in hardware, hold software as the constant. IOW, run old software on new hardware.
If both are variables, it's difficult to assess how much software has improved on it own, without the benefit of new hardware.
Linus Torvalds' comments on the future of RISC, CPU instructions, and compatibility versus "cool new features" in the /. interview were interesting.
Of course there have been some new algos and improvements to existing ones since 1984. But nothing approaching a Moore's Law. The dramatic changes the blog post highlighted are due to hardware, not software.
Take away the hardware advances (hold the hardware as a constant), then look at software as the variable. Then we can have a meaningful discussion of software improvements over time. IOW, run new software on old hardware.
To measure advances in hardware, hold software as the constant. IOW, run old software on new hardware.
If both are variables, it's difficult to assess how much software has improved on it own, without the benefit of new hardware.
Linus Torvalds' comments on the future of RISC, CPU instructions, and compatibility versus "cool new features" in the /. interview were interesting.