Maybe old software was also fast because the people building it were more likely to care about computers first and careers second. Not that they were better, just that fewer people were there because tech was the obvious high-paying path.
One of my favourite words in engineering is resourcefulness.
For simplification, you need to make a Spaghetti Bolognese for 4 people.
Person A gets $10, Person B gets $100.
Person A is forced to be resourceful, look around and do a lot of thinking. Person B can be wasteful and still be in budget.
Reality Nowadays: Person B would contract this out to Person C, who would subcontract to Person D and suddenly there is a huge scope creep and $100 is not enough.
> Sometimes, hardware is cheaper than human coordination.
A t3.small on AWS costs $182.21 a year before any discounts and has 2 CPUs and 2 GB of RAM.
So the computer to run the example at the start costs 3 hours of Engineering time.
This has... Warping effects on how hardware performance is perceived to put it mildly.
If you spend 4 hours halving that cost it takes multiple years to reclaim that investment.
Not that performance doesn't matter of course, reducing your total spend by a percentage is worthwhile, but micro optimizations become difficult when hardware is cheap and performant.
Old software wasn’t universally fast. And the load/expectations of server software was quite a bit lower due to fewer users, in the case of a web server.
I can remember the slow loading of Bryce 3D or Photoshop or Office on P133 with a 5400 RPM drive.
9 comments
[ 3.2 ms ] story [ 35.9 ms ] threadI use old software on new hardware
It's faster than new software on new hardware
For simplification, you need to make a Spaghetti Bolognese for 4 people.
Person A gets $10, Person B gets $100.
Person A is forced to be resourceful, look around and do a lot of thinking. Person B can be wasteful and still be in budget.
Reality Nowadays: Person B would contract this out to Person C, who would subcontract to Person D and suddenly there is a huge scope creep and $100 is not enough.
A t3.small on AWS costs $182.21 a year before any discounts and has 2 CPUs and 2 GB of RAM.
So the computer to run the example at the start costs 3 hours of Engineering time.
This has... Warping effects on how hardware performance is perceived to put it mildly.
If you spend 4 hours halving that cost it takes multiple years to reclaim that investment.
Not that performance doesn't matter of course, reducing your total spend by a percentage is worthwhile, but micro optimizations become difficult when hardware is cheap and performant.
I can remember the slow loading of Bryce 3D or Photoshop or Office on P133 with a 5400 RPM drive.
The nostalgia is strong with this one.