3 ms·
The is a great study showing the algorithms have improved faster than Moore's law and advances in computing in many cases, if only the abstraction allowed for f
by vmception 5y ago
The is a great study showing the algorithms have improved faster than Moore's law and advances in computing in many cases, if only the abstraction allowed for frontend and end users to experience these gains more often.
- taeric 5y agoI'm torn. A lot of the advances in algorithms required advances in memory capacity. Such that it seems odd to claim they outpaced Moore's law. True, but they could really only do it with said law.
- vmception 5y agoI think these concepts do not compete. The gains are not linearly related to the capacity or speed of memory. The memory improvements accomplish just the removal of a bottleneck. You remove a bottleneck and everything flows faster instead of pools. So the increase itself can be greater than the same as the increase in the size of the corridor.
- taeric 5y agoThey were somewhat linear with relation to capacity of memory, but they also might not have been possible with capacity from a decade ago. This is easily seen if you look at fun algorithms such as BDDs. If you have the memory for them, they are amazingly fast. But, if you don't have the memory for it, it gets you nothing. Edit: removed a "not" that I didn't mean to type. :(