3 ms·
This make me wonder how could this be applied in an interpreter. I have one with array/relational semantics and have wonderer how to pack N-values per row/colum
by mamcx 5y ago
This make me wonder how could this be applied in an interpreter. I have one with array/relational semantics and have wonderer how to pack N-values per row/column that is efficient to encode/Decode, but always fear that all the extra processing will be costly and then just let Rust do his magic.
- dragontamer 5y agoI mean, the #1 job of any optimizing programmer is rather simple when you think of it. 1. Program attempt A. 2. Program attempt B using the same interface. 3. Develop a suite of tests. 4. Profile A and B using those tests. 5. Choose the methodology that uses fewer resources (in particular: time, memory, cache). 6. Use experience from 1-5 to make attempt C. Then repeat again. #5 is worded like that on purpose. Different programs will be constrained by different resources. Time itself is often times a constraint on resources (run out of RAM, your program can continue but only through Swap. Run out of L3 cache, your program will continue, but through RAM). A methodology that uses less RAM will stay in L1 cache longer, but if everyone's data is so huge you're in DDR4, maybe a different methodology would work instead.