# 🛠 devtools

# Memory consumption in node and chrome?

v8 is the fastest JavaScript engine

Top most reason for chrome crashing today
- due to bugs in web-apps called memory leaks
There is more traffic coming from mobile and IoT, kind of restricted with memory

#cpu cores packaged into chips
sharing a global memory called RAM
Access to memory is synchronized within all cores in some fashion with limited memory bandwidth

CPU core

CPU cores itself have their own registers and L1, L2 level caches.

Additionally a third layer cache L3 shared by all the cores

Mind your memory footprint
- avoid crashes due to out-of-memory
- benefit from faster caches(which are limited in size)

light version of v8 for mobile

regEx in v8

memory occupied by regular expressions

751 bytes of machine code just for regex /./ dot operator

regex are generally executed more than once
- 1st execution uses interpreter and
- subsequent calls to regex uses optimization engine
regex improvements by 29% in real websites

~70% of manged memory is pointers (small integers)

engine starts with sentinel pointers, and shape
- chrome limits memory to 2 or 4 GB depending on system in use
Pointer Compression? as most of data (70%) is pointers
- compress the biggest chunk in memory
- disable by default for nodejs
- 4GB compressed 6 to 8 GB uncompressed, 20-30% reduction in size

30% of time is parsing and compiling
- which is almost similar to time spent on executing the script
How to you improve it?
- stack trace
- team look out the nitty gritty of it and micro optimized the shot out of it
- chrome 75 can now parse Facebook js
Too much surprise, JSON parsing is faster than js Object
- Why is that?
- JSON.parse() call expression with a single argument
- js object has far more tokens
  - each property name is identifier or a string like literal
  - values could be anything - nested Objects, Arrays
  - work harder to tokenize it correctly

JS Parsing

Pretend you are the parser and looking at source code character by character

Parsing javascript is hard because its context sensitive.

parsing JSON is comparatively easy as its context insensitive

How to use this micro optimization?
tooling can help, yet keeping you source code readable
- json-parse-webpack
- babel-plug-object-to-json-parse

const data = { foo: 123, bar: 456 };
// equivalent
const data = JSON.parse('{"foo":123,"bar":456}'); // faster

Main thread blocked for work, like user-interactions, scrolling, page load
Chrome uses resourcePreloader
- defer loading script
Google metrics called TTS - Time to Shakira
- time to load facebook shakira page
  - facebook sends 30 mb of scripts with each page load
- load google sheets