Been 30 years since I've written assembler but back in the day I considered myself part of Abrash's army. What we tend to forget about writing in assembler is each typical formatted line has a one-to-one mapping to a machine instruction unless you're using macros. It really is the most wysiwyg programming experience you can have, so it's simply not possible to get the same optimization and space usage with a higher level language because your only choice for optimizing higher level is via groups of instructions. So C optimizers are inlining groups of instructions and making best assumptions about execution flows for you while maintaining intent. So if you want to enjoy the abstractions of higher languages, and the speed at which you can develop, the only obvious run-time optimization is faster hardware to compensate for abstractions that might decompile to massive amounts of machine instructions.

Oh yeah and I'd just like to mention as an aside that assembler statements make more sense to me than bootstrap css codes - a sad statement to the art of reasonable brevity.