March 20th, 2020 at 12:12 AM
CPUs are insanely fast these days. the ~3 billion clock cycles/second can result in even more instructions per second, as modern CPUs can exucute 4-5 in parallel (7 in the case of the new iPhone). But I mean, lack of optimization is part of why software has not exactly gotten faster at the rate that CPUs have gotten faster. My Mac still takes about 5-10 seconds to open Word. Heck, a comparable text editor for the time in 2000 would have loaded in about the same amount of time on a Pentium 3.
But hey, I do know that one thing that modern CPUs are focusing on is in the arena of memory optimization. AMD is throwing a ridiculous amount of cache onto their new chips. Generally, you'll only have a very slight increase in performance for every time you double cache. Most of the cache-essential data can be stored in the first ~2MB or so, and the rest only boosts performance slightly, but when you're loading a string that could be megabytes long and you need to quickly calculate the length over and over again, more cache is good.
I'm often amazed at how well optimized glibc truly is. Clearlinux has some optimizations in addition to the standard GNU stack, which amazes me even more.
But hey, I do know that one thing that modern CPUs are focusing on is in the arena of memory optimization. AMD is throwing a ridiculous amount of cache onto their new chips. Generally, you'll only have a very slight increase in performance for every time you double cache. Most of the cache-essential data can be stored in the first ~2MB or so, and the rest only boosts performance slightly, but when you're loading a string that could be megabytes long and you need to quickly calculate the length over and over again, more cache is good.
I'm often amazed at how well optimized glibc truly is. Clearlinux has some optimizations in addition to the standard GNU stack, which amazes me even more.
