Hacker News Reader

Comment by imoreno | Hacker News Reader

> Why do we write programs with one thread that executes the instructions in order? Why not just always write fully parallelized code?

Maybe to a layperson this is an interesting question, but surely on this site the answer is obvious.

I skimmed through anyway, just on the off chance he's talking about some new GPU development I didn't know about. But no, it's just a clickbaity title for a basic "pros and cons of parallelizing" article of mediocre quality.

View on HN

You're not the target audience. This is a very well-written explainer for non-techies. I've already emailed it to a couple of friends. Lovely.

This whole site is not the target audience.

> a clickbaity title for a basic "pros and cons of parallelizing" article of mediocre quality.

Why don't you write one with good quality (if not great) that is understandable to the audience of laypersons interested in, having or asking this question.

I think the main point here is that there is no fundamental amazing insight here for professionals to ingest.

This exact question comes to my mind in the last 2 hours after reading about the new Nvidia Digit supercomputer (which will retail at $3,000) and RTX 5090 GPU (which will cost $2,000). At exactly those moments, I saw this submission here.

Teddy Wahle did a good job explaining it to non-professionals who are using gaming laptops/PCs in their daily lives for content creation and such.

If I can think this question, imagine thousands (if not millions) of people who have or will have similar questions.

Though no modem CPU actually executes instructions in order.

Because CPUs care about semantic ordering

Reply on news.ycombinator.com