Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

embedded systems are resource and cost-constrained devices. If you can use a slightly less powerful device and save 2c per device... multiply that by 10 million production run makes a really significant difference.

efficiency really matters and costs real money.

Of course if you're using dynamic allocation in an embedded system then you have other issues anyway...



> efficiency really matters and costs real money.

Sometimes.

10 million times $.02 is only $200,000.

And the probability of you shipping 10 million units of anything is vanishing small. 10,000 is more typical--at which point the savings is $200. Not having to think saves you more money.


only 200,000... granted... numbers plucked out of thin air...

vanishingly small... I'm currently designing a product that will ship around 30 million units and have done similar in the past.

So, no... the point stands.


If it really produces visible impact for the target use case for the software, in a situation like you describe, then yes.

If a developer is doing micro-optimizations without numbers of what it brings in terms of business case, then no.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: