Hacker News new | past | comments | ask | show | jobs | submit login

granted, while it's great in concept to work on ML everywhere, given that it is such a data and GPU-hungry process, the tradeoffs for an ML capable machine would make a shitty laptop. All the nVidia engineers that I see use a large "portable desktop" that has to accommodate the increased cooling and power contraints for a discrete GPU.

And, after having a number of laptops (and an iMac) die from heat death due to the ball grid array solders of the GPU wearing out, I'm done with discrete GPU's in anything other than a desktop box.

Better in this day and age to prototype on the macbook, then offload the heavy lifting to a dedicated ML box or, at the rate things are going, to Amazon...




I generally do agree, but 'laptop' is a little disingenuous when you're talking about an engineer's workstation—I get very little code written out of reach of a power socket. I'd even put up with a heavier/clunkier/louder machine (slightly) for a gpu/power boost.

They did use to make 17" laptops; calling those portable was really stretching the word.




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

Search: