> If it's operational in seconds it was hibernated or suspended.

This is not true. Are you still using a platter hard drive? (If an SSD, have you looked up benchmarks for it?)

My ~5 year old laptop used to cold boot Windows 7 in less than 10 seconds (once I'd disabled most autostarting programs, at least). It currently cold boots Ubuntu in ~5 or so; most of that time is spent displaying the UEFI and Grub splash screens. This is made possible _almost entirely_ by a Samsung Evo; I'm looking forward to getting an M.2 drive when I replace the computer.

Are you pressing the button to start and externally timing the process of arriving at a usable desktop and have you explicitly disabled fast boot? This isn't a new feature in windows 10.

Internally my computer tells me the process takes about 5 seconds from the OS start to graphical environment but in reality there are several steps. For example this doesn't account for the period of time between hitting the power button and the OS itself starting to run, entering full disk encryption password, unlocking volume.

I would be surprised if a full restart actually took so short. Maybe not loading a menu or unlocking a volume is sufficient to explain the difference?

I often hear hibernation is one of the great Mac Os Features. For my part, I decided years ago that cold booting windows works much faster than hibernating it.

I will not start to count the seconds and fight which OS boots faster, but it is certainly much faster than it was in the nineties. Boot times are certainly one thing where modern computers have significantly improved. Everyone who compares an instant on 8-bit is oversimplifying things. Try booting to ie. GEOS on one of those.

That's windows. In Linux, I get this from cold boot (on an Thinkpad T495):

    > systemd-analyze 
    Startup finished in 8.878s (firmware) + 1.666s (loader) + 1.592s (kernel) + 3.265s (userspace) = 15.403s 
    graphical.target reached after 3.176s in userspace

It's crazy that the slowest part of the whole process is the firmware stuff.

Is that wall clock time? I get 13s but actual wall clock time is more like 26 from power on to desktop. Around 30 seconds to opening a browser.

I haven't check wall clock; I would have to enable auto-login to eliminate pauses from user interaction.

> I would have to enable auto-login to eliminate pauses from user interaction.

Why? By the time you're logging in, booting is already finished.

To check at what point the system is responsive enough to launch a browser. So, not just calculating the boot times (we have that already), but from power on to launching a browser.

While that's not an entirely unreasonable idea, I note that launching a browser has its own startup time. Power on to opening a document in evince is faster than power on to launching a browser, and that's support for the original thesis in a way, but it also feels like a little much to count it against "boot time".

Just tested mine from totally off to login screen, took slightly over 12 seconds.