A Two-Second Boot Time With systemd

[GreatEmerald]

I was under the impression that this simply moved the boot loader into the logic board's firmware. This is nice and all, but all that this really accomplishes (in the context of boot time) is the reduction of one pile-of-time and an increase in some-other-pile-of-time. In other words, there is no real net-gain, unless the UEFI boot loader implementation is significantly faster than grub.

I may be mistaken about UEFI's both implementation, having read only a handful of pages of the spec.

F

I'd imagine that would still boost the speed a little, as the PC doesn't need to hand off control to a whole new environment. However, it doesn't take all that long to make the transfer, for one (although changing the VGA mode does take a little time), and the major boost in speed that Coreboot gives you is that it's especially tailored for speedy boot. UEFI is in essence an operating system in and of itself, so it takes a while to load it. Coreboot is just an extremely small program that does one thing and one thing alone, initialises hardware and hands over control as fast as possible. So using Coreboot would still be noticeably faster compared to UEFI direct boot to Linux.

[Veerappan]

There are a lot of people who still don't use suspend-to-ram, suprisingly.

I boot maybe once every 10-30 days for a variety of reasons, and the boot time is short enough now that it never concerns me.

Consider yourself lucky. My desktop never successfully wakes from suspend (Gigabyte 785G mobo, x6 1055T, Radeon HD 6850, Mint 12, previously Ubuntu 11.10). My laptop and other workstations wake properly, but not my primary desktop where it matters most to me.

[curaga]

Huh, I look out of place here not using suspend even though it works. Just no point in it when boot is 5s and procudes a much cleaner environment.

[Kano]

@russofris

Well compared to coreboot uefi is slow. But every new system has got uefi (well done w8) and the old way would be activating the csm for bios compat at the end and run old mbr grub code or use uefi to boot efi grub. So when you skip that and directly load the kernel - without initrd as that would require efi shell access (maybe patching efibootmgr to support options would be enough) you definitely save a few seconds. It also depends a bit on the uefi vendor, Insyde (which is used by ASUS) is a bit slow i would say. You definitely gain more speed than switching from sysvinit (makefile mode) to systemd.

[Nevertime]

Huh, I look out of place here not using suspend even though it works. Just no point in it when boot is 5s and procudes a much cleaner environment.

Maybe some would like a suspend option that logs you out of any sessions and logs you back in afresh on waking.


I'm personally happy with suspend (on the laptop it works reliably on) and mainly do a reboot if my batteries run out.

[curaga]

Also polar bears, suspend to ram uses some 1-3W still, no?

[johnc]

Huh, I look out of place here not using suspend even though it works. Just no point in it when boot is 5s and procudes a much cleaner environment.

True... but I tend to like it because I don't have to close down my open applications every night, which is especially useful when I'm doing something like, say, development and have 4 million billion trillion windows open.

Rumor is it does use <5 watts of power in suspend mode, though my UPS always reports zero, even with two computers in suspend. But it could be that the UPS is not capable of accurately reporting energy usage that low.

[russofris]

True... but I tend to like it because I don't have to close down my open applications every night, which is especially useful when I'm doing something like, say, development and have 4 million billion trillion windows open.

I'm in the same boat. I tend to have a browser open with tabs for each Jira ticket left open overnight. I have a text/scratch document for each work area. I have an e-mail client open so that my client-side rules can be applied immediately.

In instances where my work PC shuts down overnight, I am often confused and slow in the morning. Step 1 is trying to restore my browser session so I can figure out WTF I was working on last night. Step 2 is to start my e-mail client and letting the 400 overnight messages from APOC/DVCI regions sync (20-30 mins). Step 3 is to find my in-progress work.

It takes 1 hour and two coffees from the time I turn on my PC until I am able to do any actual work. It's gotten so bad that I had to implement a VM on my iMac specifically for on-call work (10 minute response SLA). It's sad that the windows VM on my iMac sleeps properly, but my Win 7 T400 does not.

F

[not.sure]

Huh, I look out of place here not using suspend even though it works. Just no point in it when boot is 5s and procudes a much cleaner environment.

You sound like you're using windows, and even modern versions of windows don't need a daily 'clean-environment' reboot any more.

And for a lot of us, having to start with a 'clean environment' is a total mega epic PITA punishment because it takes forever to get everything back in working order and in the right place. I don't even care how long it takes to wake up from suspend or hibernate as long as I get my shit back the way I left it.

Besides, the boot time pissing contest days will hopefully be over soon, with the trend going towards always-on or deep-sleep stuff like what win8+ultrabooks are aiming for.

[curaga]

@not.sure

Sounds like you need to set up some X startup scripts for your IM client etc

Always on, deep sleep = power wastage. Why would I lose battery life or pay for electricity overnight, when I'm not using it?