[j1vms]

As indicated in the paper, patches (currently for kernel 4.1) are here: https://github.com/jplozi/wastedcores

[rando3826 → j1vms]

Are they accepted upstream? If not, I'd like to see a link to the lkml thread for these.

[justinmk → rando3826]

I wonder if they are discouraged by the experience of Con Kolivas[1] who proposed an alternative scheduler back in 2007. (Apparently he is still maintaining his "-ck" fork of the linux kernel[2]!)

I only mention this as a historical case that has remained in my memory. Maybe Linus is willing to revisit the issue, I don't follow LKML.

[1] https://en.wikipedia.org/wiki/Con_Kolivas

[2] http://ck-hack.blogspot.com/2015/12/bfs-467-linux-43-ck3.htm...

[doe88 → justinmk]

I do not follow Kernel Dev enough to have a good representation of what happened, but it seems to me that it was another example of smart people pushed out. Nowadays I think he is just maintaining his patches from one kernel release to the next, it was smart for Con to stop interacting with them and move to other types of devs, it comes a point where you have to keep your sanity.

[LoSboccacc → justinmk]

I remember the whole debacle back then when the lkml was summarized on that website I can't remember

It'd be interesting to see if that branch exhibit the same behavior and issues.

[tremon → LoSboccacc]

Maybe KernelTrap? I was sad when it shut down, it was a very useful resource for following Linux development from the sidelines.

[LoSboccacc → tremon]

Yeah that one! The memories.

[DCoder → LoSboccacc]

You might be thinking of LWN. They covered the scheduler multiple times over the years: https://lwn.net/Kernel/Index/#Scheduler

[rincebrain → rando3826]

I can't find any evidence they were submitted upstream at any juncture.

Moreover, since (several of) the patches seem to rip out a bunch of logic in favor of very simple logic, they would probably be contentious without broad testing and probably a runtime option to configure the behavior.

[verroq → rincebrain]

And uses spaces instead of tabs. Someone will need to go fix them up.

[jeffreyrogers → verroq]

Not sure why you're being downvoted. Tabs instead of spaces is the Linux kernel style.

[melchebo → rando3826]

The conference where this will be presented starts next week. So I think they'll post it after they catched up some sleep after that.

[anonymousDan → melchebo]

Which conference, Eurosys?

[desdiv → anonymousDan]

Yes, since the filename is eurosys16-final29.pdf.