=== Guest63440 is now known as RAOF
=== cachio_afk is now known as cachio
=== JanC is now known as Guest30327
=== JanC_ is now known as JanC
[13:30] <jtaylor> have there been scheduler changes in recent kernel updates?
[13:30] <jtaylor> I have schedule and sched_clock show up at 25% in all of my profiles
[13:30] <jtaylor> that wasn't the case a few weeks ago
[14:49] <jamieiles> Hi folks, the Ubuntu-3.13.0-113.160 tag isn't present in ubuntu-trusty.git, looks like it should be b5fea3e677cf9b2f85237eed1a6c56191384342d
[14:54] <apw> jamieiles, hmmm, looking
=== JanC_ is now known as JanC
[15:27] <jamieiles> apw: I see the tag now, thanks
[15:28] <apw> jamieiles, great ...
[16:15] <tseliot> apw, rtg: are we going to get a backport of this commit in xenial? It looks trivial, and I think it would make sense to have it: https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=08b259631b5a1d912af4832847b5642f377d9101
[16:16] <rtg> tseliot, is it relevant for a 4.4 kernel ? The commit says 4.9
[16:19] <tseliot> rtg: the CPU is supported by 4.4, but then it crashes after a while, as reported here: https://www.servethehome.com/amd-ryzen-with-ubuntu-here-is-what-you-have-to-do-to-fix-constant-crashes/
[16:22] <rtg> tseliot, their fix is to run a 4.10 kernel.
[16:23] <rtg> tseliot, we'll have a 4.10 based hwe-edge kernel in Xenial fairly soon
[16:23] <tseliot> rtg: right, but they mention the actual commit (with a link): "There have been numerous reports of SMT issues and it is a known fix that got rolled out in the Linux 4.10 and 4.11 kernels as we mentioned in our release piece. "
[16:23] <tseliot> rtg: when would that be?
[16:23] <rtg> tseliot, I hope to get it into proposed today. working with apw on that
[16:24] <tseliot> rtg: ok, cool
[17:06] <dmj_s76> sforshee: I'll test out the -22 ucode when I get a chance
[18:08] <sforshee> dmj_s76: thanks!
=== mwhudson_ is now known as mwhudson
=== diddledan_ is now known as diddledan
=== ulkesh_ is now known as ulkesh
[21:22] <dmj_s76> sforshee: I tested on one machine with an 8260 and it seems the new -22 ucode doesn't have the same problem.  There could be regressions elsewhere, but at least on the machine I tested things were working fine.
=== daniel1 is now known as Odd_Bloke