On Tue, 8 Aug 2017 at 15:10, <mechanical-sympathy@googlegroups.com> wrote:
> mechanical-sympathy@googlegroups.com > <https://groups.google.com/forum/?utm_source=digest&utm_medium=email#!forum/mechanical-sympathy/topics> > Google > Groups > <https://groups.google.com/forum/?utm_source=digest&utm_medium=email/#!overview> > <https://groups.google.com/forum/?utm_source=digest&utm_medium=email/#!overview> > Today's topic summary > View all topics > <https://groups.google.com/forum/?utm_source=digest&utm_medium=email#!forum/mechanical-sympathy/topics> > > - JVM random performance <#m_6674955155299118709_group_thread_0> - 1 > Update > - failing to understand the issues with transparent huge paging > <#m_6674955155299118709_group_thread_1> - 6 Updates > > JVM random performance > <http://groups.google.com/group/mechanical-sympathy/t/b40e33933400997f?utm_source=digest&utm_medium=email> > Oleg Mazurov <omazu...@splicemachine.com>: Aug 07 10:29PM -0700 > > I ran the benchmark with a profiler and was able to reproduce both modes, > fast and slow. The difference appears to be due to how HotSpot compiles > the > DefaultMailbox.run() -> ...more > <http://groups.google.com/group/mechanical-sympathy/msg/1b88e54d0aeed?utm_source=digest&utm_medium=email> > Back to top <#m_6674955155299118709_digest_top> > failing to understand the issues with transparent huge paging > <http://groups.google.com/group/mechanical-sympathy/t/b258f37a19c23595?utm_source=digest&utm_medium=email> > Peter Veentjer <alarmnum...@gmail.com>: Aug 07 08:42AM -0700 > > Hi Everyone, > > I have a failing understanding the problem with transparent huge pages. > > I 'understand' how normal pages work. A page is typically 4kb in a virtual > address space; each process has ...more > <http://groups.google.com/group/mechanical-sympathy/msg/18b6f08ffea23?utm_source=digest&utm_medium=email> > Marshall Pierce <marsh...@mpierce.org>: Aug 07 10:54AM -0500 > > AFAIK, The issue is not huge pages that applications opt in to and > control; it's when they are "transparent", as in when the kernel > "helpfully" decides when to use them. This can lead to significant ...more > <http://groups.google.com/group/mechanical-sympathy/msg/4103f783dd05e?utm_source=digest&utm_medium=email> > Peter Veentjer <alarmnum...@gmail.com>: Aug 07 09:02AM -0700 > > Your answer makes a lot of sense. It makes it the 'transparent' part clear. > > What doesn't make sense is why the kernel would decide to switch between > large/small pages. So what is the benefit to ...more > <http://groups.google.com/group/mechanical-sympathy/msg/18c880fdbdade?utm_source=digest&utm_medium=email> > Gil Tene <g...@azul.com>: Aug 07 10:25AM -0700 > > THP certainly sits in my "just don't do it" list of tuning things due to > it's fundamental dramatic latency disruption in current implementations, > seen as occasional 10s to 100s of msec (and ...more > <http://groups.google.com/group/mechanical-sympathy/msg/19114de0afdb3?utm_source=digest&utm_medium=email> > Gil Tene <g...@azul.com>: Aug 07 11:14AM -0700 > > To highlight the problematic path in current THP kernel implementations, > here is an example call trace that can happen (pulled from the discussion > linked to below). It shows that a simple ...more > <http://groups.google.com/group/mechanical-sympathy/msg/193bced6d6d46?utm_source=digest&utm_medium=email> > "Alen Vrečko" <alen.vre...@gmail.com>: Aug 07 08:50PM +0200 > > Saw this a while back. > > https://shipilev.net/jvm-anatomy-park/2-transparent-huge-pages/ > > Basically using THP/defrag with madvise and using > -XX:+UseTransparentHugePages -XX:+AlwaysPreTouch JVM ...more > <http://groups.google.com/group/mechanical-sympathy/msg/419d14267ecec?utm_source=digest&utm_medium=email> > Back to top <#m_6674955155299118709_digest_top> > You have received this digest because you're subscribed to updates for > this group. You can change your settings on the group membership page > <https://groups.google.com/forum/?utm_source=digest&utm_medium=email#!forum/mechanical-sympathy/join> > . > To unsubscribe from this group and stop receiving emails from it send an > email to mechanical-sympathy+unsubscr...@googlegroups.com. > -- Isaiah Perumalla -- You received this message because you are subscribed to the Google Groups "mechanical-sympathy" group. To unsubscribe from this group and stop receiving emails from it, send an email to mechanical-sympathy+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.