On Friday 2012-06-15 17:10, Greg KH wrote:
Yes, using tmpfs is faster, but you loose the persistance when power goes off, and usually you don't have that much ram on your system.
Since when do you need power-persistence for build roots?
You have to read the rpms you install for the build from somewhere.
I do not think that reading the RPMs is so much of an issue (because it's generally less than 500). That which takes significant amounts of time is writing all the files that those RPMs extract: find tells me I have ~34000 files in /var/tmp/jng-openSUSE_12.1-x86_64. And that for a packge that has no BuildRequires: in the .spec file at all. (It's just repacking PHP files, which is a noarch job.) Clearly, extracting 34k files into tmpfs/SSD is a heck of a lot faster than common disks.
I have tied a tmpfs to /var/tmp/jng-openSUSE_12.1-x86_64 and factory-x86_64 so that at least the majority of my `osc build` invocations run accelerated. I can't say I see much difference to a system where /var/tmp is part of a SSD.
Really? Are you CPU bound then? If you throw a bigger processor at it, does that speed things up?
It's IO-bound, which - naturally, if tmpfs is used - becomes a pure CPU/memory-bound workload. -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org