Heavy I/O blocks FreeBSD box for several seconds
ohartman at zedat.fu-berlin.de
Thu Jul 7 23:13:43 UTC 2011
On 07/07/11 09:27, Andriy Gapon wrote:
> on 06/07/2011 21:11 Nathan Whitehorn said the following:
>> On 07/06/11 13:00, Steve Kargl wrote:
>>> AFAICT, it is a cpu affinity issue. If I launch n+1 MPI images
>>> on a system with n cpus/cores, then 2 (and sometimes 3) images
>>> are stuck on a cpu and those 2 (or 3) images ping-pong on that
>>> cpu. I recall trying to use renice(8) to force some load
>>> balancing, but vaguely remember that it did not help.
>> I've seen exactly this problem with multi-threaded math libraries, as well.
> Exactly the same? Let's see.
>> Using parallel GotoBLAS on FreeBSD gives terrible performance because the
>> threads keep migrating between CPUs, causing frequent cache misses.
> So Steve reports that if he has Nthr> Ncpu, then some threads are "over-glued"
> to a particular CPU, which results in sub-optimal scheduling for those threads.
> I have to guess that Steve would want to see the threads being shuffled between
> CPUs to produce more even CPU load.
> On the other hand, you report that your threads keep being shuffled between CPUs
> (I presume for Nthr == Ncpu case, where Nthr is a count of the number-crunching
> threads). And I guess that you want them to stay glued to particular CPUs.
> So how is this the same problem? In fact, it sounds like somewhat opposite.
> The only thing in common is that you both don't like how ULE works.
> ULE has many knobs to tune its behavior. Unfortunately they are not very well
> documented and there are too many of them. So, it's not easy to find which
> combination would be the best for a particular work-load. In your particular
> case you might want to try to increase value of kern.sched.affinity to increase
> affinity of threads to their CPUs.
Not all of those using FreeBSD are developer or experts, even experts of
a very specific
area of computer science and engineering or a particular subject of the
FreeBSD kernel and its techniques of scheduling. I'm not capable of
tuning my servers via
a lot of undocumented knobs, I'm sorry. I'd like to do if there would be
a kind of howto
> Also, please note that FreeBSD support in GotoBLAS is not equivalent to Linux
> support as I have pointed out before. On Linux they bind their threads to CPUs
> to avoid the situation that you describe. Apparently they didn't know how to do
> CPU-binding on FreeBSD, so this is not implemented. You may have a motivation
> to help them out with this.
More information about the freebsd-performance