diff options
author | Yufei Ren <renyufei83@gmail.com> | 2012-10-19 23:11:52 -0400 |
---|---|---|
committer | Jens Axboe <axboe@kernel.dk> | 2012-10-22 10:02:40 +0200 |
commit | d0b937ed8841484c2d651919943e8e4c70dfd460 (patch) | |
tree | c307023f27ea9244c08229b4ffd2fe09e267e069 /README | |
parent | f5fd0b1cd383c4a3cb8fb5078b7e93aecba8847e (diff) | |
download | fio-d0b937ed8841484c2d651919943e8e4c70dfd460.tar.gz |
Fine-grained job level numa control
Two new options, numa_cpu_nodes and numa_mem_policy, are created
for a fine-grained job level numa control. Please refer HOWTO and
README for detailed description.
A example job, examples/numa, is added as well.
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Diffstat (limited to 'README')
-rw-r--r-- | README | 14 |
1 files changed, 10 insertions, 4 deletions
@@ -233,10 +233,11 @@ The job file parameters are: readv/writev (with queuing emulation) mmap for mmap'ed io, syslet-rw for syslet driven read/write, splice for using splice/vmsplice, sg for direct SG_IO io, net - for network io, or cpuio for a cycler burner load. sg - only works on Linux on SCSI (or SCSI-like devices, such - as usb-storage or sata/libata driven) devices. Fio also - has a null io engine, which is mainly used for testing + for network io, rdma for RDMA io, or cpuio for a + cycler burner load. sg only works on Linux on + SCSI (or SCSI-like devices, such as usb-storage or + sata/libata driven) devices. Fio also has a null + io engine, which is mainly used for testing fio itself. iodepth=x For async io, allow 'x' ios in flight @@ -255,6 +256,11 @@ The job file parameters are: ratecycle=x ratemin averaged over x msecs cpumask=x Only allow job to run on CPUs defined by mask. cpus_allowed=x Like 'cpumask', but allow text setting of CPU affinity. + numa_cpu_nodes=x,y-z Allow job to run on specified NUMA nodes' CPU. + numa_mem_policy=m:x,y-z Setup numa memory allocation policy. + 'm' stands for policy, such as local, interleave, + bind, prefer, local. 'x, y-z' are numa node(s) for + memory allocation according to policy. fsync=x If writing with buffered IO, fsync after every 'x' blocks have been written. end_fsync=x If 'x', run fsync() after end-of-job. |