hi, Christoph Hellwig, On Tue, Aug 13, 2024 at 08:49:55AM +0200, Christoph Hellwig wrote: > On Tue, Aug 13, 2024 at 02:47:48PM +0800, Oliver Sang wrote: > > no in our tests, becomes even worse actually. detail data is in [1] > > Thanks. Do you have a good summary of what operations this counts? sorry I don't have many details. not sure if https://github.com/filebench/filebench/wiki is helpful for you? and a raw data run by this test looks like below 2024-08-13 02:24:56 filebench -f /lkp/benchmarks/filebench/share/filebench/workloads/randomrw.f Filebench Version 1.5-alpha3 0.000: Allocated 177MB of shared memory 0.001: Random RW Version 3.0 personality successfully loaded 0.001: Populating and pre-allocating filesets 0.001: Removing largefile1 tree (if exists) 0.002: Pre-allocating directories in largefile1 tree 0.119: Pre-allocating files in largefile1 tree 49.469: Waiting for pre-allocation to finish (in case of a parallel pre-allocation) 49.469: Population and pre-allocation of filesets completed 49.469: Starting 1 rand-rw instances 50.472: Running... 110.478: Run took 60 seconds... 110.478: Per-Operation Breakdown rand-write1 11769ops 196ops/s 1.5mb/s 5.096ms/op [0.002ms - 1567.054ms] rand-read1 49402ops 823ops/s 6.4mb/s 1.214ms/op [0.001ms - 1567.058ms] 110.478: IO Summary: 61171 ops 1019.414 ops/s 823/196 rd/wr 8.0mb/s 1.961ms/op 110.478: Shutting down processes 2024-08-13 02:26:48 sleep 100 /nfs/sdd1/largefile1 /nfs/sdd1/lost+found /nfs/sdd1/wait_for_nfs_grace_period 2024-08-13 02:28:28 rm -rf /nfs/sdd1/largefile1 /nfs/sdd1/lost+found /nfs/sdd1/wait_for_nfs_grace_period our bot parse it as: { "filebench.sum_operations": [ 61171 ], "filebench.sum_operations/s": [ 1019.414 ], "filebench.sum_reads/s": [ 823 ], "filebench.sum_writes/s": [ 196 ], "filebench.sum_bytes_mb/s": [ 8.0 ], "filebench.sum_time_ms/op": [ 1.961 ] }
