9fans archive / 1999 / 02 / 2 /    prev next

From: Franklin Robert Araujo França 973930@dcc...
Subject: [9fans] Hbench:OS on Plan9 - a comparison with Linux - more measurements
Date: Tue, 02 Feb 1999 09:38:08 +0000

Hbench:OS on Plan9 - a comparison with Linux - more measurements
HW: 200Mhz 64MB Pentium with a 3GB Quantum IDE disk
Where Plan9 was faster:
Forking a null process: 850 us x 1290 us
copying large chunks of memory (512KB)   52MB/s x 35MB/s

Where there was a tie:
Context switching latency of 20 processes 64KB each: ~400 us
Forking a null process thru APE : ~1300 us

Where Linux was clearly faster:
copying memory inside the L1 cache ( 4KB chunks)  420 MB/s x 84 MB/s
zeroing small amounts of memory(8KB)via bzero()   725 MB/s x 84MB/s
Reading sequentially a previously cached 8MB File: 37 MB/s x 0.53 MB/s
Reading sequentially a non cached 100MB file(*):  8.3 MB/s x 1.5 MB/s
Writing sequentially a 100MB file:                4.6 MB/s x 0.36MB/s
Reading ramdonly a 100MB file (8KB blocks):      0.86 MB/s x 0.15 MB/s
pipe bandwidth( 64KB buffer)                       44 MB/S x 17 mb/S
tcp bandwidth  ( 64 KB buffer)                      19MB/s x 0.61 MB/s

(*)Linux 8.3 MB/s is about 90% of the disk nominal transfer rate.
Notes:
(i) HBench and lmbench C sources can be found in:
[Brown et al 97] Aaron B. Brown & Margo I. Seltzer, "Operating
System Benchmarking in the Wake of Lmbench: A Case Study of the
Performance of NetBSD on the Intel x86 Architecture", Sigmetrics
1997; also in : www.eecs.harvard.edu/vino/perf/hbench/index.html.

[Mc Voy & Staelin 96] Larry McVoy & Carl Staedlin, "lmbench:
Portable tools for performance analysis", Proceedings of the 1996
Usenix Technical Conference, San Diego, CA, Jan 1996, 279-295; also
in : www.bitmover.com/lmbench/lmbench-usenix.ps.gz

(ii)more results in: www.dcc.unicamp.br/~celio/plan9/benchmarks

Celio Guimaraes & Franklin Franca
Institute of Computing
Unicamp, Campinas, Brazil
celio@dcc...
franklin robert araujo franca <973930@dcc...>