I have been hearing a bit of feedback from my BSD friends, several of whom
are committers, about the superiority of freeBSD over Linux. Recently I
got an earful of "BSD runs Linux binaries faster than Linux."
SO here's the result of measuring IPC speed on two machines, identical
except one turns out to be a P=II-350 instead of 400. Doesn't matter, the
test results are normalized to CPU speed on identical CPUs, and the BSD
machine was faster so they can't claim a foul.
The results labeled "wcox1" are the test compiled native on BSD. The
results labeled "wcox_lxexe" are the Linux executable running under the
Linux environment of BSD. Other than the release of a shared memory
segment failing in that mode, it did run fine.
I had to hack the Makefile and data extract script slightly to run on BSD,
I didn't get bash in a default install and just made the Makefile a bit
more portable. I'll merge that with some pthreads work I'm doing and try
to release it by the end of the week as a new ctxbench version. All
results were 1.10 release, compiled from the same source.
================================================================
Run information
================================================================
Run: wcox1
CPU_MHz 400
CPUtype P-II
HostName wcox.prodigy.com
KernelName FreeBSD 5.0-RELEASE
Ncpu 1
Run: wcox_lxexe
CPU_MHz 400
CPUtype P-II
HostName wcox.prodigy.com
KernelName FreeBSD 5.0-RELEASE
Ncpu 1
Run: 2.5.52
CPU_MHz 348.395
CPUtype Pentium II (Deschutes)
HostName oddball.prodigy.com
KernelName Linux 2.5.52
Ncpu 1
Run: 2.5.54
CPU_MHz 348.444
CPUtype Pentium II (Deschutes)
HostName oddball.prodigy.com
KernelName Linux 2.5.54
Ncpu 1
Run: 2.5.56
CPU_MHz 348.427
CPUtype Pentium II (Deschutes)
HostName oddball.prodigy.com
KernelName Linux 2.5.56
Ncpu 1
Run: 2.5.59n
CPU_MHz 348.395
CPUtype Pentium II (Deschutes)
HostName oddball.prodigy.com
KernelName Linux 2.5.59-1n
Ncpu 1
================================================================
Results by IPC type
================================================================
loops/sec
SIGUSR1 low high average avg/MHz
wcox1 9 245 94 0.236
wcox_lxexe 967 13799 5767 14.419
2.5.52 40961 41699 41290 118.517
2.5.54 39573 39699 39647 113.785
2.5.56 36875 36933 36909 105.932
2.5.59n 41310 41865 41542 119.239
loops/sec
message queue low high average avg/MHz
wcox1 24485 24792 24653 61.634
wcox_lxexe 23259 23259 23259 58.148
2.5.52 78427 79163 78675 225.823
2.5.54 74131 75390 74969 215.156
2.5.56 68177 71334 70228 201.557
2.5.59n 76381 76537 76472 219.500
loops/sec
pipes low high average avg/MHz
wcox1 31641 32284 32057 80.143
wcox_lxexe 31085 31840 31463 78.658
2.5.52 77020 77586 77365 222.061
2.5.54 70445 70572 70489 202.297
2.5.56 66728 71280 69736 200.145
2.5.59n 67564 68632 67952 195.044
loops/sec
semiphore low high average avg/MHz
wcox1 36311 36962 36678 91.697
wcox_lxexe 33709 34858 34502 86.255
2.5.52 84367 84639 84516 242.589
2.5.54 77304 77521 77379 222.072
2.5.56 77120 77358 77245 221.697
2.5.59n 83412 83720 83536 239.774
loops/sec
spin+yield low high average avg/MHz
wcox1 66551 71060 68807 172.018
wcox_lxexe 64707 68347 66356 165.890
2.5.52 219279 219658 219470 629.947
2.5.54 160157 178830 172598 495.341
2.5.56 172703 173159 172973 496.440
2.5.59n 181186 181848 181446 520.806
loops/sec
spinlock low high average avg/MHz
wcox1 50 50 50 0.125
wcox_lxexe 49 50 49 0.125
2.5.52 3 3 3 0.009
2.5.54 3 3 3 0.009
2.5.56 3 3 3 0.009
2.5.59n 3 3 3 0.009
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Fri Jan 31 2003 - 22:00:20 EST