## The Results of our Testing
## Rob Garrett <>
Well, it was my intention in this article to provide *clear* benchmark
results of FreeBSD vs Linux, but it seems that the only thing that I was
actually able to prove is that the benchmark suites currently available for
UNIX operating systems are extremely lacking. There wasn't a clear winner,
and the results of the benchmarks are conflicting.
Bonnie shows FreeBSD to be faster than RedHat in all areas ftping a 1 Gig
file shows FreeBSD to my 10 mbit link, while Linux doesn't quite saturate it,
although it's close.
Linux Results from Bonnie
-------Sequential Output-------- ---Sequential Input-- --Random--
-Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks---
Machine MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU /sec %CPU
100 5958 95.8 6358 10.6 3524 11.0 4738 73.5 8654 7.9 67.7 0.8
-------Sequential Output-------- ---Sequential Input-- --Random--
-Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks---
Machine MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU /sec %CPU
100 6015 96.3 6322 10.4 3551 10.5 5354 82.2 11057 10.9 77.7 0.8
-------Sequential Output-------- ---Sequential Input-- --Random--
-Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks---
Machine MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU /sec %CPU
100 5768 92.4 6313 9.9 3549 10.9 5329 82.8 11053 10.7 77.4 0.8
-------Sequential Output-------- ---Sequential Input-- --Random--
-Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks---
Machine MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU /sec %CPU
100 5980 96.0 6316 9.7 3471 10.5 4946 76.4 11054 8.9 78.1 0.7
FreeBSD results from Bonnie
-------Sequential Output-------- ---Sequential Input-- --Random--
-Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks---
Machine MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU /sec %CPU
100 7416 37.8 7334 10.5 7336 19.7 18124 99.9 90798 100.0 1169.3 13.
-------Sequential Output-------- ---Sequential Input-- --Random--
-Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks---
Machine MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU /sec %CPU
100 7255 37.1 7308 10.4 7346 19.8 18114 100.0 90522 100.0 1130.2 12
-------Sequential Output-------- ---Sequential Input-- --Random--
-Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks---
Machine MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU /sec %CPU
100 7383 37.7 7330 10.5 7348 19.6 18107 99.9 90438 100.0 1295.2 14.
-------Sequential Output-------- ---Sequential Input-- --Random--
-Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks---
Machine MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU /sec %CPU
100 7396 37.6 7326 10.4 7336 19.7 18122 100.0 90710 100.0 1290.2 14
However..
Iozone Results From Linux
IOZONE: auto-test mode
MB reclen bytes/sec written bytes/sec read
1 512 4369066 52428799
1 1024 8065969 52428800
1 2048 104857599 52428799
1 4096 104857600 52428800
1 8192 104857599 52428799
2 512 52428800 69905066
2 1024 104857599 69905066
2 2048 104857600 69905066
2 4096 104857599 69905066
2 8192 69905066 69905066
4 512 52428799 59918628
4 1024 83886079 69905066
4 2048 83886080 69905066
4 4096 83886079 83886080
4 8192 83886080 69905066
8 512 52428799 41943040
8 1024 76260072 59918628
8 2048 76260072 69905066
8 4096 83886079 69905066
8 8192 76260072 64527753
16 512 49344752 49344752
16 1024 69905066 59918628
16 2048 76260072 64527753
16 4096 76260072 64527753
16 8192 76260072 62137837
FreeBSD Results from Iozone
IOZONE: auto-test mode
MB reclen bytes/sec written bytes/sec read
1 512 26843545 10324440
1 1024 33554432 8947848
1 2048 44739242 8388608
1 4096 44739242 8388608
1 8192 67108864 7895160
2 512 26843545 10324440
2 1024 33554432 9586980
2 2048 44739242 8947848
2 4096 53687091 8659208
2 8192 53687091 8659208
4 512 26843545 10737418
4 1024 38347922 9586980
4 2048 53687091 8947848
4 4096 59652323 8801162
4 8192 67108864 8521760
8 512 28256363 10129639
8 1024 41297762 8801162
8 2048 51130563 8659208
8 4096 59652323 8454660
8 8192 71582788 8259552
16 512 9629971 31580641
16 1024 10900932 20069940
16 2048 11545611 119304647
16 4096 11864550 143165576
16 8192 11864550 119304647
I couldn't get ByteBench to run on RedHat, so I don't have results to show
you for that benchmark.
L M B E N C H 1 . 0 S U M M A R Y
------------------------------------
Processor, Processes - times in microseconds
--------------------------------------------
Host OS MHz Null Null Simple /bin/sh Mmap 2-proc 8-proc
Syscall Process Process Process lat ctxsw ctxsw
--------- ------------- ---- ------- ------- ------- ------- ---- ------ ------
elijah.el FreeBSD 3.0-R 352 4 0K 2K 4K 18 2 3
linux.eli Linux 2.0.35 352 1 0K 2K 12K 14 3 4
*Local* Communication latencies in microseconds
-----------------------------------------------
Host OS Pipe UDP RPC/ TCP RPC/
UDP TCP
--------- ------------- ------- ------- ------- ------- -------
elijah.el FreeBSD 3.0-R 12 48 111 55 145
linux.eli Linux 2.0.35 13 47 111 77 168
*Local* Communication bandwidths in megabytes/second
----------------------------------------------------
Host OS Pipe TCP File Mmap Bcopy Bcopy Mem Mem
reread reread (libc) (hand) read write
--------- ------------- ---- ---- ------ ------ ------ ------ ---- -----
elijah.el FreeBSD 3.0-R 182 4 125 222 124 123 300 168
linux.eli Linux 2.0.35 166 64 165 244 120 132 298 169
Memory latencies in nanoseconds
(WARNING - may not be correct, check graphs)
--------------------------------------------
Host OS MHz L1 $ L2 $ Main mem Guesses
--------- ------------- --- ---- ---- -------- -------
elijah.el FreeBSD 3.0-R 351 5 42 158
linux.eli Linux 2.0.35 352 5 42 157
All tests were done on out of the box installations from cdrom. The
version of Linux tested was RedHat 5.0, and FreeBSD was 3.0-RELEASE.
The test machine is a 350 MHz asusPbd Motherboard 128 Meg Ram, 1 Fujitsu
10.2 Gig hard drive and a generic N2000 ethernet card Ati 4 MB AGP video card
though video functions were not tested.
Conclusions
In conclusion, I would like to say that decisions concerning an operating
system should be made on several factors. The ones I consider important are
these:
- reliability
- security
- speed
- ease of use
This, of course, is just my opinion. Yours may vary.
We have received a challenge from someone on EfNet IRC in #seattle.
Hopefully we will get better results from that challenge. It's
not definite yet, but here are the details:
- The user from #seattle will configure a Linux system as best as he can, and then we'll run benchmarks on it.
- I will configure the same system using FreeBSD and run benchmarks with my configuration.
If the challenge is conducted, I will post those results in a future
article.
- Rob
Return to Issue #4
|
