Attached are a set of performance results comparing 2.6.11-rc5 under
RedHat REL4 on power5 1.5Ghz, 4cpus (smt-enabled=0), 16G RAM, with an
ext2 filesystem. -nostack is a kernel with selinux and capability.
stack is a kernel with stacker, selinux, and cap_stack.
dbench (run as dbench -c client_plain.txt 4, three times)
nostack: 872.617 882.029 870.968
stack: 799.608 798.028 800.122
hackbench (run as ./hackbench 100)
nostack: 5.064
stack: 6.721
unixbench (full report files are attached)
nostack: 494.3
stack: 447.1
In particular stacker performance impact on tests like the unixbench
file read and copy tests seem excessive.
I'm not sure whether to proceed with apachebench, or many runs of
lmbench. It seems lmbench may be more helpful in pointing toward the
culprit.
thanks,
-serge
On Wed, 2005-03-09 at 11:14 -0500, Stephen Smalley wrote:
> In the past, we haven't found kernel compile benchmark to be very
> revealing for SELinux performance analysis. dbench results would be of
> interest. More generally, you might want to repeat the tests done for
> the AVC RCU work, see
>
> http://marc.theaimsgroup.com/?l=linux-kernel&m=110054824004161&w=2
>
--
Serge Hallyn <serue@private>
BYTE UNIX Benchmarks (Version 4.1.0)
System -- Linux techman 2.6.11-rc5 #1 SMP Fri Mar 11 15:55:06 CST 2005 ppc64 ppc64 ppc64 GNU/Linux
Start Benchmark Run: Sun Mar 13 01:17:07 CST 2005
1 interactive users.
01:17:07 up 2:30, 1 user, load average: 0.00, 0.00, 0.00
lrwxrwxrwx 1 root root 4 Mar 10 22:54 /bin/sh -> bash
/bin/sh: symbolic link to `bash'
/dev/sdc4 18340280 6360604 11048044 37% /
Dhrystone 2 using register variables 4552324.0 lps (10.0 secs, 10 samples)
Double-Precision Whetstone 663.3 MWIPS (9.9 secs, 10 samples)
System Call Overhead 611750.8 lps (10.0 secs, 10 samples)
Pipe Throughput 424185.0 lps (10.0 secs, 10 samples)
Pipe-based Context Switching 104960.8 lps (10.0 secs, 10 samples)
Process Creation 5343.4 lps (30.0 secs, 3 samples)
Execl Throughput 1939.2 lps (29.9 secs, 3 samples)
File Read 1024 bufsize 2000 maxblocks 669563.0 KBps (30.0 secs, 3 samples)
File Write 1024 bufsize 2000 maxblocks 470043.0 KBps (30.0 secs, 3 samples)
File Copy 1024 bufsize 2000 maxblocks 273748.0 KBps (30.0 secs, 3 samples)
File Read 256 bufsize 500 maxblocks 193167.0 KBps (30.0 secs, 3 samples)
File Write 256 bufsize 500 maxblocks 130354.0 KBps (30.0 secs, 3 samples)
File Copy 256 bufsize 500 maxblocks 76669.0 KBps (30.0 secs, 3 samples)
File Read 4096 bufsize 8000 maxblocks 1871012.0 KBps (30.0 secs, 3 samples)
File Write 4096 bufsize 8000 maxblocks 1416786.0 KBps (30.0 secs, 3 samples)
File Copy 4096 bufsize 8000 maxblocks 794053.0 KBps (30.0 secs, 3 samples)
Shell Scripts (1 concurrent) 2773.5 lpm (60.0 secs, 3 samples)
Shell Scripts (8 concurrent) 953.2 lpm (60.0 secs, 3 samples)
Shell Scripts (16 concurrent) 514.9 lpm (60.0 secs, 3 samples)
Arithmetic Test (type = short) 365077.2 lps (10.0 secs, 3 samples)
Arithmetic Test (type = int) 386323.4 lps (10.0 secs, 3 samples)
Arithmetic Test (type = long) 386298.6 lps (10.0 secs, 3 samples)
Arithmetic Test (type = float) 192247.1 lps (10.0 secs, 3 samples)
Arithmetic Test (type = double) 192048.8 lps (10.0 secs, 3 samples)
Arithoh 4379740.0 lps (10.0 secs, 3 samples)
C Compiler Throughput 722.9 lpm (60.0 secs, 3 samples)
Dc: sqrt(2) to 99 decimal places 82060.0 lpm (30.0 secs, 3 samples)
Recursion Test--Tower of Hanoi 45530.6 lps (20.0 secs, 3 samples)
INDEX VALUES
TEST BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 4552324.0 390.1
Double-Precision Whetstone 55.0 663.3 120.6
Execl Throughput 43.0 1939.2 451.0
File Copy 1024 bufsize 2000 maxblocks 3960.0 273748.0 691.3
File Copy 256 bufsize 500 maxblocks 1655.0 76669.0 463.3
File Copy 4096 bufsize 8000 maxblocks 5800.0 794053.0 1369.1
Pipe Throughput 12440.0 424185.0 341.0
Process Creation 126.0 5343.4 424.1
Shell Scripts (8 concurrent) 6.0 953.2 1588.7
System Call Overhead 15000.0 611750.8 407.8
=========
FINAL SCORE 494.3
BYTE UNIX Benchmarks (Version 4.1.0)
System -- Linux techman 2.6.11-rc5-stack #2 SMP Fri Mar 11 16:17:52 CST 2005 ppc64 ppc64 ppc64 GNU/Linux
Start Benchmark Run: Sat Mar 12 03:53:52 CST 2005
1 interactive users.
03:53:52 up 11 min, 1 user, load average: 0.00, 0.00, 0.00
lrwxrwxrwx 1 root root 4 Mar 10 22:54 /bin/sh -> bash
/bin/sh: symbolic link to `bash'
/dev/sdc4 18340280 6360580 11048068 37% /
Dhrystone 2 using register variables 4551679.9 lps (10.0 secs, 10 samples)
Double-Precision Whetstone 662.2 MWIPS (9.9 secs, 10 samples)
System Call Overhead 610533.5 lps (10.0 secs, 10 samples)
Pipe Throughput 304645.4 lps (10.0 secs, 10 samples)
Pipe-based Context Switching 92837.0 lps (10.0 secs, 10 samples)
Process Creation 5263.2 lps (30.0 secs, 3 samples)
Execl Throughput 1870.7 lps (29.7 secs, 3 samples)
File Read 1024 bufsize 2000 maxblocks 509733.0 KBps (30.0 secs, 3 samples)
File Write 1024 bufsize 2000 maxblocks 390972.0 KBps (30.0 secs, 3 samples)
File Copy 1024 bufsize 2000 maxblocks 223054.0 KBps (30.0 secs, 3 samples)
File Read 256 bufsize 500 maxblocks 140702.0 KBps (30.0 secs, 3 samples)
File Write 256 bufsize 500 maxblocks 106027.0 KBps (30.0 secs, 3 samples)
File Copy 256 bufsize 500 maxblocks 60912.0 KBps (30.0 secs, 3 samples)
File Read 4096 bufsize 8000 maxblocks 1528229.0 KBps (30.0 secs, 3 samples)
File Write 4096 bufsize 8000 maxblocks 1211524.0 KBps (30.0 secs, 3 samples)
File Copy 4096 bufsize 8000 maxblocks 668225.0 KBps (30.0 secs, 3 samples)
Shell Scripts (1 concurrent) 2737.9 lpm (60.0 secs, 3 samples)
Shell Scripts (8 concurrent) 942.2 lpm (60.0 secs, 3 samples)
Shell Scripts (16 concurrent) 508.6 lpm (60.0 secs, 3 samples)
Arithmetic Test (type = short) 364919.8 lps (10.0 secs, 3 samples)
Arithmetic Test (type = int) 386182.5 lps (10.0 secs, 3 samples)
Arithmetic Test (type = long) 386295.8 lps (10.0 secs, 3 samples)
Arithmetic Test (type = float) 192557.2 lps (10.0 secs, 3 samples)
Arithmetic Test (type = double) 191979.7 lps (10.0 secs, 3 samples)
Arithoh 4379898.6 lps (10.0 secs, 3 samples)
C Compiler Throughput 713.2 lpm (60.0 secs, 3 samples)
Dc: sqrt(2) to 99 decimal places 79953.3 lpm (30.0 secs, 3 samples)
Recursion Test--Tower of Hanoi 45522.6 lps (20.0 secs, 3 samples)
INDEX VALUES
TEST BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 4551679.9 390.0
Double-Precision Whetstone 55.0 662.2 120.4
Execl Throughput 43.0 1870.7 435.0
File Copy 1024 bufsize 2000 maxblocks 3960.0 223054.0 563.3
File Copy 256 bufsize 500 maxblocks 1655.0 60912.0 368.0
File Copy 4096 bufsize 8000 maxblocks 5800.0 668225.0 1152.1
Pipe Throughput 12440.0 304645.4 244.9
Process Creation 126.0 5263.2 417.7
Shell Scripts (8 concurrent) 6.0 942.2 1570.3
System Call Overhead 15000.0 610533.5 407.0
=========
FINAL SCORE 447.1
This archive was generated by hypermail 2.1.3 : Wed Mar 16 2005 - 13:30:37 PST