[BENCHMARK] 2.5.55 Lmbench performance.

Aniruddha M Marathe (aniruddha.marathe@wipro.com)
Thu, 9 Jan 2003 16:52:25 +0530


Hi,
Here is a comparison of results of 2.5.55 and 2.5.54. The figures in the table below indicate median of 5 repetitions of tests. This result doesn't have many
Differences than the previous one.

2.5.55 2.5.54
==============================================================================
Processor, Processes - times in microseconds - smaller is better

1. Null call 0.41 0.44
2. Null I/O 0.58 0.82
3. stat 25 29
4. open close 26 30
5. sh proc 7117 7872
6. exec proc 1285 1528
------------------------------------------------------------------------------
Context switching - times in microseconds - smaller is better

1. 2p/0K ctxsw 1.100 1.560
------------------------------------------------------------------------------
*Local* Communication latencies in microseconds - smaller is better
1. PIPE 5.626 8.426
2. AF UNIX 13 19
3. UDP 24 30
4. TCP connection 126 146
------------------------------------------------------------------------------
File & VM system latencies in microseconds - smaller is better
1. MMAP latency 472 599
------------------------------------------------------------------------------
*Local* Communication bandwidth in MB/s - bigger is better
1. PIPE 554 634
==============================================================================

Full result
****************************************************************************
Lmbench result
kernel 2.5.55
****************************************************************************

L M B E N C H 2 . 0 S U M M A R Y
------------------------------------
(Alpha software, do not distribute)

Basic system parameters
----------------------------------------------------
Host OS Description Mhz

--------- ------------- ----------------------- ----
benchtest Linux 2.5.55 i686-pc-linux-gnu 790
benchtest Linux 2.5.55 i686-pc-linux-gnu 790
benchtest Linux 2.5.55 i686-pc-linux-gnu 790
benchtest Linux 2.5.55 i686-pc-linux-gnu 790
benchtest Linux 2.5.55 i686-pc-linux-gnu 790

Processor, Processes - times in microseconds - smaller is better
----------------------------------------------------------------
Host OS Mhz null null open selct sig sig fork exec sh
call I/O stat clos TCP inst hndl proc proc proc
--------- ------------- ---- ---- ---- ---- ---- ----- ---- ---- ---- ---- ----
benchtest Linux 2.5.55 790 0.38 0.58 25 26 20 1.20 4.70 303 1354 7228
benchtest Linux 2.5.55 790 0.41 0.58 25 26 18 1.22 4.93 239 1265 7047
benchtest Linux 2.5.55 790 0.38 0.57 25 26 21 1.20 4.92 253 1285 7061
benchtest Linux 2.5.55 790 0.41 0.58 25 26 19 1.22 4.99 291 1304 7117
benchtest Linux 2.5.55 790 0.41 0.56 25 26 17 1.20 4.93 237 1252 7080

Context switching - times in microseconds - smaller is better
-------------------------------------------------------------
Host OS 2p/0K 2p/16K 2p/64K 8p/16K 8p/64K 16p/16K 16p/64K
ctxsw ctxsw ctxsw ctxsw ctxsw ctxsw ctxsw
--------- ------------- ----- ------ ------ ------ ------ ------- -------
benchtest Linux 2.5.55 1.040 4.2600 15 4.7500 175 37 173
benchtest Linux 2.5.55 1.120 4.3800 14 6.4800 173 38 175
benchtest Linux 2.5.55 1.120 4.4500 14 8.0100 178 41 178
benchtest Linux 2.5.55 1.100 4.3800 14 5.0000 179 38 176
benchtest Linux 2.5.55 1.100 4.3700 14 6.8700 176 33 176

*Local* Communication latencies in microseconds - smaller is better
-------------------------------------------------------------------
Host OS 2p/0K Pipe AF UDP RPC/ TCP RPC/ TCP
ctxsw UNIX UDP TCP conn
--------- ------------- ----- ----- ---- ----- ----- ----- ----- ----
benchtest Linux 2.5.55 1.040 5.489 13 24 47 69 97 126
benchtest Linux 2.5.55 1.120 5.626 14 24 47 70 97 126
benchtest Linux 2.5.55 1.120 5.775 14 24 47 69 97 126
benchtest Linux 2.5.55 1.100 5.692 13 24 47 69 97 126
benchtest Linux 2.5.55 1.100 5.612 13 24 47 69 97 126

File & VM system latencies in microseconds - smaller is better
--------------------------------------------------------------
Host OS 0K File 10K File Mmap Prot Page
Create Delete Create Delete Latency Fault Fault
--------- ------------- ------ ------ ------ ------ ------- ----- -----
benchtest Linux 2.5.55 106 43 317 94 472 0.843 3.00000
benchtest Linux 2.5.55 106 44 304 96 481 0.842 3.00000
benchtest Linux 2.5.55 107 44 316 96 470 0.872 3.00000
benchtest Linux 2.5.55 107 44 302 99 470 0.849 3.00000
benchtest Linux 2.5.55 106 44 317 96 479 0.890 3.00000

*Local* Communication bandwidths in MB/s - bigger is better
-----------------------------------------------------------
Host OS Pipe AF TCP File Mmap Bcopy Bcopy Mem Mem
UNIX reread reread (libc) (hand) read write
--------- ------------- ---- ---- ---- ------ ------ ------ ------ ---- -----
benchtest Linux 2.5.55 634 133 24 308 356 124 113 356 171
benchtest Linux 2.5.55 248 135 25 286 353 123 112 354 170
benchtest Linux 2.5.55 608 129 25 304 353 123 112 353 169
benchtest Linux 2.5.55 667 138 25 301 353 123 112 353 169
benchtest Linux 2.5.55 681 108 25 295 351 124 113 351 169

Memory latencies in nanoseconds - smaller is better
(WARNING - may not be correct, check graphs)
---------------------------------------------------
Host OS Mhz L1 $ L2 $ Main mem Guesses
--------- ------------- ---- ----- ------ -------- -------
benchtest Linux 2.5.55 790 3.796 8.8720 174
benchtest Linux 2.5.55 790 3.798 8.8650 175
benchtest Linux 2.5.55 790 3.793 8.8730 175
benchtest Linux 2.5.55 790 3.793 8.8640 176
benchtest Linux 2.5.55 790 3.803 8.8670 176

Aniruddha Marathe
WIPRO Technologies, India
aniruddha.marathe@wipro.com
+91-80-5502001 to 2008 extn 5092

-
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/