Results of LMbench tests on 2.5.49 and 2.5.50

Aniruddha M Marathe (aniruddha.marathe@wipro.com)
Fri, 29 Nov 2002 14:05:16 +0530

Hello,
Here are the results of LMbench tests carried out on kernel 2.5.49 and 2.5.50. Seems that 2.5.49 outperforms 2.5.50 in all the tests that are carried out by LMbench. I did repeated testing on 2.5.50 but I am getting same results. Any clue why it is the case ?

*****************************************************************************************
RESULTS FOR 2.5.49
*****************************************************************************************

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 2.5.49 i686-pc-linux-gnu 790
benchtest Linux 2.5.49 i686-pc-linux-gnu 790
benchtest Linux 2.5.49 i686-pc-linux-gnu 790
benchtest Linux 2.5.49 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 2.5.49 790 0.38 0.59 26 27 1.20 5.95 327 1448 7579
benchtest Linux 2.5.49 790 0.40 0.58 26 27 19 1.20 6.01 301 1440 7525
benchtest Linux 2.5.49 790 0.38 0.61 26 27 22 1.21 5.99 341 1462 7654
benchtest Linux 2.5.49 790 0.38 0.61 26 27 19 1.22 6.01 301 1482 7641

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 2.5.49 1.050 4.5600 14 6.5700 171 36 175
benchtest Linux 2.5.49 1.110 4.4200 14 7.0600 176 39 176
benchtest Linux 2.5.49 1.030 4.3500 14 8.4300 178 41 177
benchtest Linux 2.5.49 1.080 4.6900 14 12 173 34 176

*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 2.5.49 1.050 5.658 16 27 51 116 212 133
benchtest Linux 2.5.49 1.110 5.672 16 28 50 167 147 133
benchtest Linux 2.5.49 1.030 5.676 16 27 116 193 3.0M
benchtest Linux 2.5.49 1.080 5.757 16 27 50 116 146 35M

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 2.5.49 109 46 410 105 466 0.986 3.00000
benchtest Linux 2.5.49 109 46 314 92 462 0.909 3.00000
benchtest Linux 2.5.49 109 47 374 104 467 0.834 3.00000
benchtest Linux 2.5.49 111 47 333 104 472 0.918 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 2.5.49 419 138 23 301 355 124 113 356 170
benchtest Linux 2.5.49 378 141 23 301 353 123 112 353 169
benchtest Linux 2.5.49 514 140 23 302 351 123 111 352 168
benchtest Linux 2.5.49 510 135 23 297 350 123 112 351 168

Memory latencies in nanoseconds - smaller is better
(WARNING - may not be correct, check graphs)
---------------------------------------------------
Host OS Mhz L1 $ L2 $ Main mem Guesses
--------- ------------- ---- ----- ------ -------- -------
benchtest Linux 2.5.49 790 3.794 60 175
benchtest Linux 2.5.49 790 3.797 8.8730 175
benchtest Linux 2.5.49 790 3.803 8.8630 176
benchtest Linux 2.5.49 790 3.795 8.8810 177

*****************************************************************************************
RESULTS FOR 2.5.50
*****************************************************************************************

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 2.5.50 i686-pc-linux-gnu 790
benchtest Linux 2.5.50 i686-pc-linux-gnu 790
benchtest Linux 2.5.50 i686-pc-linux-gnu 790
benchtest Linux 2.5.50 i686-pc-linux-gnu 790
benchtest Linux 2.5.50 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 2.5.50 790 0.45 0.80 27 29 37 1.29 6.53 435 1687 8182
benchtest Linux 2.5.50 790 0.45 0.88 27 29 35 1.31 6.53 354 1669 8228
benchtest Linux 2.5.50 790 0.43 0.81 27 29 36 1.35 6.53 447 1669 8118
benchtest Linux 2.5.50 790 0.45 0.80 27 29 37 1.29 6.53 404 1659 8201
benchtest Linux 2.5.50 790 0.43 0.87 27 29 40 1.29 6.48 458 1688 8213

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 2.5.50 1.200 4.8700 14 5.8200 177 42 178
benchtest Linux 2.5.50 1.280 4.8400 14 7.5200 177 41 179
benchtest Linux 2.5.50 1.370 4.9100 14 6.4000 180 45 181
benchtest Linux 2.5.50 1.350 4.9200 14 5.8300 178 38 181
benchtest Linux 2.5.50 1.320 5.0100 14 12 182 44 183

*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 2.5.50 1.200 8.452 20 34 59 123 154 159
benchtest Linux 2.5.50 1.280 8.532 21 34 59 124 155 159
benchtest Linux 2.5.50 1.370 8.449 21 34 59 123 155 161
benchtest Linux 2.5.50 1.350 8.328 20 34 59 123 154 160
benchtest Linux 2.5.50 1.320 8.630 20 34 59 123 155 159

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 2.5.50 119 56 389 117 600 1.025 4.00000
benchtest Linux 2.5.50 119 56 405 122 609 1.052 4.00000
benchtest Linux 2.5.50 119 56 386 124 610 1.031 4.00000
benchtest Linux 2.5.50 119 56 393 125 615 1.006 4.00000
benchtest Linux 2.5.50 119 56 391 128 618 1.013 4.00000

*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 2.5.50 323 137 22 294 356 124 113 356 171
benchtest Linux 2.5.50 433 125 22 292 352 123 112 351 168
benchtest Linux 2.5.50 285 131 22 294 352 123 112 352 168
benchtest Linux 2.5.50 335 136 22 287 350 122 112 351 168
benchtest Linux 2.5.50 391 129 22 279 348 124 113 349 167

Memory latencies in nanoseconds - smaller is better
(WARNING - may not be correct, check graphs)
---------------------------------------------------
Host OS Mhz L1 $ L2 $ Main mem Guesses
--------- ------------- ---- ----- ------ -------- -------
benchtest Linux 2.5.50 790 3.801 8.8780 175
benchtest Linux 2.5.50 790 3.808 8.8820 176
benchtest Linux 2.5.50 790 3.798 8.8730 176
benchtest Linux 2.5.50 790 3.807 8.8820 176
benchtest Linux 2.5.50 790 3.808 8.8830 177
*****************************************************************************************

regards,
---------------------------------------------------------------
Aniruddha Marathe
Systems Engineer,
4th floor, WIPRO technologies,
53/1, Hosur road,
Madivala,
Bangalore - 560068
Karnataka, India
Phone: +91-80-5502001 extension 5092
E-mail: aniruddha.marathe@wipro.com
---------------------------------------------------------------
-
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/