i'm not sure how reliable/useful the lmbench is, but i noticed that
page fault latencies of pre5 and pre6 are unusually high. i'm new at
all of this, so i don't know if this is an acceptable trade-off. I also
noticed that the pre4-ac3 showed exceptionally low local communication
latencies -- again, i don't know the real meaning of this.
brent
*personal request* (reply to damonbrent@earthlink.net to lighten
mail-list traffic)
i have a fair knowledge of C. i would like to eventually be able
to contribute to this development (or atleast entertain myself with
experiments). what is the best way to attack this learning curve?
where to best start? Thanks!
===============================================================================
(in this order)
2.2.0pre4-ac3
2.2.0pre5
2.2.0pre5-arca-VM11
2.2.0pre6
L M B E N C H 1 . 9 S U M M A R Y
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 inst hndl proc
proc proc
--------- ------------- ---- ---- ---- ---- ---- ----- ---- ---- ----
---- ----
i586-linu Linux 2.2.0-p 299 0.7 1.1 7 10 0.06K 1.9 3 0.5K
3K 16K
i586-linu Linux 2.2.0-p 299 0.7 1.1 8 10 0.06K 1.9 3 0.6K
3K 17K
i586-linu Linux 2.2.0-p 299 0.7 1.1 8 10 0.07K 2.0 3 0.6K
3K 17K
i586-linu Linux 2.2.0-p 299 0.7 1.3 7 10 0.08K 1.9 3 0.6K
3K 17K
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
--------- ------------- ----- ------ ------ ------ ------ -------
-------
i586-linu Linux 2.2.0-p 4 17 146 43 161 45 230
i586-linu Linux 2.2.0-p 3 16 162 53 200 59 243
i586-linu Linux 2.2.0-p 5 16 161 43 199 47 225
i586-linu Linux 2.2.0-p 6 15 155 43 203 56 234
*Local* Communication latencies in microseconds - smaller is better
-------------------------------------------------------------------
Host OS 2p/0K Pipe AF UDP RPC/ TCP RPC/ TCP
ctxsw UNIX UDP TCP conn
--------- ------------- ----- ----- ---- ----- ----- ----- ----- ----
i586-linu Linux 2.2.0-p 4 14 28 58 130
i586-linu Linux 2.2.0-p 3 18 39 64 160
i586-linu Linux 2.2.0-p 5 15 31 56 149
i586-linu Linux 2.2.0-p 6 16 33 67 152
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
--------- ------------- ------ ------ ------ ------ ------- -----
-----
i586-linu Linux 2.2.0-p 17 2 39 4 8253 2
1.2K
i586-linu Linux 2.2.0-p 7 2 43 4 8679 2
5.1K
i586-linu Linux 2.2.0-p 17 2 42 4 8463 2
1.3K
i586-linu Linux 2.2.0-p 17 2 40 4 8563 2
8.5K
*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
--------- ------------- ---- ---- ---- ------ ------ ------ ------ ----
-----
i586-linu Linux 2.2.0-p 116 59 -1 66 193 65 66
193 100
i586-linu Linux 2.2.0-p 114 58 -1 71 194 69 70
193 104
i586-linu Linux 2.2.0-p 76 53 29 70 193 64 65
193 100
i586-linu Linux 2.2.0-p 98 55 30 71 193 68 70
193 106
Memory latencies in nanoseconds - smaller is better
(WARNING - may not be correct, check graphs)
---------------------------------------------------
Host OS Mhz L1 $ L2 $ Main mem Guesses
--------- ------------- --- ---- ---- -------- -------
i586-linu Linux 2.2.0-p 299 6 80 157
i586-linu Linux 2.2.0-p 299 6 81 158
i586-linu Linux 2.2.0-p 299 6 92 158
i586-linu Linux 2.2.0-p 299 6 94 157
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/