Re: Comparing Java Web Server performance
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
But thats exactly my point - how can I compare the performance!?!
Our 3.2 GHz Xeons seem to be outperforming our Dual 248 Opterons,
but I don't know how I can really compare this...
Is idle a good way of comparing this?
take the values from /proc/stat twice, (the first line)
time1: cpu 343278033 23768 19953941 1442746591 4716086 1372872 5551609
time2: cpu 343278589 23768 19953979 1442748407 4716092 1372875 5551621
delta: 556 0 38 1816 6 3 12
total = 2431
idle% = 1816/2431 * 100
= 74.70%
???
Any other suggestions?
Regards
Andrew
On Dec 1, 2005, at 3:10 PM, Chris Wagner wrote:
I would ignore the hyperthreading and go with the Opterons.
Depending on
the process's thread patterns the HT can make a difference or be
totally
useless. I like to use this analogy. Threads execute by putting
their
"people" in a train of cars and sending it for a loop through the
roller
coaster. Hyperthreading lets u put somebody else's people in any
unused
cars. It's too much of a crap shoot for me and who knows what
other kind of
overhead it's producing. Not to mention how it complicates
figuring cpu
usage. The only way I can think to compare them is to do
benchmarking with
and without the HT turned on. It's really application specific.
At 01:52 PM 12/1/2005 +0100, Andrew Miehs wrote:
One problem that I am having though is trying to find a good way to
compare 2 proc Opteron machines
vs 2 proc Xeons with Hyperthreading....
--
REMEMBER THE WORLD TRADE CENTER ---=< WTC 911 >=--
"...ne cede malis"
00000100
--
To UNSUBSCRIBE, email to debian-isp-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact
listmaster@lists.debian.org
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (Darwin)
iD8DBQFDjwrJW126qUNSzvURAgmMAJ0Uzq2yAWqWhCl0GmnUrRmL6BtCxACgjK/u
+J2W8U9r3zSFPxPUkJuo2RE=
=X7bS
-----END PGP SIGNATURE-----
Reply to: