[NCLUG] July 1, 2003 talk - Linux System Performance

jbass at dmsd.com jbass at dmsd.com
Wed Jun 25 14:00:42 MDT 2003


Thanks Mike!

Looks like there is a counter overflow problem in the test for very
fast machines :(

Not entirely a suprise for code that has existed since PDP-11 and Vax
days.

Will look into it, and we can try this again.

John

	Thus spake jbass at dmsd.com:

	 [trimmed]

	> In preparation for the talk I seek a broad range of current system
	> data on memory/cache/TLB performance, as well as getting people to
	> think about the issues before the talk. Please compile and run the
	> memory/cache/TLB benchmark below and return the results to me with
	> a description of the system including CPU's, motherboard, chipset,
	> memory type and size. You might find it interesting to run the test
	> both single user and multiuser from an xterm. Other interesting
	> tests are with various I/O active in the background (disk, network,
	> etc) and multiple sessions on machines with two or more processors.

	 Here's the quick results of mbench runs on 3 of my machines:
	 
	 http://www.verinet.com/~mike/mbench.html

	-- 
	   Mike Loseke    | ... Logically incoherent, semantically
	 mike at verinet.com | incomprehensible, and legally ... impeccable!



More information about the NCLUG mailing list