compare application performance on solaris 8 and Ubuntu

unix

    Next

  • 1. nfsd -a 16
    Hello, I have what seems to be an overload problem, regarding nfs. There is about 7-9 simultaneous nfs mounts from various clients, and there are 16 server threads to process their requests. Still, partiularly one of the clients experience quite a fragmented nfs connection (nfs server xxxxx not responding, still trying) When the nfs deamons are started by nfsd -a 16, it means that 16 concurrent threads are started, over all available transport protocols (typically tcp and udp), according to the manual. However, does anybody know whether each server thread handles all protocols, or whether there is one server thread per protocol. (i.e. in our case there would be 8 tcp server threads and 8 udp server threads.) If only the tcp protocol should be used (all clients are requested to mount by tcp protocol), would it make any difference, load-wise, to start the deamons by explicitly specifying with protocol to use ? (e.g. nfsd -p tcp 16) If not, should I just increase the number of server threads (e.g. to 32) ? Thanks Tor Bye
  • 2. Flow Control
    how would I determine if flow control is turned on for a Solaris Server I have multiple servers using multiple cards. Hme0 ce0 and ge0
  • 3. Looooooooping Server
    Hi, Yesterday I had a box that half way through boot up would panic. Unfortunately eprom auto-boot was set to true. I tried to send a break signal throughout the whole boot cycle but the machine would not respond. This could be because it was a rsc connection ??? The only way I managed to get the box to the ok prompt was to power it off, pull the internal disks and power the box back on. Surely there must be a simpler solution to this, any one have any ideas ? RSC Version: 2.2 RSC Bootmon version: 2.0.0 RSC Firmware version: 2.2.1 Sun-Fire-280R

compare application performance on solaris 8 and Ubuntu

Postby Patricia » Thu, 20 Apr 2006 05:02:55 GMT

Hi all,

I ran the completely same application on Solaris 8 and Linux(Ubuntu)
respectively. But the execution on Linux is significantly faster than
that on Solaris (about twice faster than on Solaris). the hardware
information is:

Solaris: 900 MHz UltraSPARCC III Cu with 1Gb RAM
Linux: Intel(R) Pentium(R) M processor 1.20GHz,512 memory

Could anyone tell me the possible reason for this please?

Thank you very much.


Re: compare application performance on solaris 8 and Ubuntu

Postby Chris Ridd » Thu, 20 Apr 2006 05:09:20 GMT

On 18/4/06 9:02, in article
 XXXX@XXXXX.COM , "Patricia"




Solaris 8 is inefficient on x86? Honestly, without more information it is
hard to give a good answer.

However it is said that Solaris 10 is significantly faster on x86 hardware
than previous versions of Solaris, so you should probably try to compare
recent OS versions.

Cheers,

Chris 


Re: compare application performance on solaris 8 and Ubuntu

Postby tunla » Thu, 20 Apr 2006 05:19:54 GMT




Reason 1 )  Solaris 8 is seven  years  old.

Reason 2 ) the Pentium M 1.2 GHz    is more than twice  as fast   as
the 900MHz USIII
                 IRC  PentiumM  is almost twice the performance  of a "
normal"  Pentium 4 at the
                 same clock speed.


  Solaris 10   has been  Speed optimezed at the system call level , and
is maybe
 twice the throughput of  of solaris 8, I dont have the theoretical
figure right now.

  So  to me  you question  sound like:

  Why is this new formula 1  racing car faster than  that  7 year old
BMW over there.

  You cant  compare systems like that.

   Install solaris 10 on the  Pentium M machine and try the same
workload and then ask you
   question. If you have any at that point.

  Regards 
//Lars


Re: compare application performance on solaris 8 and Ubuntu

Postby ptribble » Thu, 20 Apr 2006 05:46:21 GMT

In article < XXXX@XXXXX.COM >,
	"Patricia" < XXXX@XXXXX.COM > writes:

Well, it's not exactly the same application, is it, unless it's a
script.

What does the application do? What language is it programmed in? How
was it compiled - and with which compiler? Does it generate any I/O
activity?

-- 
-Peter Tribble
L.I.S., University of Hertfordshire -  http://www.**--****.com/ 
 http://www.**--****.com/ ://ptribble.blogspot.com/

Re: compare application performance on solaris 8 and Ubuntu

Postby Casper H.S. Dik » Thu, 20 Apr 2006 06:02:28 GMT

"Patricia" < XXXX@XXXXX.COM > writes:




They're not the same system.

Tell us about the application, the compiler used, the options
used to compile, the version of the operating systems, etc.

Casper
-- 
Expressed in this posting are my opinions.  They are in no way related
to opinions held by my employer, Sun Microsystems.
Statements on Sun products included here are not gospel and may
be fiction rather than truth.

Re: compare application performance on solaris 8 and Ubuntu

Postby Michael Vilain » Thu, 20 Apr 2006 17:18:14 GMT

In article < XXXX@XXXXX.COM >,




Funny, you don't look trollish...

-- 
DeeDee, don't press that button!  DeeDee!  NO!  Dee...




Re: compare application performance on solaris 8 and Ubuntu

Postby Colin B. » Fri, 21 Apr 2006 00:15:18 GMT



Could it possibly be that you're testing different things on totally
different hardware?

On the Sun side, you've got a ~5 year old processor with a seven year old
OS. How old is the Pentium M 1.2GHz, and how recent is Ubuntu? A whole lot
more recent, I can tell you that much.

Try installing Solaris 10/x86 on the Pentium machine, use the same compiler
for the app in both environments, and then tell us what the speed is like.
For bonus points, try the test with both gcc and Sun's compiler suite; For
further bonus points, try installing Linux on the Sun machine, and see what
the performance is like.

Colin

Similar Threads:

1.compare application performance on solaris 8 and Ubuntu

2.How to compare System Performance and Disk IO Performance

Hi AIX User


How to compare System  Performance and Disk IO Performance ?

Old System : p5 9117-570, 4 x 1.65GHz
New System : p5 9117-MMA , 4 x 4.2GHz

We want compare  System  Performance and Disk IO Performance


I just have
Multi-user Performance (rPerf, SPEC CPU2006)
19.66 for Old System
38.76 for New System


moonhk
GMT+8

3.How to compare System Performance and Disk IO Performance

On Apr 10, 9:49 am, moonhkt < XXXX@XXXXX.COM > wrote:
> Hi AIX User
>
> How to compare System  Performance and Disk IO Performance ?
>
> Old System : p5 9117-570, 4 x 1.65GHz
> New System : p5 9117-MMA , 4 x 4.2GHz
Looks like you answered your first question by yourself. The second
one is on a first approach  simply not related to the named hardware.

Or to say in other words:
Disk I/O is determined by the Disk Type, Disk Amount, Disk
Configuration (Raid Type ), Connection Type ( FC,SSA,SCSI,iSCSI,SAS ),
Conection Speed , HBA , Filesystem Type and configuration
(JFS,JFS2,GPFS) and ..... the CPU

cheers
Hajo

4.10 reasons to love Ubuntu, Performance, performance performance

5.[News] [Linux] Ubuntu Linux Expands in Europe, Sidux (Debian) and Mint Linux (Ubuntu) Compared

6. Sun LDAP server on Solaris 10 compared to Solaris 8

7. Comparing Sparc and Intel PC performance

8. [News] Why Ubuntu Linux Beats Windows Vista, Recommended Applications for Ubuntu



Return to unix

 

Who is online

Users browsing this forum: No registered users and 15 guest