|
Replies:
18
-
Last Post:
Aug 28, 2008 10:54 AM
by: esaxe
|
Threads:
[
Previous
|
Next
]
|
|
Posts:
3
From:
Registered:
6/18/07
|
|
|
|
Project proposal: "Solaris Enhancements for
AMD-based Platforms"
Posted:
Oct 30, 2007 1:53 PM
|
|
I would like to propose creation of a new project titled "Solaris
Enhancements for AMD-based Platforms".
The project will address various features that are specific to platforms
based on AMD processors, such as
- IOMMU support
- NUMA topology, particularly how it affects IO performance
- Observability (performance counters, instruction-based
sampling)
- Power management
- RAS features
- New instruction support, new CPUID features
Since what's described above covers fairly diverse range of subjects,
the proposed project will serve as an umbrella for sub-projects, each of
them covering a particular area related to improving Solaris behavior on
systems built around AMD processors (as well as chipsets and graphics
components)
I think this project would be of interest to a number of OpenSolaris
communities but I am asking the performance community for sponsorship as
it appears to be the most relevant.
Thanks.
-boris
_______________________________________________
perf-discuss mailing list
perf-discuss at opensolaris dot org
|
|
|
Posts:
198
From:
US
Registered:
3/9/05
|
|
|
|
Re: Project proposal: "Solaris Enhancements for
AMD-based Platforms"
Posted:
Oct 30, 2007 2:31 PM
in response to: bostrovs
|
|
Ostrovsky, Boris wrote:
> I would like to propose creation of a new project titled "Solaris
> Enhancements for AMD-based Platforms".
>
> The project will address various features that are specific to platforms
> based on AMD processors, such as
> - IOMMU support
> - NUMA topology, particularly how it affects IO performance
> - Observability (performance counters, instruction-based
> sampling)
> - Power management
> - RAS features
> - New instruction support, new CPUID features
>
> Since what's described above covers fairly diverse range of subjects,
> the proposed project will serve as an umbrella for sub-projects, each of
> them covering a particular area related to improving Solaris behavior on
> systems built around AMD processors (as well as chipsets and graphics
> components)
>
> I think this project would be of interest to a number of OpenSolaris
> communities but I am asking the performance community for sponsorship as
> it appears to be the most relevant.
>
+1 from me.
-Eric
_______________________________________________
perf-discuss mailing list
perf-discuss at opensolaris dot org
|
|
|
|
|
Posts:
44
From:
US
Registered:
3/9/05
|
|
|
|
Re: Project proposal: "Solaris Enhancements for
AMD-based Platforms"
Posted:
Oct 30, 2007 3:35 PM
in response to: bostrovs
|
|
Ostrovsky, Boris wrote:
> I would like to propose creation of a new project titled "Solaris
> Enhancements for AMD-based Platforms".
>
> The project will address various features that are specific to platforms
> based on AMD processors, such as
> - IOMMU support
> - NUMA topology, particularly how it affects IO performance
> - Observability (performance counters, instruction-based
> sampling)
> - Power management
> - RAS features
> - New instruction support, new CPUID features
>
> Since what's described above covers fairly diverse range of subjects,
> the proposed project will serve as an umbrella for sub-projects, each of
> them covering a particular area related to improving Solaris behavior on
> systems built around AMD processors (as well as chipsets and graphics
> components)
>
> I think this project would be of interest to a number of OpenSolaris
> communities but I am asking the performance community for sponsorship as
> it appears to be the most relevant.
>
+1 from me too.
Jonathan
_______________________________________________
perf-discuss mailing list
perf-discuss at opensolaris dot org
|
|
|
|
|
Posts:
3,657
From:
DE
Registered:
6/16/05
|
|
|
|
Re: Project proposal: "Solaris Enhancements
forAMD-based Platforms"
Posted:
Oct 31, 2007 8:13 PM
in response to: bostrovs
|
|
"Ostrovsky, Boris" wrote:
> I would like to propose creation of a new project titled "Solaris
> Enhancements for AMD-based Platforms".
>
> The project will address various features that are specific to platforms
> based on AMD processors, such as
> - IOMMU support
> - NUMA topology, particularly how it affects IO performance
> - Observability (performance counters, instruction-based
> sampling)
> - Power management
> - RAS features
> - New instruction support, new CPUID features
>
> Since what's described above covers fairly diverse range of subjects,
> the proposed project will serve as an umbrella for sub-projects, each of
> them covering a particular area related to improving Solaris behavior on
> systems built around AMD processors (as well as chipsets and graphics
> components)
>
> I think this project would be of interest to a number of OpenSolaris
> communities but I am asking the performance community for sponsorship as
> it appears to be the most relevant.
+1 from me...
... it may be nice to see the new AMD64 128bit floating-point stuff
supported in Solaris... may be a usefull for HPC customers... :-)
----
Bye,
Roland
--
__ . . __
(o.\ \/ /.o) roland dot mainz at nrubsig dot org
\__\/\/__/ MPEG specialist, C&&JAVA&&Sun&&Unix programmer
/O /==\ O\ TEL +49 641 7950090
(;O/ \/ \O;)
_______________________________________________
perf-discuss mailing list
perf-discuss at opensolaris dot org
|
|
|
|
|
Posts:
3
From:
Registered:
6/18/07
|
|
|
|
Re: Project proposal: "Solaris Enhancements
forAMD-based Platforms"
Posted:
Nov 1, 2007 8:09 AM
in response to: gisburn
|
|
Roland Mainz wrote:
>
>
> ... it may be nice to see the new AMD64 128bit floating-point stuff
> supported in Solaris... may be a usefull for HPC customers... :-)
>
There is no need to change anything in software to take advantage of
wider path (128bit) in FP unit,
if that's what you are referring to.
There are some gotchas such as data alignment (though I think CR 6525798
addressed this) and
dependencies but in general you will get performance boost without need
to recompile.
-boris
_______________________________________________
perf-discuss mailing list
perf-discuss at opensolaris dot org
|
|
|
|
|
Posts:
3,657
From:
DE
Registered:
6/16/05
|
|
|
|
Re: Project proposal:"Solaris Enhancements
forAMD-based Platforms"
Posted:
Nov 1, 2007 3:33 PM
in response to: bostrovs
|
|
Boris Ostrovsky wrote:
> Roland Mainz wrote:
> > ... it may be nice to see the new AMD64 128bit floating-point stuff
> > supported in Solaris... may be a usefull for HPC customers... :-)
>
> There is no need to change anything in software to take advantage of
> wider path (128bit) in FP unit,
> if that's what you are referring to.
I was referring to something I read recently in a german computer
magazine tha the upcoming AMD CPUs have some kind of special 128 FP
instruction stuff (disclimer: I have no clue what exactly te stuff was)
...
... if this stuff includes some kind of special instructions it may be
nice to reflect this via matching flags in the $ /usr/bin/isalist #
output.
> There are some gotchas such as data alignment (though I think CR 6525798
> addressed this) and
> dependencies but in general you will get performance boost without need
> to recompile.
Ok... but see above...
----
Bye,
Roland
--
__ . . __
(o.\ \/ /.o) roland dot mainz at nrubsig dot org
\__\/\/__/ MPEG specialist, C&&JAVA&&Sun&&Unix programmer
/O /==\ O\ TEL +49 641 7950090
(;O/ \/ \O;)
_______________________________________________
perf-discuss mailing list
perf-discuss at opensolaris dot org
|
|
|
|
|
Rayson Ho
rayrayson@gmail.com
|
|
|
|
Re: Project proposal:"Solaris Enhancements
forAMD-based Platforms"
Posted:
Nov 1, 2007 3:52 PM
in response to: gisburn
|
|
On Nov 1, 2007 5:33 PM, Roland Mainz <roland dot mainz at nrubsig dot org> wrote:
> I was referring to something I read recently in a german computer
> magazine tha the upcoming AMD CPUs have some kind of special 128 FP
> instruction stuff (disclimer: I have no clue what exactly te stuff was)
It may be referring to the SSE5 instruction set...
BTW, Sun engineers continue to add new features in the compiler for AMD:
http://blogs.sun.com/tatkar/entry/sun_studio_patch_supports_barcelona
Rayson
> ...
> ... if this stuff includes some kind of special instructions it may be
> nice to reflect this via matching flags in the $ /usr/bin/isalist #
> output.
>
> > There are some gotchas such as data alignment (though I think CR 6525798
> > addressed this) and
> > dependencies but in general you will get performance boost without need
> > to recompile.
>
> Ok... but see above...
>
> ----
>
> Bye,
> Roland
>
> --
> __ . . __
> (o.\ \/ /.o) roland dot mainz at nrubsig dot org
> \__\/\/__/ MPEG specialist, C&&JAVA&&Sun&&Unix programmer
> /O /==\ O\ TEL +49 641 7950090
> (;O/ \/ \O;)
> _______________________________________________
>
> perf-discuss mailing list
> perf-discuss at opensolaris dot org
>
_______________________________________________
perf-discuss mailing list
perf-discuss at opensolaris dot org
|
|
|
|
|
Posts:
3
From:
Registered:
6/18/07
|
|
|
|
Re: Project proposal:"Solaris Enhancements
forAMD-based Platforms"
Posted:
Nov 2, 2007 3:09 PM
in response to: Rayson Ho
|
|
> -----Original Message-----
> From: perf-discuss-bounces at opensolaris dot org [mailto:perf-discuss-
> bounces at opensolaris dot org] On Behalf Of Rayson Ho
> Sent: Thursday, November 01, 2007 6:52 PM
> To: perf-discuss at opensolaris dot org
> Subject: Re: [perf-discuss] Project proposal:"Solaris Enhancements
forAMD-
> based Platforms"
>
> On Nov 1, 2007 5:33 PM, Roland Mainz <roland dot mainz at nrubsig dot org> wrote:
> > I was referring to something I read recently in a german computer
> > magazine tha the upcoming AMD CPUs have some kind of special 128 FP
> > instruction stuff (disclimer: I have no clue what exactly te stuff
was)
>
> It may be referring to the SSE5 instruction set...
This is a brief description of SSE5 instruction set.
http://developer.amd.com/sse5.jsp
>
> BTW, Sun engineers continue to add new features in the compiler for
AMD:
> http://blogs.sun.com/tatkar/entry/sun_studio_patch_supports_barcelona
>
> Rayson
>
>
>
>
> > ...
> > ... if this stuff includes some kind of special instructions it may
be
> > nice to reflect this via matching flags in the $ /usr/bin/isalist #
> > output.
That's a good idea. isalist doesn't report any of SSE* now. getisax()
does.
-boris
> >
> > > There are some gotchas such as data alignment (though I think CR
> 6525798
> > > addressed this) and
> > > dependencies but in general you will get performance boost without
> need
> > > to recompile.
> >
> > Ok... but see above...
> >
_______________________________________________
perf-discuss mailing list
perf-discuss at opensolaris dot org
|
|
|
|
|
Posts:
3,657
From:
DE
Registered:
6/16/05
|
|
|
|
Re: Project proposal:"Solaris
EnhancementsforAMD-based Platforms"
Posted:
Nov 2, 2007 4:23 PM
in response to: bostrovs
|
|
"Ostrovsky, Boris" wrote:
> > On Nov 1, 2007 5:33 PM, Roland Mainz <roland dot mainz at nrubsig dot org> wrote:
> > > I was referring to something I read recently in a german computer
> > > magazine tha the upcoming AMD CPUs have some kind of special 128 FP
> > > instruction stuff (disclimer: I have no clue what exactly te stuff
> was)
> >
> > It may be referring to the SSE5 instruction set...
>
> This is a brief description of SSE5 instruction set.
>
> http://developer.amd.com/sse5.jsp
Thanks! :-)
> > BTW, Sun engineers continue to add new features in the compiler for
> AMD:
> > http://blogs.sun.com/tatkar/entry/sun_studio_patch_supports_barcelona
> >
> > > ...
> > > ... if this stuff includes some kind of special instructions it may
> be
> > > nice to reflect this via matching flags in the $ /usr/bin/isalist #
> > > output.
>
> That's a good idea. isalist doesn't report any of SSE* now. getisax()
> does.
Umpf... the problem is that something like |getisax()| isn't used by
"isaexec" (/usr/lib/isaexec is a tool which acts as a "switch" which
"redirects" requests like /usr/bin/ksh93 (which is a hardlink to
/usr/lib/isaexec) to either /usr/bin/amd64/ksh93 or /usr/bin/i86/ksh93
depending on whether the matching ISA is supported or not (that way we
have a 64bit korn shell on 64bit platforms and in theory could allow
even accerlated versions like a normal 64bit AMD64 binary and a SSE5
binary))
BTW: far-fetched dreaming...
... do you accept ideas for more instruction extensions ? If "yes" - one
idea would be an "asyncronous block copy extension" which works like
this:
You issue an instruction which works like |memmove()| (e.g. copy memory,
even overlapping) but continues execution until a 2nd
wait_for_block_copy instruction is reached. That way the block copy
could be fully asyncronous to the normal pipeline operation and would
not block until the "wait_for_block_copy"-instruction is reached. This
needs to work with multiple (where "multiple" means an "unlimited"
nesting depth) block copies issued, e.g. something like...
-- snip --
instr_mem_move_start a b 25
some_other_instructions
instr_mem_move_start c d 15
instr_mem_move_start e f 15
some_other_instructions
some_other_instructions
call_a_subroutine
some_other_instructions
some_other_instructions
wait_for_block_copy # <--- waits for "e f 15"-copy
wait_for_block_copy # <--- waits for "c d 15"-copy
wait_for_block_copy # <--- waits for "a b 25"-copy
-- snip --
(e.g. you end-up of having some kind of memory copy pipeline which works
in parallel to the normla integer/f
|
|
|
|
|
Posts:
1,188
From:
US
Registered:
3/9/05
|
|
|
|
Re: Project proposal: "Solaris Enhancements for
AMD-based Platforms"
Posted:
Nov 2, 2007 5:50 PM
in response to: bostrovs
|
|
Ostrovsky, Boris wrote:
> I would like to propose creation of a new project titled "Solaris
> Enhancements for AMD-based Platforms".
>
> The project will address various features that are specific to platforms
> based on AMD processors, such as
> - IOMMU support
> - NUMA topology, particularly how it affects IO performance
> - Observability (performance counters, instruction-based
> sampling)
> - Power management
> - RAS features
> - New instruction support, new CPUID features
>
> Since what's described above covers fairly diverse range of subjects,
> the proposed project will serve as an umbrella for sub-projects, each of
> them covering a particular area related to improving Solaris behavior on
> systems built around AMD processors (as well as chipsets and graphics
> components)
>
> I think this project would be of interest to a number of OpenSolaris
> communities but I am asking the performance community for sponsorship as
> it appears to be the most relevant.
>
+1
=- Bart
--
Bart Smaalders Solaris Kernel Performance
barts at cyber dot eng dot sun dot com http://blogs.sun.com/barts
_______________________________________________
perf-discuss mailing list
perf-discuss at opensolaris dot org
|
|
|
|
|
Posts:
198
From:
US
Registered:
3/9/05
|
|
|
|
Re: Project proposal: "Solaris Enhancements for
AMD-based Platforms"
Posted:
Nov 9, 2007 5:05 PM
in response to: bostrovs
|
|
Ostrovsky, Boris wrote:
> I would like to propose creation of a new project titled "Solaris
> Enhancements for AMD-based Platforms".
>
I believe you have endorsement from the performance community. Any
objections?
-Eric
_______________________________________________
perf-discuss mailing list
perf-discuss at opensolaris dot org
|
|
|
|
|
Posts:
1
From:
SG
Registered:
9/27/05
|
|
|
|
Re: Project proposal: "Solaris Enhancements for AMD-based Platforms"
Posted:
Aug 28, 2008 5:55 AM
in response to: esaxe
To: Communities » performance » discuss
|
|
+1 for me as well
|
|
|
|
|
Darshak Modi
darshak.modi@eliteco...
|
|
|
|
Oracle 10g R1 performs extremely BAD
Posted:
Aug 28, 2008 6:11 AM
in response to: tayboonl
|
|
Hi All,
We are planning to move our application to T2 architecture blades.
When we tried to install Oracle and import the data, it took nearly 5-6
hours which on a normal PC takes 30 mins.
The server is 64 threads, 64GB RAM. The CPU is completely idle, cores are
idle, memory remains free still its now working
When I see in vmstat it says the blocked threads increases. In ps I can see
CPU time taken is 30 minutes where total time taken is almost 5 hrs.
Regards,
Darshak
_______________________________________________
perf-discuss mailing list
perf-discuss at opensolaris dot org
|
|
|
|
|
Posts:
124
From:
Lodz, Poland
Registered:
3/3/07
|
|
|
|
Re: Oracle 10g R1 performs extremely BAD
Posted:
Aug 28, 2008 6:33 AM
in response to: Darshak Modi
|
|
On Thu, Aug 28, 2008 at 06:41:42PM +0530, Darshak Modi wrote:
> Hi All,
> We are planning to move our application to T2 architecture blades.
> When we tried to install Oracle and import the data, it took nearly 5-6
> hours which on a normal PC takes 30 mins.
> The server is 64 threads, 64GB RAM. The CPU is completely idle, cores are
> idle, memory remains free still its now working
>
> When I see in vmstat it says the blocked threads increases. In ps I can see
> CPU time taken is 30 minutes where total time taken is almost 5 hrs.
T2 architecture is not very good (in performance terms) for single
threaded tasks. So import/export usually will take longer then before.
If you use Oracle 10 you can use parallel option of export/import
(it is called Data Pump).
http://articles.techrepublic.com.com/5100-10878_11-5800445.html
Regards
Przemek
--
http://przemol.blogspot.com/
---------------------------------------------------------------
Nasilaja sie kradzieze.
Mieszkancy osiedli zaniepokojeni.
Prosimy o pomoc w tej sprawie >>> http://link.interia.pl/f1eef
_______________________________________________
perf-discuss mailing list
perf-discuss at opensolaris dot org
|
|
|
|
|
Darshak Modi
darshak.modi@eliteco...
|
|
|
|
Re: Oracle 10g R1 performs extremely BAD
Posted:
Aug 28, 2008 6:59 AM
in response to: przemolb
|
|
We want to use Oracle on T2 machines. Is there any configuration to run
default processes in Paralle.
Thanks
Darshak
-----Original Message-----
From: perf-discuss-bounces at opensolaris dot org
[mailto:perf-discuss-bounces at opensolaris dot org] On Behalf Of
przemolicc at poczta dot fm
Sent: Thursday, August 28, 2008 7:04 PM
To: perf-discuss at opensolaris dot org
Subject: Re: [perf-discuss] Oracle 10g R1 performs extremely BAD
On Thu, Aug 28, 2008 at 06:41:42PM +0530, Darshak Modi wrote:
> Hi All,
> We are planning to move our application to T2 architecture blades.
> When we tried to install Oracle and import the data, it took nearly 5-6
> hours which on a normal PC takes 30 mins.
> The server is 64 threads, 64GB RAM. The CPU is completely idle, cores are
> idle, memory remains free still its now working
>
> When I see in vmstat it says the blocked threads increases. In ps I can
see
> CPU time taken is 30 minutes where total time taken is almost 5 hrs.
T2 architecture is not very good (in performance terms) for single
threaded tasks. So import/export usually will take longer then before.
If you use Oracle 10 you can use parallel option of export/import
(it is called Data Pump).
http://articles.techrepublic.com.com/5100-10878_11-5800445.html
Regards
Przemek
--
http://przemol.blogspot.com/
---------------------------------------------------------------
Nasilaja sie kradzieze.
Mieszkancy osiedli zaniepokojeni.
Prosimy o pomoc w tej sprawie >>> http://link.interia.pl/f1eef
_______________________________________________
perf-discuss mailing list
perf-discuss at opensolaris dot org
_______________________________________________
perf-discuss mailing list
perf-discuss at opensolaris dot org
|
|
|
|
|
Tanel Poder
tanel.poder.003@mail...
|
|
|
|
Re: Oracle 10g R1 performs extremely BAD
Posted:
Aug 28, 2008 6:39 AM
in response to: Darshak Modi
|
|
Hi Darshak,
The key question is how did you try to import data?
1) Using an Oracle data loading tool or a custom written one?
2) Using only single thread or in parallel?
If only 30 minutes of total 5 hours elapsed time was spent on CPU, the load
process was waiting on something a lot. Maybe for IO (especially if the
loading program commits too frequently) or just network rountrips (if
loading from a remote server with too small insert array sizes).
These are just guesses though, but the good thing is that you can measure
from inside Oracle, where the loading time is spent.
There are many tools for querying Oracle's instrumentation views
(V$SESSION_EVENT and V$SESSTAT for example), I have written a convenient
script for this (which doesn't require any installations nor db changes).
You can get it from here along some instructions:
http://blog.tanelpoder.com/2007/12/06/oracle-session-snapper-v106-released/
You should run it with let say 1-minute interval on the individual loading
session or all sessions in db. You can post the result so I (or other Oracle
people here) can advise...
--
Regards,
Tanel Poder
http://blog.tanelpoder.com
http://n.otepad.com - n.ote this!
> -----Original Message-----
> From: perf-discuss-bounces at opensolaris dot org
> [mailto:perf-discuss-bounces at opensolaris dot org] On Behalf Of
> Darshak Modi
> Sent: Thursday, August 28, 2008 21:12
> To: perf-discuss at opensolaris dot org
> Subject: [perf-discuss] Oracle 10g R1 performs extremely BAD
>
> Hi All,
> We are planning to move our application to T2 architecture blades.
> When we tried to install Oracle and import the data, it took
> nearly 5-6 hours which on a normal PC takes 30 mins.
> The server is 64 threads, 64GB RAM. The CPU is completely
> idle, cores are idle, memory remains free still its now working
>
> When I see in vmstat it says the blocked threads increases.
> In ps I can see CPU time taken is 30 minutes where total time
> taken is almost 5 hrs.
>
>
>
> Regards,
> Darshak
>
> _______________________________________________
> perf-discuss mailing list
> perf-discuss at opensolaris dot org
_______________________________________________
perf-discuss mailing list
perf-discuss at opensolaris dot org
|
|
|
|
|
Darshak Modi
darshak.modi@eliteco...
|
|
|
|
Re: Oracle 10g R1 performs extremely BAD
Posted:
Aug 28, 2008 6:58 AM
in response to: Tanel Poder
|
|
This is standard imp tool of oracle.
In vmstat it show blocked. In mpstat and iostat there are no huge write/read
going on. How will I find that its waiting for IO ?
Regards,
Darshak
-----Original Message-----
From: Tanel Poder [mailto:tanel dot poder dot 003 at mail dot ee]
Sent: Thursday, August 28, 2008 7:10 PM
To: 'Darshak Modi'; perf-discuss at opensolaris dot org
Subject: RE: [perf-discuss] Oracle 10g R1 performs extremely BAD
Hi Darshak,
The key question is how did you try to import data?
1) Using an Oracle data loading tool or a custom written one?
2) Using only single thread or in parallel?
If only 30 minutes of total 5 hours elapsed time was spent on CPU, the load
process was waiting on something a lot. Maybe for IO (especially if the
loading program commits too frequently) or just network rountrips (if
loading from a remote server with too small insert array sizes).
These are just guesses though, but the good thing is that you can measure
from inside Oracle, where the loading time is spent.
There are many tools for querying Oracle's instrumentation views
(V$SESSION_EVENT and V$SESSTAT for example), I have written a convenient
script for this (which doesn't require any installations nor db changes).
You can get it from here along some instructions:
http://blog.tanelpoder.com/2007/12/06/oracle-session-snapper-v106-released/
You should run it with let say 1-minute interval on the individual loading
session or all sessions in db. You can post the result so I (or other Oracle
people here) can advise...
--
Regards,
Tanel Poder
http://blog.tanelpoder.com
http://n.otepad.com - n.ote this!
> -----Original Message-----
> From: perf-discuss-bounces at opensolaris dot org
> [mailto:perf-discuss-bounces at opensolaris dot org] On Behalf Of
> Darshak Modi
> Sent: Thursday, August 28, 2008 21:12
> To: perf-discuss at opensolaris dot org
> Subject: [perf-discuss] Oracle 10g R1 performs extremely BAD
>
> Hi All,
> We are planning to move our application to T2 architecture blades.
> When we tried to install Oracle and import the data, it took
> nearly 5-6 hours which on a normal PC takes 30 mins.
> The server is 64 threads, 64GB RAM. The CPU is completely
> idle, cores are idle, memory remains free still its now working
>
> When I see in vmstat it says the blocked threads increases.
> In ps I can see CPU time taken is 30 minutes where total time
> taken is almost 5 hrs.
>
>
>
> Regards,
> Darshak
>
> _______________________________________________
> perf-discuss mailing list
> perf-discuss at opensolaris dot org
_______________________________________________
perf-discuss mailing list
perf-discuss at opensolaris dot org
|
|
|
|
|
Tanel Poder
tanel.poder.003@mail...
|
|
|
|
Re: [perf-discuss] Oracle 10g R1 performs extremely BAD
Posted:
Aug 28, 2008 9:17 AM
in response to: Darshak Modi
|
|
Yep, in addition to T2 having a good throughput only when the job is heavily
parallelized, the standard imp is also a quite low-perfomring import tool.
And it can't be parallelized (unless you manually export to multiple
different files and then run multiple imp's in parallel).
You should really be using Oracle's datapump (expdp and impdp). This can do
parallel execution too.
I noticed that you didn't mention whether it was the imp process or Oracle's
background server process which had consumed 30minutes of CPU time during 5
hour execution (and thus had spent rest of 4h 30minutes waiting for
something). If it was the imp process, then this can be understood as it has
to sit idle while Oracle server is serving the array insert request. If it
was the Oracle server process, then I would use my snapper script to
understand where this extra time was spent (as Oracle records exactly on
what the waiting time was spent).
There are few options for tuning the imp speed, set recordlength=65535 and
buffer=1048576 for example in imp command line.
The above options don't help (they probably wont help much) then look into
expdp/impdp...
--
Regards,
Tanel Poder
http://blog.tanelpoder.com
http://n.otepad.com - n.ote this!
> -----Original Message-----
> From: Darshak Modi [mailto:darshak dot modi at elitecore dot com]
> Sent: Thursday, August 28, 2008 21:59
> To: 'Tanel Poder'; perf-discuss at opensolaris dot org
> Subject: RE: [perf-discuss] Oracle 10g R1 performs extremely BAD
>
> This is standard imp tool of oracle.
>
> In vmstat it show blocked. In mpstat and iostat there are no
> huge write/read going on. How will I find that its waiting for IO ?
>
>
> Regards,
> Darshak
>
> -----Original Message-----
> From: Tanel Poder [mailto:tanel dot poder dot 003 at mail dot ee]
> Sent: Thursday, August 28, 2008 7:10 PM
> To: 'Darshak Modi'; perf-discuss at opensolaris dot org
> Subject: RE: [perf-discuss] Oracle 10g R1 performs extremely BAD
>
> Hi Darshak,
>
> The key question is how did you try to import data?
>
> 1) Using an Oracle data loading tool or a custom written one?
>
> 2) Using only single thread or in parallel?
>
> If only 30 minutes of total 5 hours elapsed time was spent on
> CPU, the load process was waiting on something a lot. Maybe
> for IO (especially if the loading program commits too
> frequently) or just network rountrips (if loading from a
> remote server with too small insert array sizes).
>
> These are just guesses though, but the good thing is that you
> can measure from inside Oracle, where the loading time is spent.
>
> There are many tools for querying Oracle's instrumentation
> views (V$SESSION_EVENT and V$SESSTAT for example), I have
> written a convenient script for this (which doesn't require
> any installations nor db changes).
>
> You can get it from here along some instructions:
>
> http://blog.tanelpoder.com/2007/12/06/oracle-session-snapper-v
106-released/
>
> You should run it with let say 1-minute interval on the
> individual loading session or all sessions in db. You can
> post the result so I (or other Oracle people here) can advise...
>
> --
> Regards,
> Tanel Poder
> http://blog.tanelpoder.com
> http://n.otepad.com - n.ote this!
>
>
> > -----Original Message-----
> > From: perf-discuss-bounces at opensolaris dot org
> > [mailto:perf-discuss-bounces at opensolaris dot org] On Behalf Of Darshak
> > Modi
> > Sent: Thursday, August 28, 2008 21:12
> > To: perf-discuss at opensolaris dot org
> > Subject: [perf-discuss] Oracle 10g R1 performs extremely BAD
> >
> > Hi All,
> > We are planning to move our application to T2 architecture blades.
> > When we tried to install Oracle and import the data, it took nearly
> > 5-6 hours which on a normal PC takes 30 mins.
> > The server is 64 threads, 64GB RAM. The CPU is completely
> idle, cores
> > are idle, memory remains free still its now working
> >
> > When I see in vmstat it says the blocked threads increases.
> > In ps I can see CPU time taken is 30 minutes where total
> time taken is
> > almost 5 hrs.
> >
> >
> >
> > Regards,
> > Darshak
> >
> > _______________________________________________
> > perf-discuss mailing list
> > perf-discuss at opensolaris dot org
>
>
_______________________________________________
perf-discuss mailing list
perf-discuss at opensolaris dot org
|
|
|
|
|
Posts:
198
From:
US
Registered:
3/9/05
|
|
|
|
Re: [perf-discuss] Project proposal: "Solaris Enhancements for
AMD-based Platforms"
Posted:
Aug 28, 2008 10:54 AM
in response to: tayboonl
|
|
Hi Raymond,
Raymond Tay Boon Leong wrote:
> +1 for me as well
>
This project is already underway...
http://www.opensolaris.org/os/project/amd/
-Eric
_______________________________________________
perf-discuss mailing list
perf-discuss at opensolaris dot org
|
|
|
|
|
