Thread: RE: [Fault-injection-developer] Need for benchmarking
Status: Alpha
Brought to you by:
rustyl
Menu
▾
▴
fault-injection-developer
|
From: Gao, K. <kev...@in...> - 2002-12-16 00:33:10
|
We can do some tiny program to test the time of MMIO read,
write operation in such situation:
common kernel.
Kernel with our FITH, but no faultset.
Kernel with our FITH, but the address of faultset is not associate with =
the
operation.
Kernel with our FITH, corresponding faultset but no action.
Kernel with our FITH, and injection corresponding fault into the =
operation.
Any other ideas?
Kevin
-----Original Message-----
From: Rusty Lynch [mailto:ru...@li...]=20
Sent: 2002=C4=EA12=D4=C214=C8=D5 0:36
To: fau...@li...
Subject: [Fault-injection-developer] Need for benchmarking
We could use some standardized benchmarking to=20
periodically run on our code base to get a feel for how
much overhead our changes are adding.
For example our current code adds a hook to the
page fault handler. It would be really easy to thrash
the kernel with even the smallest of tweaks to that
code.
Any idea's? Kevin, have planned on putting something
together?
-rustyl=20
-------------------------------------------------------
This sf.net email is sponsored by:
With Great Power, Comes Great Responsibility=20
Learn to use your power at OSDN's High Performance Computing Channel
http://hpc.devchannel.org/
_______________________________________________
Fault-injection-developer mailing list
Fau...@li...
https://lists.sourceforge.net/lists/listinfo/fault-injection-developer
|
|
From: Zhuang, L. <lou...@in...> - 2002-12-16 00:47:37
|
Maybe we can use LTP to test performance.
-----Original Message-----
From: Gao, Kevin [mailto:kev...@in...]
Sent: Monday, December 16, 2002 8:31 AM
To: 'Rusty Lynch'; fau...@li...
Subject: RE: [Fault-injection-developer] Need for benchmarking
We can do some tiny program to test the time of MMIO read,
write operation in such situation:
common kernel.
Kernel with our FITH, but no faultset.
Kernel with our FITH, but the address of faultset is not associate with =
the
operation.
Kernel with our FITH, corresponding faultset but no action.
Kernel with our FITH, and injection corresponding fault into the =
operation.
Any other ideas?
Kevin
-----Original Message-----
From: Rusty Lynch [mailto:ru...@li...]
Sent: 2002=C4=EA12=D4=C214=C8=D5 0:36
To: fau...@li...
Subject: [Fault-injection-developer] Need for benchmarking
We could use some standardized benchmarking to
periodically run on our code base to get a feel for how
much overhead our changes are adding.
For example our current code adds a hook to the
page fault handler. It would be really easy to thrash
the kernel with even the smallest of tweaks to that
code.
Any idea's? Kevin, have planned on putting something
together?
-rustyl
-------------------------------------------------------
This sf.net email is sponsored by:
With Great Power, Comes Great Responsibility
Learn to use your power at OSDN's High Performance Computing Channel
http://hpc.devchannel.org/
_______________________________________________
Fault-injection-developer mailing list
Fau...@li...
https://lists.sourceforge.net/lists/listinfo/fault-injection-developer
-------------------------------------------------------
This sf.net email is sponsored by:
With Great Power, Comes Great Responsibility
Learn to use your power at OSDN's High Performance Computing Channel
http://hpc.devchannel.org/
_______________________________________________
Fault-injection-developer mailing list
Fau...@li...
https://lists.sourceforge.net/lists/listinfo/fault-injection-developer
|
|
From: Gao, K. <kev...@in...> - 2002-12-16 00:54:26
|
Yes, great idea!
-----Original Message-----
From: Zhuang, Louis=20
Sent: 2002=C4=EA12=D4=C216=C8=D5 8:45
To: Gao, Kevin; 'Rusty Lynch';
fau...@li...
Subject: RE: [Fault-injection-developer] Need for benchmarking
Maybe we can use LTP to test performance.
-----Original Message-----
From: Gao, Kevin [mailto:kev...@in...]
Sent: Monday, December 16, 2002 8:31 AM
To: 'Rusty Lynch'; fau...@li...
Subject: RE: [Fault-injection-developer] Need for benchmarking
We can do some tiny program to test the time of MMIO read,
write operation in such situation:
common kernel.
Kernel with our FITH, but no faultset.
Kernel with our FITH, but the address of faultset is not associate with =
the
operation.
Kernel with our FITH, corresponding faultset but no action.
Kernel with our FITH, and injection corresponding fault into the =
operation.
Any other ideas?
Kevin
-----Original Message-----
From: Rusty Lynch [mailto:ru...@li...]
Sent: 2002=C4=EA12=D4=C214=C8=D5 0:36
To: fau...@li...
Subject: [Fault-injection-developer] Need for benchmarking
We could use some standardized benchmarking to
periodically run on our code base to get a feel for how
much overhead our changes are adding.
For example our current code adds a hook to the
page fault handler. It would be really easy to thrash
the kernel with even the smallest of tweaks to that
code.
Any idea's? Kevin, have planned on putting something
together?
-rustyl
-------------------------------------------------------
This sf.net email is sponsored by:
With Great Power, Comes Great Responsibility
Learn to use your power at OSDN's High Performance Computing Channel
http://hpc.devchannel.org/
_______________________________________________
Fault-injection-developer mailing list
Fau...@li...
https://lists.sourceforge.net/lists/listinfo/fault-injection-developer
-------------------------------------------------------
This sf.net email is sponsored by:
With Great Power, Comes Great Responsibility
Learn to use your power at OSDN's High Performance Computing Channel
http://hpc.devchannel.org/
_______________________________________________
Fault-injection-developer mailing list
Fau...@li...
https://lists.sourceforge.net/lists/listinfo/fault-injection-developer
|
|
From: Lynch, R. <rus...@in...> - 2002-12-16 03:51:18
|
There have been people posting benchmarking results on LKML. I haven't = looked into the results or which tool they are using to generate the results, but I am guessing that they would do the job. The iterations you listed sound fine. -rustyl > -----Original Message----- > From: Gao, Kevin [mailto:kev...@in...] > Sent: Sunday, December 15, 2002 4:31 PM > To: 'Rusty Lynch'; fau...@li... > Subject: RE: [Fault-injection-developer] Need for benchmarking >=20 >=20 > We can do some tiny program to test the time of MMIO read, > write operation in such situation: >=20 > common kernel. > Kernel with our FITH, but no faultset. > Kernel with our FITH, but the address of faultset is not=20 > associate with the > operation. > Kernel with our FITH, corresponding faultset but no action. > Kernel with our FITH, and injection corresponding fault into=20 > the operation. >=20 > Any other ideas? >=20 > Kevin >=20 >=20 > -----Original Message----- > From: Rusty Lynch [mailto:ru...@li...]=20 > Sent: 2002=C4=EA12=D4=C214=C8=D5 0:36 > To: fau...@li... > Subject: [Fault-injection-developer] Need for benchmarking >=20 >=20 > We could use some standardized benchmarking to=20 > periodically run on our code base to get a feel for how > much overhead our changes are adding. >=20 > For example our current code adds a hook to the > page fault handler. It would be really easy to thrash > the kernel with even the smallest of tweaks to that > code. >=20 > Any idea's? Kevin, have planned on putting something > together? >=20 > -rustyl=20 >=20 >=20 > ------------------------------------------------------- > This sf.net email is sponsored by: > With Great Power, Comes Great Responsibility=20 > Learn to use your power at OSDN's High Performance Computing Channel > http://hpc.devchannel.org/ > _______________________________________________ > Fault-injection-developer mailing list > Fau...@li... > = https://lists.sourceforge.net/lists/listinfo/fault-injection-developer >=20 >=20 > ------------------------------------------------------- > This sf.net email is sponsored by: > With Great Power, Comes Great Responsibility=20 > Learn to use your power at OSDN's High Performance Computing Channel > http://hpc.devchannel.org/ > _______________________________________________ > Fault-injection-developer mailing list > Fau...@li... > = https://lists.sourceforge.net/lists/listinfo/fault-injection-developer >=20 |
|
From: Lynch, R. <rus...@in...> - 2002-12-16 03:54:09
|
Does LTP contain a benchmarking tool? We could definitely=20 run it to make sure we didn't accidentally break something=20 else in the kernel. BTW, do not overlook using STP on OSDL as a testing tool. It's easy to use and there are a bunch of machines waiting to be utilized. -rusty > -----Original Message----- > From: Zhuang, Louis [mailto:lou...@in...] > Sent: Sunday, December 15, 2002 4:45 PM > To: Gao, Kevin; 'Rusty Lynch';=20 > fau...@li... > Subject: RE: [Fault-injection-developer] Need for benchmarking >=20 >=20 > Maybe we can use LTP to test performance. >=20 > -----Original Message----- > From: Gao, Kevin [mailto:kev...@in...] > Sent: Monday, December 16, 2002 8:31 AM > To: 'Rusty Lynch'; fau...@li... > Subject: RE: [Fault-injection-developer] Need for benchmarking >=20 > We can do some tiny program to test the time of MMIO read, > write operation in such situation: >=20 > common kernel. > Kernel with our FITH, but no faultset. > Kernel with our FITH, but the address of faultset is not=20 > associate with the > operation. > Kernel with our FITH, corresponding faultset but no action. > Kernel with our FITH, and injection corresponding fault into=20 > the operation. >=20 > Any other ideas? >=20 > Kevin >=20 >=20 > -----Original Message----- > From: Rusty Lynch [mailto:ru...@li...] > Sent: 2002=C4=EA12=D4=C214=C8=D5 0:36 > To: fau...@li... > Subject: [Fault-injection-developer] Need for benchmarking >=20 >=20 > We could use some standardized benchmarking to > periodically run on our code base to get a feel for how > much overhead our changes are adding. >=20 > For example our current code adds a hook to the > page fault handler. It would be really easy to thrash > the kernel with even the smallest of tweaks to that > code. >=20 > Any idea's? Kevin, have planned on putting something > together? >=20 > -rustyl >=20 >=20 > ------------------------------------------------------- > This sf.net email is sponsored by: > With Great Power, Comes Great Responsibility > Learn to use your power at OSDN's High Performance Computing Channel > http://hpc.devchannel.org/ > _______________________________________________ > Fault-injection-developer mailing list > Fau...@li... > = https://lists.sourceforge.net/lists/listinfo/fault-injection-developer >=20 >=20 > ------------------------------------------------------- > This sf.net email is sponsored by: > With Great Power, Comes Great Responsibility > Learn to use your power at OSDN's High Performance Computing Channel > http://hpc.devchannel.org/ > _______________________________________________ > Fault-injection-developer mailing list > Fau...@li... > = https://lists.sourceforge.net/lists/listinfo/fault-injection-developer >=20 >=20 > ------------------------------------------------------- > This sf.net email is sponsored by: > With Great Power, Comes Great Responsibility=20 > Learn to use your power at OSDN's High Performance Computing Channel > http://hpc.devchannel.org/ > _______________________________________________ > Fault-injection-developer mailing list > Fau...@li... > = https://lists.sourceforge.net/lists/listinfo/fault-injection-developer >=20 |
|
From: Wang, S. <sta...@in...> - 2002-12-16 04:58:35
|
We've download "lmbench". And it may help us :) Stan > -----Original Message----- > From: Lynch, Rusty [mailto:rus...@in...] > Sent: 2002=C4=EA12=D4=C216=C8=D5 11:54 > To: Zhuang, Louis; Gao, Kevin; 'Rusty Lynch'; > fau...@li... > Subject: RE: [Fault-injection-developer] Need for benchmarking >=20 >=20 > Does LTP contain a benchmarking tool? We could definitely=20 > run it to make sure we didn't accidentally break something=20 > else in the kernel. >=20 > BTW, do not overlook using STP on OSDL as a testing tool. > It's easy to use and there are a bunch of machines waiting > to be utilized. >=20 > -rusty >=20 > > -----Original Message----- > > From: Zhuang, Louis [mailto:lou...@in...] > > Sent: Sunday, December 15, 2002 4:45 PM > > To: Gao, Kevin; 'Rusty Lynch';=20 > > fau...@li... > > Subject: RE: [Fault-injection-developer] Need for benchmarking > >=20 > >=20 > > Maybe we can use LTP to test performance. > >=20 > > -----Original Message----- > > From: Gao, Kevin [mailto:kev...@in...] > > Sent: Monday, December 16, 2002 8:31 AM > > To: 'Rusty Lynch'; fau...@li... > > Subject: RE: [Fault-injection-developer] Need for benchmarking > >=20 > > We can do some tiny program to test the time of MMIO read, > > write operation in such situation: > >=20 > > common kernel. > > Kernel with our FITH, but no faultset. > > Kernel with our FITH, but the address of faultset is not=20 > > associate with the > > operation. > > Kernel with our FITH, corresponding faultset but no action. > > Kernel with our FITH, and injection corresponding fault into=20 > > the operation. > >=20 > > Any other ideas? > >=20 > > Kevin > >=20 > >=20 > > -----Original Message----- > > From: Rusty Lynch [mailto:ru...@li...] > > Sent: 2002=C4=EA12=D4=C214=C8=D5 0:36 > > To: fau...@li... > > Subject: [Fault-injection-developer] Need for benchmarking > >=20 > >=20 > > We could use some standardized benchmarking to > > periodically run on our code base to get a feel for how > > much overhead our changes are adding. > >=20 > > For example our current code adds a hook to the > > page fault handler. It would be really easy to thrash > > the kernel with even the smallest of tweaks to that > > code. > >=20 > > Any idea's? Kevin, have planned on putting something > > together? > >=20 > > -rustyl > >=20 > >=20 > > ------------------------------------------------------- > > This sf.net email is sponsored by: > > With Great Power, Comes Great Responsibility > > Learn to use your power at OSDN's High Performance Computing = Channel > > http://hpc.devchannel.org/ > > _______________________________________________ > > Fault-injection-developer mailing list > > Fau...@li... > >=20 > = https://lists.sourceforge.net/lists/listinfo/fault-injection-developer > >=20 > >=20 > > ------------------------------------------------------- > > This sf.net email is sponsored by: > > With Great Power, Comes Great Responsibility > > Learn to use your power at OSDN's High Performance Computing = Channel > > http://hpc.devchannel.org/ > > _______________________________________________ > > Fault-injection-developer mailing list > > Fau...@li... > >=20 > = https://lists.sourceforge.net/lists/listinfo/fault-injection-developer > >=20 > >=20 > > ------------------------------------------------------- > > This sf.net email is sponsored by: > > With Great Power, Comes Great Responsibility=20 > > Learn to use your power at OSDN's High Performance Computing = Channel > > http://hpc.devchannel.org/ > > _______________________________________________ > > Fault-injection-developer mailing list > > Fau...@li... > >=20 > = https://lists.sourceforge.net/lists/listinfo/fault-injection-developer > >=20 >=20 >=20 > ------------------------------------------------------- > This sf.net email is sponsored by: > With Great Power, Comes Great Responsibility=20 > Learn to use your power at OSDN's High Performance Computing Channel > http://hpc.devchannel.org/ > _______________________________________________ > Fault-injection-developer mailing list > Fau...@li... > = https://lists.sourceforge.net/lists/listinfo/fault-injection-developer >=20 |
|
From: Rusty L. <ru...@li...> - 2002-12-16 05:06:11
|
excellent. If we can start getting regular results
posted then we can get a clue as to what kind of overhead
we are causing.
Seems like something that could be automated if you have
the hardware available.
--rusty
----- Original Message -----
From: "Wang, Stanley" <sta...@in...>
To: "Lynch, Rusty" <rus...@in...>; "Zhuang, Louis" <louis.zhuang=
@intel.com>; "Gao, Kevin" <kev...@in...>; "'Rusty
Lynch'" <ru...@li...>; <fau...@li...ur=
ceforge.net>
Sent: Sunday, December 15, 2002 8:56 PM
Subject: RE: [Fault-injection-developer] Need for benchmarking
> We've download "lmbench". And it may help us :)
>
> Stan
>
> > -----Original Message-----
> > From: Lynch, Rusty [mailto:rus...@in...]
> > Sent: 2002=C4=EA12=D4=C216=C8=D5 11:54
> > To: Zhuang, Louis; Gao, Kevin; 'Rusty Lynch';
> > fau...@li...
> > Subject: RE: [Fault-injection-developer] Need for benchmarking
> >
> >
> > Does LTP contain a benchmarking tool? We could definitely
> > run it to make sure we didn't accidentally break something
> > else in the kernel.
> >
> > BTW, do not overlook using STP on OSDL as a testing tool.
> > It's easy to use and there are a bunch of machines waiting
> > to be utilized.
> >
> > -rusty
> >
> > > -----Original Message-----
> > > From: Zhuang, Louis [mailto:lou...@in...]
> > > Sent: Sunday, December 15, 2002 4:45 PM
> > > To: Gao, Kevin; 'Rusty Lynch';
> > > fau...@li...
> > > Subject: RE: [Fault-injection-developer] Need for benchmarking
> > >
> > >
> > > Maybe we can use LTP to test performance.
> > >
> > > -----Original Message-----
> > > From: Gao, Kevin [mailto:kev...@in...]
> > > Sent: Monday, December 16, 2002 8:31 AM
> > > To: 'Rusty Lynch'; fau...@li...
> > > Subject: RE: [Fault-injection-developer] Need for benchmarking
> > >
> > > We can do some tiny program to test the time of MMIO read,
> > > write operation in such situation:
> > >
> > > common kernel.
> > > Kernel with our FITH, but no faultset.
> > > Kernel with our FITH, but the address of faultset is not
> > > associate with the
> > > operation.
> > > Kernel with our FITH, corresponding faultset but no action.
> > > Kernel with our FITH, and injection corresponding fault into
> > > the operation.
> > >
> > > Any other ideas?
> > >
> > > Kevin
> > >
> > >
> > > -----Original Message-----
> > > From: Rusty Lynch [mailto:ru...@li...]
> > > Sent: 2002=C4=EA12=D4=C214=C8=D5 0:36
> > > To: fau...@li...
> > > Subject: [Fault-injection-developer] Need for benchmarking
> > >
> > >
> > > We could use some standardized benchmarking to
> > > periodically run on our code base to get a feel for how
> > > much overhead our changes are adding.
> > >
> > > For example our current code adds a hook to the
> > > page fault handler. It would be really easy to thrash
> > > the kernel with even the smallest of tweaks to that
> > > code.
> > >
> > > Any idea's? Kevin, have planned on putting something
> > > together?
> > >
> > > -rustyl
> > >
> > >
> > > -------------------------------------------------------
> > > This sf.net email is sponsored by:
> > > With Great Power, Comes Great Responsibility
> > > Learn to use your power at OSDN's High Performance Computing Channe=
l
> > > http://hpc.devchannel.org/
> > > _______________________________________________
> > > Fault-injection-developer mailing list
> > > Fau...@li...
> > >
> > https://lists.sourceforge.net/lists/listinfo/fault-injection-develope=
r
> > >
> > >
> > > -------------------------------------------------------
> > > This sf.net email is sponsored by:
> > > With Great Power, Comes Great Responsibility
> > > Learn to use your power at OSDN's High Performance Computing Channe=
l
> > > http://hpc.devchannel.org/
> > > _______________________________________________
> > > Fault-injection-developer mailing list
> > > Fau...@li...
> > >
> > https://lists.sourceforge.net/lists/listinfo/fault-injection-develope=
r
> > >
> > >
> > > -------------------------------------------------------
> > > This sf.net email is sponsored by:
> > > With Great Power, Comes Great Responsibility
> > > Learn to use your power at OSDN's High Performance Computing Channe=
l
> > > http://hpc.devchannel.org/
> > > _______________________________________________
> > > Fault-injection-developer mailing list
> > > Fau...@li...
> > >
> > https://lists.sourceforge.net/lists/listinfo/fault-injection-develope=
r
> > >
> >
> >
> > -------------------------------------------------------
> > This sf.net email is sponsored by:
> > With Great Power, Comes Great Responsibility
> > Learn to use your power at OSDN's High Performance Computing Channel
> > http://hpc.devchannel.org/
> > _______________________________________________
> > Fault-injection-developer mailing list
> > Fau...@li...
> > https://lists.sourceforge.net/lists/listinfo/fault-injection-develope=
r
> >
>
>
> -------------------------------------------------------
> This sf.net email is sponsored by:
> With Great Power, Comes Great Responsibility
> Learn to use your power at OSDN's High Performance Computing Channel
> http://hpc.devchannel.org/
> _______________________________________________
> Fault-injection-developer mailing list
> Fau...@li...
> https://lists.sourceforge.net/lists/listinfo/fault-injection-developer
|
×
Want the latest updates on software, tech news, and AI?
Get latest updates about software, tech news, and AI from SourceForge directly in your inbox once a month.
Thanks for helping keep SourceForge clean.
X