Sorry, it took me so long to reply, I was out of office these days.
The paper I sent, the author did the test using 100% CPU, but not in my
case.
But since I did not look into the code of TIPC, I am afraid I cannot
explain it. I can only
show you my result and hope you get the same or even bett
On Sep 1, 2011, at 7:05 AM, Xin He wrote:
> And get the result as in appendix. It seems that TCP has better performances
> with smaller message while TIPC with larger message.
Interesting. Any idea why? From the TIPC paper you sent, one of TIPC's
strengths was that it was supposed to be faste
hi, I found the reason. It is because besides the direct links between 2
PCs, there is another link going through many switches and TCP BTL seems
to use
this slower link. So I run again with eth0 only.
So I build ompi with: ./configure --disable-mpi-f90 --disable-mpi-f77
--disable-mpi-cxx --di
Thanks for the links.
I found a link(below) to compare TIPC, TCP and SCTP. But it uses some old
version of TIPC(1.7.3). Do you have any similar tests but on the latest
version of TIPC, TCP and SCTP. That will be more helpful to convince people
to use TIPC.
Another thing I am interested is wheth
On Aug 29, 2011, at 3:51 AM, Xin He wrote:
>> -
>> $ mpirun --mca btl tcp,self --bynode -np 2 --mca btl_tcp_if_include eth0
>> hostname
>> svbu-mpi008
>> svbu-mpi009
>> $ mpirun --mca btl tcp,self --bynode -np 2 --mca btl_tcp_if_include eth0
>> IMB-MPI1 PingPong
>> #-
Yes, it is Gigabytes Ethernet. I configure ompi again using "./configure
--disable-mpi-f90 --disable-mpi-f77 --disable-mpi-cxx --disable-vt
--disable-io-romio --prefix=/usr --with-platform=optimized"
and run IMB-MPI1 again with "mpirun --mca btl tcp,self -n 2 --hostfile
my_hostfile --bynode ./IM
Is your interconnect Gigabytes Ethernet? It's very surprised to see TCP BTL
just got 33MBytes peak BW on your cluster. I did a similar test on an amd
cluster with gigabytes Ethernet. As following shows, the TCP BTL's BW is
similar with your tipc(112MBytes/s). Could you redo the test with 2
proce
On 08/25/2011 03:14 PM, Jeff Squyres wrote:
On Aug 25, 2011, at 8:25 AM, Xin He wrote:
Can you edit your configure.m4 directly and test it and whatnot? I provided
the configure.m4 as a starting point for you. :-) It shouldn't be hard to
make it check linux/tipc.h instead of tipc.h. I'm
On Aug 25, 2011, at 8:25 AM, Xin He wrote:
>> Can you edit your configure.m4 directly and test it and whatnot? I provided
>> the configure.m4 as a starting point for you. :-) It shouldn't be hard to
>> make it check linux/tipc.h instead of tipc.h. I'm happy to give you direct
>> write acces
On 08/23/2011 04:35 PM, Jeff Squyres wrote:
On Aug 23, 2011, at 9:54 AM, Xin He wrote:
Hi, I modified the code, copyright comments added.
I added your fixes to https://bitbucket.org/jsquyres/ompi-tipc.
And about configure.m4, sorry I was not clear before, tipc.h is under
/usr/include/linux/
On Aug 23, 2011, at 9:54 AM, Xin He wrote:
> Hi, I modified the code, copyright comments added.
I added your fixes to https://bitbucket.org/jsquyres/ompi-tipc.
> And about configure.m4, sorry I was not clear before, tipc.h is under
> /usr/include/linux/tipc.h, not under include directly.
Can y
Hi, I modified the code, copyright comments added.
And about configure.m4, sorry I was not clear before, tipc.h is under
/usr/include/linux/tipc.h, not under include directly.
I have done some tests using tools like NetPIPE, osu and IMB and the
result shows that TIPC BTL has a better performa
Ok. For the moment, you might want to leave the priority alone and see how it
goes. You can always manually turn off the SM BTL to test performance with and
without it. If it turns out to be better than the SM BTL, we can play the
priority tricks.
On Aug 17, 2011, at 10:09 AM, Xin He wrote:
No there is no library that must be linked to. :-)
About the performance compared to SM, I have not tested that yet. So
far, I compared it with TCP. It has better performances under some
circumstances, not all.
Now I am working with profiling tools, hope to find reasons and improve it.
/Xin
BTW, is there a libtipc that must be linked against? If so, can you give me a
symbol name to check for in there?
On Aug 17, 2011, at 9:53 AM, Jeff Squyres wrote:
> I put it here:
>
>https://bitbucket.org/jsquyres/ompi-tipc/overview
>
> You can clone that repo with the Mercurial distribute
I put it here:
https://bitbucket.org/jsquyres/ompi-tipc/overview
You can clone that repo with the Mercurial distributed version control tool.
I'll add a configure.m4 shortly; possibly today. You can test it for me. :-)
For the SM stuff, perhaps TIPC should just have a higher priority than
It is a single component.
And could someone write a configure file for me?
structure sockaddr_tipc (defined in tipc.h)is a good sign we have tipc.
And also TIPC cannot use with SM component, because TIPC use shared
memory as well for communication between processes on the same node.
Please k
Is your code self-contained in a single component?
If it's a small (compressed) tarball, just send it to the list. Otherwise, you
might want to post it somewhere like bitbucket.org where people can download
and look at it.
On Aug 17, 2011, at 4:00 AM, Xin He wrote:
> Hi developers,
>
> I ha
18 matches
Mail list logo
