Dear Jeff, Sorry I could not test the cluster earlier but I am having problems with one compute node.(I will have to replace it!). So I will have to repeat this test with 15 nodes. Yes I had 4 NIC cards on the head node and it was only eth3 that was the gigabit NIC which was communicating to other eth1 gigabit Nics on the compute nodes through a gigabit switch. So though I did not specify the ethernet interface in the switch --mca pml teg I was getting good performance but in --mca btl tcp not specifying the interface seems to create problems. I wiped out the Linux FC3 installation and tried again with Oscar 4.2 but am having problems with --mca btl tcp switch. mpirun --mca btl tcp --prefix /home/allan/openmpi --hostfile aa -np 16 ./xhpl The hostfile aa contains the 16 hosts to So to recap the cluster is only connected to itself through the giga bit 16 port switch through gigabit ethernet cards to form a LAN with an IP for each. There is an extra ethernet card built into the compute motherboards that is 10/100Mbps that is not connected to anything yet. Please can you tell me the right mpirun command line for btl tcp for my setup? Is the hostfile right? for the mpirun command above? Should it include which is the head node from where I am invoking mpirun? Or should it not have the head node? Thank you, Allan Message: 2 Date: Sun, 13 Nov 2005 15:51:30 -0500 From: Jeff Squyres <> Subject: Re: [O-MPI users] HPL anf TCP To: Open MPI Users <> Message-ID: <> Content-Type: text/plain; charset=US-ASCII; format=flowed On Nov 3, 2005, at 8:35 PM, Allan Menezes wrote:
> 1. No, I have 4 NICs on the head node and two on each of the 15 other 
> compute nodes. I use the realtek 8169 gigabit ethernet cards on the 
> compute nodes as eth1 or eth0(one only) connected to a gigabit 
> ethernet switch with bisection bandwidth of 32Gbps and a sk98lin 
> driver 3Com built in gigabit ethernet NIC card on the head node(eth3). 
> The other ethernet cards 10/100M on the head node handle a network 
> laser printer(eth0) and eth2 (10/100M) internet access. Eth1 is a 
> spare 10/100M which I can remove. The compute nodes each have two 
> ethernet cards one 10/100Mbps ethernet not connected to anything(built 
> in to M/B) and a PCI realtek 8169 gigabit ethernet connected to the 
> TCP network LAN(Gigabit). When I tried it without the switches -mca 
> pml teg the maximum performace I would get with it was 9GFlops for P=4 
> Q=4 N=approx 12- 16 thousand and NB ridiculously low at 10 block size. 
> If I tried bigger block sizes it would run for along time for large N 
> ~ 16,000 unless I killed xhpl. I use atlas BLAS 3.7.11 libs compiled 
> for each node and linked to HPL when creating xhpl. I also use open 
> mpi mpicc in the hpl make file for compile and link both.  Maybe I 
> should according to the new faq use the TCP switch to use eth3 on the 
> head node?

So if I'm reading that right, there's only one network that connects 
the head node and the compute nodes, right?

That's right!
> 2. I have 512MB of memory per node which is 8 GB total, so I can 
> safely go upto N=22,000 24,000. I used sizes of 22000 for TCP teg and 
> did not run into problems. But if I do not specify the switches 
> suggested by Tim I get bad performance for N = 12000.

I must admit that I'm still befuddled by this -- we are absolutely 
unable to duplicate this behavior.  It *sounds* like there is some 
network mismatching going on in here -- that the tcp btl is somehow 
routing information differently than the tcp ptl (and therefore taking 
longer -- timing out and the like).

We did make some improvements to the tcp subnet mask matching code for 
rc5; I had to ask again, but could you try with the latest nightly 
snapshot tarball?
I will try it in the near future if time permits with the latest 1.0 snapshot and report back.
I had to "re-image" my cluster so I have some more work to do

> 4. My cluster is an experimental Basement Cluster [BSquared = Brampton 
> Beowulf] built out of x86 Durons(6), 2 athlons, 2 semprons, two P4s, 2 
> 64 bit x86_64 AMD64 ATHLONS and two AMD x86_64 SEmprons(754 pin) for a 
> total of 16 machines running FC3 and Oscar beta cluster software. I 
> have not tried it with the latest open mpi snapshot yet but I will to 
> night. I think I should reinstall FC3 on the head node P4 2.8GHz and 
> reinstall all the compute nodes with Oscar beta Nov 3, 2005 and open 
> mpi of todays Nov 3, 2005 1.0 snapshot and try again. I could have 
> made an errror somewhere before. It should not take me long. But I 
> doubt it as MPICH2 and open mpi with the switches pml teg give good 
> comparable performance. I was not using jumo MTU frames either just 
> 1500bytes. It is not homogenous (BSquared) but a good test set up.
> If you have any advice, Please tell me and I could try it out.
> Thank you and good luck!
> Allan
> On Oct 27, 2005, at 10:19 AM, Jeff Squyres wrote:
>> > On Oct 19, 2005, at 12:04 AM, Allan Menezes wrote:
>> >
>> >
>>> >> We've done linpack runs recently w/ Infiniband, which result in
>>> >> performance
>>> >> comparable to mvapich, but not w/ the tcp port. Can you try
>>> >> running w/
>>> >> an
>>> >> earlier version, specify on the command line:
>>> >>
>>> >> -mca pml teg
>>> >> Hi Tim,
>>> >>   I tried the same cluster (16 node x86) with the switches -mca 
>>> pml
>>> >> teg and I get good performance of 24.52Gflops at N=22500
>>> >> and Block size NB=120.
>>> >> My command line now looks like :
>>> >> a1> mpirun -mca pls_rsh_orted /home/allan/openmpi/bin/orted -mca 
>>> pml
>>> >> teg -hostile aa -np 16 ./xhpl
>>> >> hostfile = aa, containing the addresses of the 16 machines.
>>> >> I am using a GS116 16 port netgear Gigabit ethernet switch with 
>>> Gnet
>>> >> realtek gig ethernet cards
>>> >> Why, PLEASE, do these switches pml teg make such a difference? 
>>> It's
>>> >> 2.6 times more performance in GFlops than what I was getting 
>>> without
>>> >> them.
>>> >> I tried version rc3 and not an earlier version.
>>> >> Thank you very much for your assistance!
>>> >>
>> >
>> > Sorry for the delay in replying to this...
>> >
>> > The "pml teg" switch tells Open MPI to use the 2nd generation TCP
>> > implementation rather than the 3rd generation TCP.  More 
>> specifically,
>> > the "PML" is the point-to-point management layer.  There are 2
>> > different components for this -- teg (2nd generation) and ob1 (3rd
>> > generation).  "ob1" is the default; specifying "--mca pml teg" tells
>> > Open MPI to use the "teg" component instead of ob1.
>> >
>> > Note, however, that teg and ob1 know nothing about TCP -- it's the 
>> 2nd
>> > order implications that make the difference here.  teg and ob1 use
>> > different back-end components to talk across networks:
>> >
>> > - teg uses PTL components (point-to-point transport layer -- 2nd 
>> gen)
>> > - ob1 uses BTL components (byte transfer layer -- 3rd gen)
>> >
>> > We obviously have TCP implementations for both the PTL and BTL.
>> > Considerable time was spent optimizing the TCP PTL (i.e., 2nd gen).
>> > Unfortunately, as yet, little time has been spent optimizing the TCP
>> > BTL (i.e., 3rd gen) -- it was a simple port, nothing more.
>> >
>> > We have spent the majority of our time, so far, optimizing the 
>> Myrinet
>> > and Infiniband BTLs (therefore showing that excellent performance is
>> > achievable in the BTLs).  However, I'm quite disappointed by the TCP
>> > BTL performance -- it sounds like we have a protocol mismatch that 
>> is
>> > arbitrarily slowing everything down, and something that needs to be
>> > fixed before 1.0 (it's not a problem with the BTL design, since IB 
>> and
>> > Myrinet performance is quite good -- just a problem/bug in the TCP
>> > implementation of the BTL).  That much performance degradation is
>> > clearly unacceptable.
>> >
>> > --
>> > {+} Jeff Squyres
>> > {+} The Open MPI Project
>> > {+}
>> >
>> > _______________________________________________
>> > users mailing list
>> >
>> >
>> >
> -- 
> _______________________________________________
> users mailing list

-- {+} Jeff Squyres {+} The Open MPI Project {+} ------------------------------