Lawrence Berkeley National Lab

Wow, LBL and Bereley is really cool. It’s basically at a large mountain. You can see San Francisco really well and the nature is great. I guess it’s really expensive :). My talk went very good and the listeners were very interested (we had many discussions).
Here the view from the Lab to San Francisco:

I’m now at Livermore. The hotel had a free wine hour today ;). But I missed it (had a fully-planned day). Ah, anyway, I’m here to work. The visit at the lab was very productive! I hope to work with some of those people soon ;).

I’ll proceed to the HOTI conference tomorrow ;).

SC07, Cluster Challenge and no sleep

Update: visit also my Cluster Challenge page 🙂

Not much to add to the tagline. I was at SC07 :). Oh, if you don’t know this conference … it’s one of the biggest in CS. There were about 9000 attendees this year. Actually I was supposed to go there to give my talk , but it turned out to be different. Shortly before the conference, I got suckered into this cluster challenge thing. It promised to be a lot of fun, and it was eventually. But it also costed a lot … beginning from huge time investments before the conference, weekends in the lab over sleepless night during the conference and finally some more problems that shall remain unmentioned. We will see if I will do it again …

My talk was a catastrophe. It was well prepared (in breaks during the cluster challenge preparation) … so a test-presentation went pretty good. But I did not think about a possible risk … technology. My MacBook stopped working properly as soon as I started the presentation (in front of the whole SC audience!!). Ok, so we had a backup-PC and I had my slides on a backup USB stick that Microsoft gave me for free before that. I should not have trusted this thing! So the talk went on .. pretty ok, but suddenly the USB stick (plugged to a MS Windows PC!) stopped working. The fiddling began again and finally we (the chairman and me) made it work again. Oh my god … everything failed! A total nightmare! But the audience was pretty patient (I guess they know about technical problems pretty well ;)).

Some pictures from my talk:

Yes, there were quite some interested people asking questions afterwards!

Ah, and something else … I lost a bet (as you can see :)):

Yes, I impersonated Pikachu! I turned into the Stony Magnet ;). But yeah guys … it did not work … so the real Stony magnet (which worked *amazingly* well!) is here:

EuroPVM

Some pictures from the conference …

I am talking about non-blocking collectives here – we had an interesting question/discussion session afterwards :).

  The reception in the French Senate – pretty classy ;).

Not too much more going on so far (besides some “pub-events” :-))

Houston …

For those of you who don’t know, I’ve been at the HPCC07 conference on Houston. I gave a talk and chaired a session there … and was hanging out a lot (of course :)). I’ll just post some commented pictures as I know that you’re all tired of reading:

I lived in the Texas Medical Center (apparently the biggest Medical Center worldwide) … look at those hospitals!!!

Just some of them actually …. there were dozens of those hospital skyscrapers!

and full of puddles like in Bloomington. I felt like home!

but then a policeman with a cowboy-hat … not home anymore. Certainly not!

more indications of being in Texas … they served Mexican food (here Taco chips and Guacamole) at the conference …

Oh yes … the tour guide was standing in front of this picture named “Flowers” and discussed for about 10 minutes that this is her favorite picture. So guys, does this remind you of some earlier talk of some talks given in Chemnitz. I guess somebody has something in common with Texans ;).

More Texas oddities at the sidewalk …. weird people!

Yes, I did really chair a session … was pretty nice :)!

Oh, this was the only techno club in Houston … apparently closed since a while :-(.

That’s all folks!

HPL MPI/NBC Patch

Yes, it is as cryptic as it sounds :). I finally (after a couple of months) finished to merge Christian’s MPI_Bcast() patch with my LibNBC patch to enable the usage of non-blocking collectives in HPL. The performance has to be investigated in more detail, LibNBC will probably provide benefits if lookahead is used. But this needs clearly more investigation. I posted it on the webpage because some people asked me to publish it … so feel free to play around with the patches. I’d also be happy to receive any feedback (yes, even bugs :)).

http://www.unixer.de/research/nbcoll/hpl/

The MPI_Bcast patch seems to break several MPI libraries (e.g. Open MPI (not the trunk 🙂 and MPICH2) because it uses really huge datatypes. It works with MVAPICH and newer Open MPI trunk versions.