Hi All,
Please remember to read the filesystem papers before the meeting today!! To help you, please answer the following three questions via email directly to me (don’t post the answer on the blog).
- What is the ciod in the BG/P paper? What service does it provide or role does it play in the system?
- Does BG/P run Linux on the PowerPC 450s?
- How do object-based storage systems get higher bandwidth than server-based storage systems even though both use centralized servers for metadata?
Also, please post your status as a response to this blog entry.
See ya at 3:30 … same Bat Time, same Bat Cave.
Ron
Classes, classes, classes! Embedded systems is taking more of my time than I would like but it’s going well. I’m learning Latex and Linux better and Andy and I are getting a good handle on the networking and optimization part of the DBES.
I’ve signed up for the breadth qualifier and am going to sign up for the depth but I need to talk to you about my committee. I’m getting my Plan of Study together as well.
I’m going to Charleston this Friday for the innovation project awards.
Last week:
finished testing the core with a standalone application, NOT NAMD.
tested core with values pulled out of NAMD’s flow, works correctly.
completed the first draft for 3 chapters so far, waiting on Ron’s and Robin’s comments.
Started with the outline evaluation chapter
This week:
actually put the core into NAMD and run a few tests
Complete the evaluation chapter and start with the conclusion
Make corrections to what’s written so far
Finishing up the Native Port Interface updates. Ran into a few problems where the MPMC stops responding. It is a weird bug that has stumped me for about half a week (and still is frustrating me). I also want to finish testing it with the MPMC simulation I built so that Sid can use it for his NAMD tests.
I spoke with Jason about the Leon processor and he is not pursuing it due to the overall size, speed and lack of integration with existing EDK cores. If someone is interested in continuing where Ashwin and I left off, let us know, it is just a matter of getting the SW to run and it could be something that initially is done via on-chip memory only.
Started working on conjugate gradient hardware.
Have to help Liu with GPU code.
Just finished testing single board Barrier.
Ready to do multi board test.
Reduce should not be hard because we
can use most of Barrier’s code. If things
go smoothly, we can have reduce this week.
I benchmarked AIREN driver on our custom network
and saw a packet loss when the loads scaled up.
I plan to work on this issue this week.
Last Week
1. Was working on MPI code for various atom configurations,
2. Plotted speed and time results for different atom configurations
This week
1. Working on to get the APU – FCM core working on Linux 2.6
this week:
1. Finish up the intorduction and related work (gate3) of the blast journal.
2. Still working on the (N X 1) pthread implementation of BLAST. The (1 X M) configuration using pthreads is done.