Posts by Rocco Moretti

1) Message boards : Feedback : Higher quorum (Message 5526)
Posted 13 Apr 2012 by Rocco Moretti
Post:
I thought tasks using the same starting point (same random seed) would produce the same result, provided the machines return the same number of models.


Theoretically, yes, the same seed should produce the same results. Unfortunately, because of slight differences in processors and timing, etc., that's not always the case. Because of the way protocols run, slight numerical differences can change the trajectory significantly (i.e. the butterfly effect).

Because we're not really interested in producing a "canonical" result for each workunit (it's more important to produce a valid distribution across the entire job), and the main thing Ralph is intended to catch is things which cause the application to error out, it's not really worth trying compare the same workunit between different machines.
2) Message boards : Feedback : Higher quorum (Message 5524)
Posted 12 Apr 2012 by Rocco Moretti
Post:
Rosetta (and Ralph) uses a protocol which is stochastic, rather than deterministic. This means that unlike many other projects, there isn't a single "canonical" result for each workunit - the same workunit will produce slightly different results when run on different machines, or even re-run by the same machine. Therefore, there is little value in trying to compare two different runs of a workunit against each other.

The primary goal with the testing on Ralph is to check that there aren't any subtle bugs with the job or the application which would result in jobs terminating with errors on certain platforms or certain small percentages of cases.
3) Message boards : RALPH@home bug list : MiniRosetta Beta 3.26 (Message 5521)
Posted 10 Apr 2012 by Rocco Moretti
Post:
18,800 seconds for 1.95 credits, that's incredible:

Any idea what went so terribly wrong?


From the stderr out, it looks like your boinc client actually ran the executable twice. Once for 99 decoys, and the second time for just a single decoy - the output file of which likely overwrote the output file of the first time around. This means that although you crunched for 100 decoys worth of time, you only sent back (and got credit for) one decoy.

Why boinc re-ran the minirosetta application, I don't know - it might have something to do with the "No heartbeat from core client for 30 sec - exiting" line. If I had to guess, your Boinc manager was not running or unresponsive when the minirosetta application finished, so it didn't recognize that it was done, causing it to restart it and overwrite the results.
4) Message boards : Number crunching : Ralph support multi-thread? (Message 5483)
Posted 25 Mar 2012 by Rocco Moretti
Post:
Multi-threading would be a headache to support for no practical gain. Protocols typically function by taking a single structure through a series of steps, which means there's choke points which would limit the speed increase you'd get from multithreading. Rosetta is really set up for trivial parallelism. Instead of multithreading a single structure, the better way to handle multiple CPUs is to run multiple structures at the same time, and the easiest way of doing that is to run multiple independent instances of the program. (Which happens to be a good match for the boinc setup.)

If you want to make full use of your 8-core CPU, go into your boinc computing preferences, and increase the "On multiprocessors, use at most" settings. This way you should crunch on eight work units at once (one for each CPU). Although each will run at about the same speed as a single work unit on a single CPU, you'll still come out ahead credit/scientific-result-wise of if we implemented multi-threading.
5) Message boards : RALPH@home bug list : Rosetta@Home Version 3.23 (Message 5475)
Posted 11 Mar 2012 by Rocco Moretti
Post:
1) If it is "cutting edge", why 3.23? I think it's a 3.50 or 4.0 release...


From a software perspective, the change wasn't all that great. The general framework of the minirosetta client is still the same, it's just we added a bit of new functionality on top of that. Most updates to Rosetta@home are like that - they're not so much to increase stability or fix bugs (although we do do that), but rather to implement new scientific functionality.

In some respects, version numbering is meaningless. As long as different versions have different numbers, there really isn't a uniform standard on how to do software versioning. For the boinc minirosetta application, we're currently using a major version of 3 (as it's based on the Rosetta3 code) and a minor version which increments by one on each new release. As boinc updates the code automatically, a more sophisticated versioning scheme really isn't necessary. Run Rosetta@home, and you'll get the latest-and-greatest, whether that's called 3.23, 4.0 or XF2012

2) "we had to change how we build"...have you change the compiler? Is this a "prelude" for gpu client? :-P


Sorry, a little more prosaic. We had hardware issues on the system we normally use to build the client, so we had to set up a new build system.
6) Message boards : RALPH@home bug list : Rosetta@Home Version 3.23 (Message 5472)
Posted 9 Mar 2012 by Rocco Moretti
Post:
3.22 seems to be very stable.
With 3.23 do you rewrite all the binaries?? :-)


The reason for the update is new & improved protocols. We're testing out new ways of doing things (particularly in the area of structure prediction), and 3.23 simply can't do these new protocols. It's cutting edge science (we hope), so we need to update the application to handle it.

The "alpha in the truest sense of the word" bit is due to the fact that, for technical reasons, we've had to change how we build the minirosetta application binaries. While we don't anticipate that there will be significant issues due to the change, we're aware that there could be unforeseen issues.
7) Message boards : RALPH@home bug list : Rosetta Mini Beta 3.20 (Message 5442)
Posted 14 Jan 2012 by Rocco Moretti
Post:
This update should fix the graphical client issues some users were reporting on Rosetta@home. (Along with other general updates & fixes.)

Please post issues and bugs here.
8) Message boards : RALPH@home bug list : Rosetta mini 3.18 (Message 5408)
Posted 1 Nov 2011 by Rocco Moretti
Post:
New version of rosetta mini. New function or a simply debug?


As mentioned by cmiles over in the Rosetta@home forums, minirosetta 3.18 is identical to minirosetta_beta 3.17, and is identical to the version (3.17) currently being run on Rosetta@home. (The difference in numbering is because of technical reasons.)

9) Message boards : Current tests : Ralph & Rosetta optimized for GPU CUDA ? (Message 5388)
Posted 9 Sep 2011 by Rocco Moretti
Post:
I don't think anyone would argue that speeding up Ralph/Rosetta@home would be anything but a positive.

That said, from what I understand, GPU programing is *hard*, especially if you don't have a naturally parallel application. SETI@Home has it a bit easy, as Fourier transforms/signal processing matches GPU processors very well.

In contrast, because of the way Rosetta works internally, it's non-trivial to convert to GPU programing style. From what I understand, while bits and pieces might be easily moved to a GPU, you'd be killed by the data transfer to and from the GPU when switching between GPU-using sections and CPU-using sections. My sketchy understanding is that on several occasions groups with experience doing GPU programming have taken a look at the Rosetta code with little to nothing ultimately coming of it.

That said, one of the researchers in the lab was/is working on a "skunkworks"-type project regarding incorporating the use of GPUs into Rosetta - though I hesitate to mention this because it's very uncertain if anything will actually come of it, or if it does, when. Even if there are parts of Rosetta that are GPU-ized, there's no guarantee that it would be those parts which would make a significant difference in the typical runs performed on Rosetta@Home.

Short answer: No one would refuse a GPU version of Rosetta, but the considered opinion is that making it would likely be a lot of effort for not all that much actual speedup.
10) Message boards : Current tests : A lot of wu....use it for science? (Message 5365)
Posted 29 Jul 2011 by Rocco Moretti
Post:
Ralph is the "alpha" site for Rosetta@home - it's where we test new versions of the executable, and also where we do small-scale tests of jobs before we send it out to R@h, to make sure there aren't any issue with them. There really isn't a distinction between the type of science on Ralph vs. Rosetta@home.

If we're not testing a new executable version or if no one has new jobs to test, there won't be much activity on Ralph. On the other hand, if we're testing an executable or if people are starting a lot of new jobs, there will be a bunch of work units.

11) Message boards : RALPH@home bug list : Minirosetta Beta 3.14 (Message 5345)
Posted 5 Jul 2011 by Rocco Moretti
Post:
ERROR: unrecognized aa LIG


Sorry about that - there was a file missing from the input files. It should be corrected in newer submissions.

ERROR: Cannot open PDB file "2p9hA_suc_0001.pdb"


A different input file issue - also should be corrected with newer submissions.

--

(I will double check my input files before submitting.
I will double check my input files before submitting.
I will double check my input files before submitting. ...)






©2019 University of Washington
http://www.bakerlab.org