Posts by [VENETO] boboviz

1) Message boards : Number crunching : R@H Scientists/Coders: An analysis of the Rosetta binaries... (Message 90744)
Posted 18 days ago by Profile [VENETO] boboviz
Post:
... and GCC is now 8.3 version (with 9.1 ready for release).


Today GCC9.1 is published, with a lot of new features (see my post on Ralph).
Others features and tricks
Link time and inter-procedural
Usability improvements
OpenMP5 support
2) Message boards : Rosetta@home Science : The Audacious Project (Message 90736)
Posted 22 days ago by Profile [VENETO] boboviz
Post:
SSE2 is fine. I just want them to feel guilty, that's all.


If you are interested, there is a very good thread on this argument (with rsj5 clarification on optimization)
Optimization
3) Message boards : Number crunching : R@H Scientists/Coders: An analysis of the Rosetta binaries... (Message 90735)
Posted 22 days ago by Profile [VENETO] boboviz
Post:
Last admin's post of this thread was 2015 :-(
Since these days, Microsoft releases Visual Studio 2019 and GCC is now 8.3 version (with 9.1 ready for release).
4) Message boards : Rosetta@home Science : The Audacious Project (Message 90733)
Posted 22 days ago by Profile [VENETO] boboviz
Post:
The Ryzen 3000 looks interesting. It will have better double precision at least; maybe better AVX2?

You are too ambitious!
I'm satisfied with the sse2 (p.s. SSE2 was released for first time in 2000 from Intel and 2003 from Amd, 16 years ago!!) and it's used in a lot of projects (Tn-grid, Seti, Lhc, etc).
Boinc supports detecting Sse2 (on Windows) since release 6.10.19 (May 2010), so it's not a boinc's infrastructure issue.
But this is an old story
5) Message boards : Rosetta@home Science : The Audacious Project (Message 90729)
Posted 23 days ago by Profile [VENETO] boboviz
Post:
Good stuff. I hope this means they will send us a wider variety of work and more of it.


+1
And more "efficent" apps - aka optimizations :-P
6) Message boards : Number crunching : Rosetta screen saver issue. (Message 90718)
Posted 27 days ago by Profile [VENETO] boboviz
Post:
You have 3 Windows machines with 4 processors. 2 have 4gb of memory and the AMD machine has 8gb of memory.
I would suspect that you are running out of physical memory and paging to disk. Rosetta becomes a memory hog at the checkpoint times and uses 1gb of memory or more.


As usual: the possibility to select this kind of simulations in user's profile (like gpugrid - LARGE wus, LITTLE wus)
7) Message boards : Number crunching : Validators not running (Message 90702)
Posted 21 Apr 2019 by Profile [VENETO] boboviz
Post:
+1. I have over 125 wus pending.

Happy Easter!!
8) Message boards : Rosetta@home Science : The Audacious Project (Message 90689)
Posted 19 Apr 2019 by Profile [VENETO] boboviz
Post:
IPD at Audacious

Protein design receiver $45 milion in founding through The Audacious Project:
The core of this plan is to make IPD the “Bell Labs of Protein Design.” During its heyday in the 1950s, Bell Labs drove the innovations that laid the foundations of the modern digital era.The collaborative environment of Bell Labs attracted top research talent and made it it a hotbed of scientific discovery and technological achievement. Just as Bell Labs helped create the digital revolution, the Institute for Protein Design will work to fuel the protein design revolution. By expanding the Institute, doubling its faculty and training a new generation of scientists, IPD will accelerate the pace of discovery and dissemination of new protein technology
9) Message boards : Number crunching : Rosetta 4.0+ (Message 90682)
Posted 17 Apr 2019 by Profile [VENETO] boboviz
Post:
Again, a lot of "C++ out of memory" error

1068437282
1068437281
1068437279
1068437273
1068437317
etc

Please, fix it
10) Message boards : Cafe Rosetta : proteins----- protein research (Message 90681)
Posted 17 Apr 2019 by Profile [VENETO] boboviz
Post:
https://deepmind.com/blog/alphafold/


And it's open source on github!
11) Message boards : Number crunching : long/large work units, cpu_run_time limit and how to check 'progress'? (Message 90675)
Posted 15 Apr 2019 by Profile [VENETO] boboviz
Post:
errors can be the results themselves, e.g. if a researcher generates lots of aribtrary models and maybe only 1 in 1,000,000 is a model (protein) that would assemble and run to completion, all 999,999 would *run to failure* error and that last 1 in 1,000,000 runs to completion


Errors are results in test projects (like Ralph), cause debugging is welcome.
I'm thinking about technical error like "validation error", "c++(out of memory) error", etc, in production projects, like Rosetta.
12) Message boards : Number crunching : long/large work units, cpu_run_time limit and how to check 'progress'? (Message 90672)
Posted 14 Apr 2019 by Profile [VENETO] boboviz
Post:
Why isn't this new and possibly disruptive work done on RALPH? Seems like RALPH is the place where Rosetta experimentation takes place and not on the main Rosetta@home. The RALPH volunteers are expecting this and it does not disrupt those who don't want to be messed up.


+1.
I crunch on both Ralph and Rosetta.
When i crunch on Ralph i have no problems with crash, errors, etc. It's normal in beta test.
When i crunch on Rosetta i would like stability and no errors.
13) Message boards : Number crunching : long/large work units, cpu_run_time limit and how to check 'progress'? (Message 90664)
Posted 12 Apr 2019 by Profile [VENETO] boboviz
Post:
next time try to do a full proper suspend for all the tasks before you shutdown. that may make a difference
i'm not sure why but for that batch, perhaps i'm lucky, i'm able to continue from that point forwards after restarting


Nope. Restarting from 0% after pause and reboot :-(
14) Message boards : Number crunching : long/large work units, cpu_run_time limit and how to check 'progress'? (Message 90660)
Posted 12 Apr 2019 by Profile [VENETO] boboviz
Post:
Unfortunately, it does not work that way. The task has to reach a point where it can completely store and reload itself. So, it is not possible to call something in the task and command it to take a checkpoint now.


And, after a reboot, all my "_robetta_cstwt_5.0*" restart from 0%
5hs of crunching lost...
15) Message boards : Number crunching : GPU calculation (Message 90656)
Posted 11 Apr 2019 by Profile [VENETO] boboviz
Post:
i'd think the 'gpu intensive ones can literally be a different 'class' of apps. e.g. Tensorflow and or Caffe/Torch
and those apps generates the cpu based algorithms optimizations for rosetta@home itself. i.e. Tensorflow and or Caffe/Torch does all the deep learning data mining and generate likely folds and r@h explore the folds and confirm the folds for unknown molecules / proteins


Today you can "combine" different approach to bionformatical problems, like this:
The application software programs used for the pipeline include:

AutoDock Vina, a molecular docking program
NAMD, a molecular dynamics program
TensorFlow, a machine learning toolkit

Each compute node includes one AMD EPYC processor, 64 GB of memory, one InfiniBand adaptor, and four Radeon Instinct™ MI25 Accelerators, which combine AMD GPU technologies and the ROCm software platform

Cpu+gpu+Ai
We've got some interesting times ahead
16) Message boards : Number crunching : GPU calculation (Message 90654)
Posted 10 Apr 2019 by Profile [VENETO] boboviz
Post:
gpu use may need redesign of the code base a very expensive (in terms of man-effort) task and makes it 'brittle'
i.e. researchers and developers may find it much harder to do new protein algorithms


I'm agree with all your consideration. Not all code is "gpu-able".
But EVERY single step like gpgpu (with Cuda/Opencl), or AI (Tensorflow/Torch), or CNN or "simple" cpu optimization (SSE/Avx) that accelerate the science is welcome!!
17) Message boards : Number crunching : GPU calculation (Message 90636)
Posted 8 Apr 2019 by Profile [VENETO] boboviz
Post:
Some steps towards Gpu (and AI)??
Tensorflow in Rosetta

Add ability to link Tensorflow's C libraries, to allow evaluation of feedforward neural networks on CPU or GPU.
Note that this is _not_ related to (or intended to compete with) the work being done to evaluate the Rosetta scoring function on the GPU using Torch. This is intended to allow evaluation within Rosetta of Tensorflow neural networks that were previously trained outside of Rosetta, for machine learning-related applications.
18) Message boards : Number crunching : Rosetta 4.0+ (Message 90631)
Posted 7 Apr 2019 by Profile [VENETO] boboviz
Post:
But researchers are working on new methods that will eventually get put into production on R@h. I'm not sure about the timeline though.


Come on guys, we are ready for a lot of new science... :-))
19) Message boards : Number crunching : Rosetta 4.0+ (Message 90630)
Posted 7 Apr 2019 by Profile [VENETO] boboviz
Post:
1066755948

after 6hrs of calculation
WARNING! cannot get file size for default.out.gz: could not open file.
Output exists: default.out.gz Size: -1
InternalDecoyCount: 0 (GZ)
-----
0
-----
Stream information inconsistent.
Writing W_0000001
======================================================
DONE :: 1 starting structures 21607.1 cpu seconds
This process generated 1 decoys from 1 attempts
======================================================
10:23:44 (11200): called boinc_finish(0)

</stderr_txt>
<message>
upload failure: <file_xfer_error>
<file_name>rb_04_05_2574_2712__t000__0_C2_SAVE_ALL_OUT_IGNORE_THE_REST_827890_682_0_r1680908808_0</file_name>
<error_code>-240 (stat() failed)</error_code>
</file_xfer_error>
</message>
20) Message boards : Number crunching : long/large work units, cpu_run_time limit and how to check 'progress'? (Message 90618)
Posted 5 Apr 2019 by Profile [VENETO] boboviz
Post:
ok finally it completes after 5 hours of run time, 6 hours elapsed, suspended once in between, no fanfare
http://boinc.bakerlab.org/rosetta/result.php?resultid=1066394376
a single decoy in that 5 hours


Same here on my Xeon
Runtime 2hs, 6hs of calculation, 1 decoy
These are big proteins, i think


Next 20



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