This application update includes an improved and optimized energy function developed mainly by Frank DiMaio and Hahnbeom Park, and others which improves on all scientific benchmarks. This is an exciting development which we would like to push out and test further. This update also includes new protocols since it uses current Rosetta source. At this point I have included a 64bit linux version but did not compile with SSE optimizations as it will require more testing. I did turn on SSE for the windows app.

Please post bugs/issues here regarding the 3.61 minirosetta_beta app update.

---

Awesome! I have my i7 fully stocked on ralph WUs.

So far 3 errors, 2 of them failed immediately, the other @ 1000 secs. I'm currently running 6 WUs around the 25 min mark (1600 secs) so far with no errors though.

WUs that failed:
https://ralph.bakerlab.org/workunit.php?wuid=3003399
https://ralph.bakerlab.org/workunit.php?wuid=3006069
https://ralph.bakerlab.org/workunit.php?wuid=3006096

Just finished 3 WUs successfully, but there are plenty that fail right at the start:

<core_client_version>7.6.6</core_client_version>
<![CDATA[
<message>
(unknown error) - exit code -1 (0xffffffff)
</message>
<stderr_txt>
[2015- 8-28 16:50:38:] :: BOINC:: Initializing ... ok.
[2015- 8-28 16:50:38:] :: BOINC :: boinc_init()
BOINC:: Setting up shared resources ... ok.
BOINC:: Setting up semaphores ... ok.
BOINC:: Updating status ... ok.
BOINC:: Registering timer callback... ok.
BOINC:: Worker initialized successfully. 
command: projects/ralph.bakerlab.org/minirosetta_beta_3.61_windows_x86_64.exe -run:protocol relax -ex1 -ex2aro -in:file:boinc_wu_zip FFF_939e816d73e5d4a3e6faf001ad588d96_hh5_Wed_Jun__3_15_28_03_PDT_2015.zip -in:file:s 00001.pdb -in:file:native 00001.pdb -in:file:fullatom -relax::default_repeats 15 -out:file:silent default.out -out:file:silent_struct_type score -silent_gz -relax:cyclic_peptide -mute all -nstruct 10000 -cpu_run_time 3600 -checkpoint_interval 120 -database minirosetta_database -in::file::zip minirosetta_database.zip -boinc::watchdog -run::rng mt19937 -constant_seed -jran 3184233
Registering options.. 
Registered extra options.
Initializing broker options ...
Registered extra options.
Initializing core...
Initializing options.... ok 
Options::initialize()
Options::adding_options()
Options::initialize() Check specs.
Options::initialize()  End reached
ERROR: Option matching -relax:cyclic_peptide not found in command line top-level context

</stderr_txt>
]]>

So far I received two work units. One of them finished correctly, the other one failed with the following error:

ERROR: Option matching -relax:cyclic_peptide not found in command line top-level context

.

Yeah, I received this error, too.



Task ID 3404721

<core_client_version>7.4.23</core_client_version>
<![CDATA[
<message>
process exited with code 255 (0xff, -1)
</message>
<stderr_txt>
[2015- 8-29 11:29:13:] :: BOINC:: Initializing ... ok.
[2015- 8-29 11:29:13:] :: BOINC :: boinc_init()
BOINC:: Setting up shared resources ... ok.
BOINC:: Setting up semaphores ... ok.
BOINC:: Updating status ... ok.
BOINC:: Registering timer callback... ok.
BOINC:: Worker initialized successfully.
command: ../../projects/ralph.bakerlab.org/minirosetta_beta_3.61_x86_64-pc-linux-gnu -run:protocol relax -ex1 -ex2aro -in:file:boinc_wu_zip FFF_939e816d73e5d4a3e6faf001ad588d96_hh5_Wed_Jun__3_15_28_03_PDT_2015.zip -in:file:s 00001.pdb -in:file:native 00001.pdb -in:file:fullatom -relax::default_repeats 15 -out:file:silent default.out -out:file:silent_struct_type score -silent_gz -relax:cyclic_peptide -mute all -nstruct 10000 -cpu_run_time 3600 -checkpoint_interval 120 -database minirosetta_database -in::file::zip minirosetta_database.zip -boinc::watchdog -run::rng mt19937 -constant_seed -jran 3184272
Registering options..
Registered extra options.
Initializing broker options ...
Registered extra options.
Initializing core...
Initializing options.... ok
Options::initialize()
Options::adding_options()
Options::initialize() Check specs.
Options::initialize() End reached
ERROR: Option matching -relax:cyclic_peptide not found in command line top-level context

</stderr_txt>
]]>

Yup. Most are good on my linux hosts, with a handful of immediate compute errors.

stderr out

<core_client_version>7.4.22</core_client_version>
<![CDATA[
<message>
process exited with code 255 (0xff, -1)
</message>
<stderr_txt>
...
command: ../../projects/ralph.bakerlab.org/minirosetta_beta_3.61_x86_64-pc-linux-gnu -run:protocol relax -ex1 -ex2aro -in:file:boinc_wu_zip FFF_f9e68a71c8f649cfc9b9f525ee9b6a38_hh9_gb_Wed_Jun_10_18_48_55_PDT_2015.zip -in:file:s 00001.pdb -in:file:native 00001.pdb -in:file:fullatom -relax::default_repeats 15 -relax:cyclic_peptide -score::weights talaris2013_cst_modified.wts -out:file:silent default.out -out:file:silent_struct_type score -silent_gz -mute all -nstruct 10000 -cpu_run_time 3600 -checkpoint_interval 120 -database minirosetta_database -in::file::zip minirosetta_database.zip -boinc::watchdog -run::rng mt19937 -constant_seed -jran 3171761
Registering options..
ERROR: Option matching -relax:cyclic_peptide not found in command line top-level context

My guess is something strange in the configuration on those particular WU's. Maybe the -relax:cyclic_peptide option is missing a required parameter there?

At this point I have included a 64bit linux version but did not compile with SSE optimizations as it will require more testing. I did turn on SSE for the windows app.

From rosetta forum:

1) Please, insert brief description of app in Applications Page, like this.

Will do.

For the 3.62 update, I reverted the windows application back to the 32bit non-sse build, and also reverted the linux application to the 32bit build (for the 64bit apps). There was a significant fraction of errors which prompted the change. We will do another update soon related to the "Option matching -relax:cyclic_peptide not found in command line top-level context" errors.

---

For the 3.62 update, I reverted the windows application back to the 32bit non-sse build, and also reverted the linux application to the 32bit build (for the 64bit apps). There was a significant fraction of errors which prompted the change. We will do another update soon related to the "Option matching -relax:cyclic_peptide not found in command line top-level context" errors.

Errors in addition to the "Option matching -relax:cyclic_peptide not found in command line top-level context" ??

So far, so good on 64-bit Linux.

For the 3.62 update, I reverted the windows application back to the 32bit non-sse build, and also reverted the linux application to the 32bit build (for the 64bit apps). There was a significant fraction of errors which prompted the change. We will do another update soon related to the "Option matching -relax:cyclic_peptide not found in command line top-level context" errors.

on Linux-64 got some

ERROR: Option matching -abinitio:cyclic_peptide not found in command line top-level context

where the error string is a little bit different to yours
eg this workunit

on Linux-64 got some

ERROR: Option matching -abinitio:cyclic_peptide not found in command line top-level context

where the error string is a little bit different to yours
eg this workunit

Seems to be the same error of 3.54 version

Hello,

As a general software engineering/debugging principal, I would not recommend mixing cpu optimizations and new code in the same build. If a WU fails, what caused it to fail? The new code or one of the optimizations?

Instead, make as many branches as you need off the branch of the latest stable build in production, and start testing the optimizations one at a time. Test against work units that have already been run successfully on the stable code base without the optimizations.

By following this procedure, you will only be testing one change at a time, for instance enabling SSE or SSE2. So, debugging will be far more straightforward in the case of any WU failures.

Cheers,

Ed

The protocol code for the majority of jobs (and all the test jobs) hasn't changed. BUT, for this cyclic peptide error, the developer did not check his code into the main source repo as he should have long ago. In an effort to save time a while ago, we incorporated his code before he checked it in under the understanding he would check it in soon after, which he didn't. But on a positive note, he is busy trying to make and characterize his cyclic peptide designs now.

---

The protocol code for the majority of jobs (and all the test jobs) hasn't changed. BUT, for this cyclic peptide error, the developer did not check his code into the main source repo as he should have long ago. In an effort to save time a while ago, we incorporated his code before he checked it in under the understanding he would check it in soon after, which he didn't. But on a positive note, he is busy trying to make and characterize his cyclic peptide designs now.

What kind of errors was the SSE extension giving then?

Remember that improving performance directly improves scientific benchmarks. There is lots of compute power in RALPH to bug-out any problems that come from using SSE/SSE2 (thought technically there shouldn't be any, since 99% of users are using SSE2 capable CPUs).

Thanks, David.
So, to be clear, would the following release notes reflect the version of 3.62 that went to production?

Improvements:
- Improved and optimized energy function developed mainly by Frank DiMaio and Hahnbeom Park and others which improves on all scientific benchmarks
- This update also includes new protocols since it uses current Rosetta source (Ed - What are the *new* protocols that were not in 3.59?)

Not Changed:
- reverted the windows application back to the 32bit non-sse build
- reverted the linux application to the 32bit build (for the 64bit apps), non-sse build

Not Changed:
- reverted the windows application back to the 32bit non-sse build
- reverted the linux application to the 32bit build (for the 64bit apps), non-sse build

I cannot understand if rosetta's admins are working together with rjs5, sesef or other volunteers expert on optimization.

Any news regarding the 3.63 version?

---

Any news regarding the 3.63 version?

New exe, new screensaver, new db.
Optimizations or only bugfixes?

In Linux libglut.so.3 errors all over the place, perhaps looking in wrong directory?