search

choderalab / ensembler

Automated omics-scale protein modeling and simulation setup.
http://ensembler.readthedocs.io/
GNU General Public License v2.0
52 stars 21 forks source link

Timings for various stages? #8

Closed jchodera closed 9 years ago

jchodera commented 9 years ago

What are the current timings per model for various stages of the TK pipeline?

danielparton commented 9 years ago

The Modeller step takes about 45 s for each model. The implicit solvent MD step takes about 160 s for each model.

On Wed, Oct 8, 2014 at 10:08 AM, John Chodera notifications@github.com wrote:

What are the current timings per model for various stages of the TK pipeline?

— Reply to this email directly or view it on GitHub https://github.com/choderalab/msmseeder/issues/8#issuecomment-58362624.

jchodera commented 9 years ago

Are these the only time-consuming steps?

Would we then approximate the total work as total = ntargets * ntemplates * (45 + 160)?

For ntargets = 500, ntemplates = 4000, that gives us ~4750 GPU-days.

We would need to trim this down somehow. Even if we eliminate the implicit solvent step, that's ~1050 GPU-days. That suggests we may need to restrict the model building to some minimal level of sequence identity.

danielparton commented 9 years ago

These should be the major bottlenecks. There is also the explicit solvent MD step, which we are skipping at the moment. I'll look into restricting model building on the basis of sequence identity, as well as possibly reducing the length of the implicit solvent MD simulation.

On Wed, Oct 8, 2014 at 5:47 PM, John Chodera notifications@github.com wrote:

Are these the only time-consuming steps?

Would we then approximate the total work as total = ntargets * ntemplates

  • (45 + 160)?

For ntargets = 500, ntemplates = 4000, that gives us ~4750 GPU-days.

We would need to trim this down somehow. Even if we eliminate the implicit solvent step, that's ~1050 GPU-days. That suggests we may need to restrict the model building to some minimal level of sequence identity.

— Reply to this email directly or view it on GitHub https://github.com/choderalab/msmseeder/issues/8#issuecomment-58433236.

jchodera commented 9 years ago

The figure now compiles all of this information.