search

amkozlov / raxml-ng

RAxML Next Generation: faster, easier-to-use and more flexible
GNU Affero General Public License v3.0
376 stars 62 forks source link

Strange running time? #112

Closed avihani closed 3 years ago

avihani commented 3 years ago

Hi, I have been running a raxml-ng tree construction for about 18 hours now on our lab "server" computer. I wasn't sure how many bootstraps to do, so I ran the following command:

(I used modeltest-ng to identify the "ideal" a.a. substitution model)

raxml-ng -all --msa new_alignment_file2.fasta --model JTT+I+G4 --bs-trees autoMRE

The log file is currently reading:

RAxML-NG v. 1.0.1 released on 19.09.2020 by The Exelixis Lab. Developed by: Alexey M. Kozlov and Alexandros Stamatakis. Contributors: Diego Darriba, Tomas Flouri, Benoit Morel, Sarah Lutteropp, Ben Bettisworth. Latest version: https://github.com/amkozlov/raxml-ng Questions/problems/suggestions? Please visit: https://groups.google.com/forum/#!forum/raxml

System: AMD Opteron(tm) Processor 6348, 24 cores, 62 GB RAM

RAxML-NG was called at 12-Jan-2021 01:24:40 as follows:

raxml-ng -all --msa new_alignment_file2.fasta --model JTT+I+G4 --bs-trees autoMRE

Analysis options: run mode: ML tree search + bootstrapping (Felsenstein Bootstrap) start tree(s): random (10) + parsimony (10) bootstrap replicates: max: 1000 + bootstopping (autoMRE, cutoff: 0.030000) random seed: 1610432680 tip-inner: OFF pattern compression: ON per-rate scalers: OFF site repeats: ON branch lengths: proportional (ML estimate, algorithm: NR-FAST) SIMD kernels: AVX parallelization: coarse-grained (auto), PTHREADS (auto)

[00:00:00] Reading alignment from file: new_alignment_file2.fasta [00:00:00] Loaded alignment with 1107 taxa and 532 sites

Alignment comprises 1 partitions and 457 patterns

Partition 0: noname Model: JTT+I+G4m Alignment sites / patterns: 532 / 457 Gaps: 40.94 % Invariant sites: 24.62 %

NOTE: Binary MSA file created: new_alignment_file2.fasta.raxml.rba

Parallelization scheme autoconfig: 12 worker(s) x 2 thread(s)

Parallel reduction/worker buffer size: 1 KB / 0 KB

[00:00:00] Generating 10 random starting tree(s) with 1107 taxa [00:00:00] Generating 10 parsimony starting tree(s) with 1107 taxa [00:00:19] Data distribution: max. partitions/sites/weight per thread: 1 / 229 / 18320 [00:00:21] Data distribution: max. searches per worker: 86

Starting ML tree search with 20 distinct starting trees

I don't really understand threading vs core performance in reference to software optimizations, so I didn't specify & and let the software decide on it's own. But reading through some issues others have had, incorrectly specifying core/thread count leads to a severe attenuation of the software performance?

Of course, I can run the tree construction with say 100 bootstraps or subsetting data (and I likely will, just to make sure everything else is going OK), but does this lengthy time frame seem consistent with the fasta file I have inputted (descriptions of the fasta file are provided in the log I have described above)?

Edit: Apologies for creating this "issue", the log just printed:

[17:17:59] [worker #10] ML tree search #11, logLikelihood: -302170.878412

Perhaps I perform tree construction on a cluster instead of our struggling computer. Closing this issue.