search

DeltaGroupNJUPT / Vina-GPU-2.0

Vina-GPU 2.0 accelerates AutoDock Vina and its related commonly derived docking methods, such as QuickVina 2 and QuickVina-W with GPUs.
Apache License 2.0
93 stars 30 forks source link

Stack execution errors/ Compilation Errors #36

Open StephanWatkins opened 10 months ago

StephanWatkins commented 10 months ago

I am able to get Vina2.0-GPU and QuickVina2-GPU to compile, and both are able to run the test sets included. However when I run them in batch, or a single run from the batch in a shell, I get segfaults, and can not figure out why. I wondered if anyone else had the same problem, and if there was a solution?

What I have tried, setting in the batch or command shell the ulimit -s to 16, 32 MB all the way up to the max ram available, the device has 6, the PC 64 Gb

My output $ Reading input ... done. Setting up the scoring function ... done. Using heuristic search_depth Analyzing the binding site ... done. GPU Platform: NVIDIA CUDA GPU Device: NVIDIA GeForce RTX 2060 Using random seed: -1196833273

Build kernel 1 from source OpenCL version: 3.0 Build kernel 2 from source OpenCL version: 3.0 Vina-GPU: ./lib/main_procedure_cl.cpp:477: void main_procedure_cl(cache&, const std::vector&, const precalculate&, parallel_mc, const vec&, const vec&, int, std::vector<boost::ptr_vector >&): Assertion `m.num_other_pairs() == 0' failed. Aborted (core dumped)

The only changes to the Make file were the locations of the thread.cpp and once.cpp (thread.hpp and once.hpp) which are installed one directory lower in boost/thread/ . of coarse their the .hpp, not .cpp.

I also played on either, as it said W had hard limits in box size using 25x25x25 and 30x25x30, The number of atoms in the lignads are around 60-70 on average however if I use 30 atom ligands it is the same, but still the test sets run.

if i run make clean make

all that happens is the ""Vina-GPU: ./lib/main_procedure_cl.cpp:477: void main_procedure_cl(cache&, const std::vector&, const precalculate&, parallel_mc, const vec&, const vec&, int, std::vector<boost::ptr_vector >&): Assertion `m.num_other_pairs() == 0' failed.""

is missing in the trace and it just says Aborted (core dumped)

?

StephanWatkins commented 10 months ago

its the use of --flex in the input, I was unaware you could not use flexible receptor amino acids...this after trying to trace the error ten million ways...and installing boost 1.83 from source, so now the kernals run good.