Pramod Kumbhar
Pramod Kumbhar
- [ ] Generating kernel with `./bin/nmodl hh.mod llvm --ir` but still `InstanceStruct` generation and overall cleaning missing here.
Consider a sample MOD file (provided by @alkino): ```python NEURON { SUFFIX glia } STATE { C1 C2 } BREAKPOINT { SOLVE kstates METHOD sparse } FUNCTION alfa(v(mV)) { alfa...
Current implementation of `net_receive` printing routine : ```c++ void CodegenCVisitor::print_net_receive_common_code(Block* node, bool need_mech_inst) { .... /// rename variables if they are actually used auto parameters = info.net_receive_node->get_parameters(); if (!parameters.empty()) {...
Multiple users have seen very high compilation times. We need to look into what is taking time to build the NMODL project. We should break the build time to make...
See #43. We have been introducing more spaghetti in the code generation passes. This needs to be fixes as soon as ISPC backend and solvers are integrated/tested.
In https://github.com/BlueBrain/nmodl/pull/531 we want to create simple expression like below for the `for` loop initialisation: ``` id = 0 ``` Here we expect `0` to be an integer. But currently...
Some of the constructs in MOD file are non-thread safe (e.g. LINEAR, NON-LINEAR, DISCRETE, PARTIAL). In this case, MOD2C/NEURON generate code with : ```cpp /* Created by Language version: 7.5.0...
See the https://github.com/BlueBrain/nmodl/pull/645#issuecomment-1095075632: ``` libc++abi: terminating with uncaught exception of type std::runtime_error: Error: only assignment is supported for CodegenAtomicStatement The reason is that mech->ion_ina[ion_ina_id] += mech->ina[id] uses + with CodegenAtomicStatement....
Here is summary of issues I see while building coreneuron+nmodl with XL compiler: System details : [Marconi100](https://www.hpc.cineca.it/hardware/marconi100) Modules: ```console $ module list Currently Loaded Modulefiles: 1) profile/base 2) python/3.8.2 3)...
Today compilers like NVHPC produces tooooooo many warnings while building NMODL. Not necessary to fix all warnings and some can be just ignored. As part of this ticket, we should...