-
Notifications
You must be signed in to change notification settings - Fork 38
Integrate changes from NERSC GPU hackathon. #713
Commits on Nov 23, 2021
-
Configuration menu - View commit details
-
Copy full SHA for d452e1a - Browse repository at this point
Copy the full SHA d452e1aView commit details -
[Hackathon] disable a lot of CI (#694)
* Disable cmake-format and clang-format checks. * Disable GitLab CI except for NMODL + GPU.
Configuration menu - View commit details
-
Copy full SHA for 8ab49e9 - Browse repository at this point
Copy the full SHA 8ab49e9View commit details
Commits on Nov 25, 2021
-
[Hackathon] Add a temporary option for benchmark data. (#695)
* Add a hackathon-specific argument for benchmarks. * Add a reference comparison for channel-benchmark.
Configuration menu - View commit details
-
Copy full SHA for 560cc3f - Browse repository at this point
Copy the full SHA 560cc3fView commit details
Commits on Nov 26, 2021
-
Minor changes for building on perlmutter (#697)
* create build/benchmark folder before trying to use it * run nrnivmodl-core in parallel than serially (too slow)
Configuration menu - View commit details
-
Copy full SHA for de4e433 - Browse repository at this point
Copy the full SHA de4e433View commit details
Commits on Nov 29, 2021
-
Configuration menu - View commit details
-
Copy full SHA for 81dd5ef - Browse repository at this point
Copy the full SHA 81dd5efView commit details -
Set by default the number of warps to execute in a large reasonable n…
…umber and update the related documentation (#700)
Configuration menu - View commit details
-
Copy full SHA for 3e394c4 - Browse repository at this point
Copy the full SHA 3e394c4View commit details
Commits on Dec 1, 2021
-
Add memory pool for Random123 streams. (#702)
* Add memory pool for Random123 streams. This speeds up initialisation when running on GPU. * Make Boost optional.
Configuration menu - View commit details
-
Copy full SHA for a8bb716 - Browse repository at this point
Copy the full SHA a8bb716View commit details
Commits on Dec 2, 2021
-
Configuration menu - View commit details
-
Copy full SHA for 9649814 - Browse repository at this point
Copy the full SHA 9649814View commit details
Commits on Dec 7, 2021
-
Basic OpenACC -> OpenMP migration. (#693)
* Simplify unified memory logic. * Pass -mp=gpu when we pass -acc * Pass -gpu=lineinfo for better debug information. * Pass -Minfo=accel,mp for better compile time diagnostics. * Add nrn_pragma_{acc,omp} macros for single-source Open{ACC,MP} support. * Call omp_set_default_device. * Drop cc60 because of OpenMP offload incompatibility. * Add --gpu to test. * Default (BB5-valid) CORENRN_EXTERNAL_BENCHMARK_DATA. * Remove cuda_add_library. * Don't print number of GPUs when quiet. * Set OMP_NUM_THREADS=1 for lfp_test. * Update NMODL to emit nrn_pragma{acc,omp} macros. Co-authored-by: Pramod Kumbhar <[email protected]>
Configuration menu - View commit details
-
Copy full SHA for 21dc2c8 - Browse repository at this point
Copy the full SHA 21dc2c8View commit details
Commits on Dec 9, 2021
-
GPU data management using OpenACC as well as OpenMP API (#704)
* Add wrapper functions for using OpenMP or OpenACC API * Add -mp=gpu in order to link gpu runtime with tests as well * Avoid copying VecPlay members twice otherwise association fails with OpenMP * IvocVect members t_ and y_ were copied twice * only discon_indices_ is pointer and hence that needs to be copied
Configuration menu - View commit details
-
Copy full SHA for 02abf78 - Browse repository at this point
Copy the full SHA 02abf78View commit details
Commits on Dec 10, 2021
-
Christos Kotsalos authored
Dec 10, 2021 Configuration menu - View commit details
-
Copy full SHA for 57f7724 - Browse repository at this point
Copy the full SHA 57f7724View commit details
Commits on Dec 13, 2021
-
Configuration menu - View commit details
-
Copy full SHA for 56889cc - Browse repository at this point
Copy the full SHA 56889ccView commit details -
solve_interleaved2_launcher (CUDA interface) : fixing size of blocksP…
…erGrid & threadsPerBlock (#710)
Christos Kotsalos authoredDec 13, 2021 Configuration menu - View commit details
-
Copy full SHA for 01a39d7 - Browse repository at this point
Copy the full SHA 01a39d7View commit details -
OpenMP offload: use #pragma instead of runtime API (#708)
* Use #pragma omp instead of runtime API in `cnrn_target_{copyin,delete}` * Fix `VecPlayContinuous::discon_indices_` device transfer. * Name `cnrn_target_` wrappers more consistently. Co-authored-by: Olli Lupton <[email protected]>
Configuration menu - View commit details
-
Copy full SHA for 0fe815e - Browse repository at this point
Copy the full SHA 0fe815eView commit details
Commits on Dec 14, 2021
-
We prefer selective host-to-device updates.
Configuration menu - View commit details
-
Copy full SHA for 78081b4 - Browse repository at this point
Copy the full SHA 78081b4View commit details -
Fixes and improvements from LLVM/XLC work. (#716)
Code fixes for XLC and Clang execution without build system changes. This mainly adds missing OpenMP pragmas and makes cnrn_target_ wrappers visible to NMODL.
Configuration menu - View commit details
-
Copy full SHA for 781d34f - Browse repository at this point
Copy the full SHA 781d34fView commit details
Commits on Dec 16, 2021
-
Use pragmas instead of omp_get_mapped_ptr (#705)
omp_get_mapped_ptr was added in OpenMP 5.1 and is not widely supported. With this change then calling cnrn_target_deviceptr on a pointer that is not present on the device is a hard error instead of returning nullptr, so avoid calling it for artificial cells.
Configuration menu - View commit details
-
Copy full SHA for 1f01552 - Browse repository at this point
Copy the full SHA 1f01552View commit details
Commits on Dec 17, 2021
-
GPU implementation improvements (#718)
* Set nwarp to very big number for optimal parallelization and improve a bit grid config of CUDA solve_interleaved2
Configuration menu - View commit details
-
Copy full SHA for d03c45f - Browse repository at this point
Copy the full SHA d03c45fView commit details -
More CI + disable OpenACC in OpenMP builds (#717)
* Re-enable GitLab CI. * Add NMODL + OpenACC test. * Restore {clang,cmake}-format checks. * Prefer OpenACC with MOD2C. * Do not enable OpenACC in NMODL + OpenMP mode. * Convert more #pragma acc to nrn_pragma_acc(...). * Call cudaSetDevice in OpenMP mode. Co-authored-by: Ioannis Magkanaris <[email protected]>
Configuration menu - View commit details
-
Copy full SHA for 3fc7037 - Browse repository at this point
Copy the full SHA 3fc7037View commit details -
Configuration menu - View commit details
-
Copy full SHA for 9a98f73 - Browse repository at this point
Copy the full SHA 9a98f73View commit details
Commits on Dec 21, 2021
-
Configuration menu - View commit details
-
Copy full SHA for 5ce52d5 - Browse repository at this point
Copy the full SHA 5ce52d5View commit details -
Configuration menu - View commit details
-
Copy full SHA for a6c7078 - Browse repository at this point
Copy the full SHA a6c7078View commit details -
Configuration menu - View commit details
-
Copy full SHA for 6b8b6c3 - Browse repository at this point
Copy the full SHA 6b8b6c3View commit details -
Presumably this was working before because our nvhpc localrc files accidentally included CUDA include directories before BlueBrain/spack#1392.
Configuration menu - View commit details
-
Copy full SHA for 531c4fe - Browse repository at this point
Copy the full SHA 531c4feView commit details -
Configuration menu - View commit details
-
Copy full SHA for e3aeafc - Browse repository at this point
Copy the full SHA e3aeafcView commit details
Commits on Dec 22, 2021
-
* Compile NVHPC+Open{ACC,MP} with -cuda. * Pull in NMODL+Eigen fixes to make this work.
Configuration menu - View commit details
-
Copy full SHA for 9fddc7d - Browse repository at this point
Copy the full SHA 9fddc7dView commit details -
Configuration menu - View commit details
-
Copy full SHA for 1fbba17 - Browse repository at this point
Copy the full SHA 1fbba17View commit details -
Configuration menu - View commit details
-
Copy full SHA for 847d415 - Browse repository at this point
Copy the full SHA 847d415View commit details -
Configuration menu - View commit details
-
Copy full SHA for 53b0c5f - Browse repository at this point
Copy the full SHA 53b0c5fView commit details -
Configuration menu - View commit details
-
Copy full SHA for 2c7377c - Browse repository at this point
Copy the full SHA 2c7377cView commit details -
Configuration menu - View commit details
-
Copy full SHA for 5c5b8a3 - Browse repository at this point
Copy the full SHA 5c5b8a3View commit details