Closed Ahish-aeon closed 4 years ago
Hi. As stated in GeNN's installation instructions, you'll have to set the GENN_PATH
and CUDA_PATH
environment variables to the respective directories. From your log, it seems that you have instead set them to a list of paths. Guessing from the entries in these lists, try using the following settings:
GENN_PATH=/home/ahish/SNN_MNIST/Brian2GeNN/genn
CUDA_PATH=/usr/local/cuda-10.0
Hi Dr. Marcel, Thank you for your quick reply. I did change the paths and still facing issues with the interfacing. Please find the logfile and brain code attached to this email. Kindly let me know what's going wrong. Hope to hear from you soon. Thank you for your support! brian_debug_oxJfvc.log brian_debug_oxJfvc.txt error.txt genn_trial1.txt
Ah, I see you are using GeNN 4.x. This version did not exist yet when Brian2GeNN was last released, and they are not compatible at the moment. Please install GeNN 3.x instead.
Thank you for the support Dr. Marcel,
I have been using multiple runs in my code, GeNN builds the model every run. This actually slows down the simulation. Kindly let me know if you have any workarounds for this. Thanks again!
Closing this issue since the original problem was about using GeNN 4 which is now supported via #101 .
Sorry about never replying to the last comment: running several builds is indeed slow with Brian2GeNN, there's currently no way to avoid it except for avoiding multiple runs in the first place. Sometimes this is possible e.g. by using a run_regularly
function that sets/resets variables after a certain time.
Hello All,
I have been trying to run a simple brain2 code on GPU using GeNN. However, there are some issues with that. Please help resolve this issue and use GPU capabilities for simulating SNN. I have attached the error files and brain2code, stdout and stderr files were empty. Hope to hear soon. Thank you!
Brian2Cuda.txt Error_Brian2GeNN.txt Error_log1.txt