-
Notifications
You must be signed in to change notification settings - Fork 1
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
run ends with cudaMemcpy() error #33
Comments
Debugging eventually revealed the
in a loop relating to output variables. For one of the variables being output ( I added a conditional check for this value ( A run with Even for a partial run, Will keep you posted on output results and further debug re: We can talk more details next week as well if you like. |
Hum... the E variable is a bit of a special one because it is allocated and freed in the wave step. so it is not a valid pointer if the wave loop is not running and it may be a ghost pointer if it works after the wave step. I't a bit of an ancillary output and I'm not sure I ever output it (H = E8/(rhog) so I output H). I might need to remove it from the output list or directly allocate it once for all. This was done when GPUs add 32Mb of RAM but now memory is cheap and it would remove the overhead of reallocating it every step... |
I have made a new branch called CheapMem where I moved all the mem allocation to the main function and ran a quick test that seems to work. |
Thanks Cyp. Will try this out. |
@CyprienBosserelle should I switch to the |
Hi @CyprienBosserelle, as per today's chat, please find attached the params in question with nx and ny changed as per discussion and flow set to 1.
The text was updated successfully, but these errors were encountered: