Ok, then letās check your hardware
- Could you please try to execute the workflow on your Intel HD GPU? I assume your laptop has one:
run("CLIJ2 Macro Extensions", "cl_device=HD");
- Could you please check if the NVidia driver installed on your machine is recent? According to the website it should be approximately version 451.48
https://www.nvidia.com/Download/driverResults.aspx/161304/en-us - Could you please run CLInfo and copy-paste the log output here?
run("CLIJ2 Macro Extensions", "cl_device=");
Ext.CLIJ_clInfo();
No problem, as images are variables:
Ext.CLIJ2_push(input);
if (user_input) {
Ext.CLIJ2_gaussianBlur2D(input, output, 1, 1);
} else {
output = input;
}
Ext.CLIJ2_pull(output);
Some āapparent inplaceā methods still exist in CLIJx:
run("CLIJ2 Macro Extensions", "cl_device=");
Ext.CLIJ2_help("inplace");
Having a way for doing what you want in a more systematic may end up on the roadmap for clEsperanto as python users demand it. Eventually code will likely look like this:
image = gaussianBlur(image, 2);
Until they are generally availabe, some details need to be streamlined under the hood. In the meantime, you can implement such methods yourself if you like:
function gaussianBlur(input, sigma) {
Ext.CLIJ2_gaussianBlur2D(input, output, sigma, sigma);
return output;
}
Intransparent ārenamingā might confuse users. However, I see where you want to go and can promise that weāre moving in the same direction
Let me know if the driver thingy solves your GPU issue.
Cheers,
Robert