Open pr4deepr opened 4 years ago
haesleinhuepf
commented
4 years ago Hey @pr4deepr ,
can you check if the issue persists after trying the solution explained here? https://clij.github.io/clij2-docs/troubleshooting#macos_login
Thanks!
Cheers, Robert
pr4deepr
commented
4 years ago If its the energy saver mode and automatic graphics switching, yes I tried it. Came across the imagej forum post. The laptop doesn't have this option as I think its only got one graphics unit??
haesleinhuepf
commented
4 years ago Hm, I'm afraid, there is not much I can do about this.
Is there a better way to handle this error, especially as all of FIJI crashes?
It's the operating system which kills Fiji. We can't catch that.
I need to refer to the hardware section of the installation instructions: Buy a proper GPU if you want to do GPU-accelerated image processing ;-) https://clij.github.io/assistant/installation#hardware
pr4deepr
commented
4 years ago It's the operating system which kills Fiji. We can't catch that.
Good to know.
I need to refer to the hardware section of the installation instructions: Buy a proper GPU if you want to do GPU-accelerated image processing ;-) https://clij.github.io/assistant/installation#hardware
Yea, fair enough. Thats pretty detailed. 😄
In my case, running most filters worked on the 1.5GB integrated card, but median was the one causing issues. Assuming its to do with the compute capabilities??? Curious to know why its the case. Also, Is it possible to mention the filters that may cause issues on lower end cards or cards just at the limit where CLIJ2/x works?
With most common use cases or users you have interacted with, do they usually have quite high graphics card specs? I know its becoming increasingly common to have dedicated GPUs..
haesleinhuepf
commented
4 years ago In my case, running most filters worked on the 1.5GB integrated card, but median was the one causing issues. Assuming its to do with the compute capabilities??? Curious to know why its the case.
Me too! ;-)
Also, Is it possible to mention the filters that may cause issues on lower end cards or cards just at the limit where CLIJ2/x works?
Before I warn users of specific hardware, I would like to be sure that the hardware is causing the issue. E.g. a median 3D filter on 1024x1024x30 works on an Intel GPU with 1.6 GB memory. I just tried:
You can also see that it doesn't use the full memory. Thus, question: What parameters did you enter when applying the filter?
pr4deepr
commented
4 years ago What parameters did you enter when applying the filter?
Yea, it wasn't filling up the memory, which was why I got confused. I used a 2x2x2 kernel and a 3x3x3 kernel. I may try a few different parameters, kernel sizes, image sizes and get back to you on this.
Hi So, I was trying a workflow on a MacPro notebook which has 1.5GB GPU memory (Intel Graphics; inbuilt I think). It was a bunch of CLIJ2 workflows together and FIJI kept crashing. I narrowed it down to the fact that was the median filter causing it (1024x1024x30, 16-bit image). Smaller images seemed to work ok, but it was quite variable. I have attached an error log of testing out the CLIJ2_median filter at the end of this post. I am using the median 3D filter in FIJI instead for now (under Filters; can also use the one under 3D imageJ suite for now)
I had the same issue happen when I used the labelling ALX workflow. FIJI crashes.
Is there a better way to handle this error, especially as all of FIJI crashes?