Hi,
> I am just curious is it possible to use a value between 1 and 2 for
this option, for example "--pad 1.5", to both save memory and avoid
artifacts?
This is a good idea but currently not possible.
> some obvious artifacts on the edges on the reconstruction
If the artifact is at the corners and outside the mask,
you don't have to worry.
If the artifact at the corners overlaps with your molecule,
your box is probably too tight. You should extract in a larger box.
This is essentially the same as your suggestion of padding by a
value between 1 and 2.
By the way, with such a large particle, you might want to try
Ewald sphere correction with relion_reconstruct.
Since it uses only CPU and not limited by GPU memory, you can use
--pad 2 (default) there.
Best regards,
Takanori Nakane
On 2018/11/09 20:30, Jerry Wang wrote:
> Hi All
>
> I am working on a large particle dataset which has 800X800 box size, and then as expected at finer angular sampling the GPU is out of memory. I understand using "--pad 1" or "skip padding" will save memory and speed up the process. But then there is some obvious artifacts on the edges on the reconstruction volume.
>
> I am just curious is it possible to use a value between 1 and 2 for this option, for example "--pad 1.5", to both save memory and avoid artifacts? And does Relion support non-integer as input for this option?
>
> Many thanks!
> Regards,
> Jerry
>
> ########################################################################
>
> To unsubscribe from the CCPEM list, click the following link:
> https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCPEM&A=1
>
########################################################################
To unsubscribe from the CCPEM list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCPEM&A=1
|