A few days to normalise over 1000 subjects is not an especially long time. I'd suggest trying to figure out what went wrong and running things again. If you used the T1w to drive the spatial normalisation, then check that these are in good alignment with the fMRI (Check Reg button). My own experience of the ABIDE dataset is that they are pretty messy, with a lot of cereballa cut off some of the positioning information a bit unreliable. Most of the T1w images can be normalised pretty easily though, so success is likely to depend on how accurately the T1w can be aligned with the fMRI.
Best regards,
-John