Do 3dDeconvolve using residual

Hi all,
I have several questions when I did 3dDeconvolve in afni. I have two runs for each subject and my processing steps are as follows:
1. I input two runs with -input option, so this two runs would be concatenated automatically, here I am confused with the -noblock option, the 3dDeconvolve help file says that if I want to consider two runs as a big run, then I should use -noblock, but I am not clear about what does it mean. In addition, what’s the difference between input multiple runs using -input and -concat?
2. I preprocessed two runs separately, so each run has a motion file. I concatenated two motion files of two runs as one big motion file and insert it as a part of baseline model in 3dDeconvolve. Then I got a residual file for each subject.
3. I tried to use the residual to estimate my stimulus related response. In this step, should I prepare my stim files using global time?(for example, in first run, the stimulus onset time is 20, 80, 160; in second run, it is 40, 90, 150; suppose there are 200 seconds each run, should I prepare my stim file as 20 80 160 240 290 350? )

Looking forward to your response, thanks very much!