In this demo, all the students were asked to collaborate and compute the total CS courses taken by everyone in the class. An interesting thing I observed in the demo was in the initial few seconds, everyone calculated local sums (sums of courses taken by neighboring students), then one of student from each group communicated the sum to other groups in the same row. This resulted in row-wise sum and which led to addition of sums from different rows. I think the last sum calculated was between the front half and back half of the class (~130 + ~140). Following things can be observed from this demo:
Locality helps and hence natural instinct was to calculate local sums first (and not between 2 people in different corners)
A leader (master) was chosen (although arbitrary in this case ... the person was more likely to be in the middle of the row / class ) for communicating the group sum to the next group.
As we progressed in the computation, more and more number of "threads" (students) were idle and in fact there was a lot more noise due to idle students which might have hindered communication.
This comment was marked helpful 3 times.
mitraraman
A point brought up in another part of the lecture slides claimed that some students were not paying attention which caused them to spend more time calculating their answers. This alludes to some threads that could exist in a program that are estimated to take a certain amount of time to finish the work but perhaps spend more time performing the operation because of idle behavior or other circumstances that could randomly occur. As we could see in the demo, this hindered the ability for the class to efficiently get a correct answer to the problem.
In this demo, all the students were asked to collaborate and compute the total CS courses taken by everyone in the class. An interesting thing I observed in the demo was in the initial few seconds, everyone calculated local sums (sums of courses taken by neighboring students), then one of student from each group communicated the sum to other groups in the same row. This resulted in row-wise sum and which led to addition of sums from different rows. I think the last sum calculated was between the front half and back half of the class (~130 + ~140). Following things can be observed from this demo:
This comment was marked helpful 3 times.
A point brought up in another part of the lecture slides claimed that some students were not paying attention which caused them to spend more time calculating their answers. This alludes to some threads that could exist in a program that are estimated to take a certain amount of time to finish the work but perhaps spend more time performing the operation because of idle behavior or other circumstances that could randomly occur. As we could see in the demo, this hindered the ability for the class to efficiently get a correct answer to the problem.
This comment was marked helpful 1 times.