It seems that the above program only computes red cells and does not mention how to deal with black cells. Did I miss something?
ChandlerBing
Guessing from the note included in the slide, I think there would be a similar procedure for the black cells too.
lament
I believe the comment under the slide title comments on why they are only showing the red update.
sam
It seems to me that the data parallel model can be implemented using ISPC. Is there any other way to implement data parallel model?
Sherry
Is reduceAdd supposed to be smart enough to keep a local version myDiff to avoid locks in each iteration?
kayvonf
@Sherry. That would be up to the implementor of the data-parallel language. Hopefully they do a good job! However, notice that the presence of the high-level primitive provides the opportunity for the language implementor to do so.
It seems that the above program only computes red cells and does not mention how to deal with black cells. Did I miss something?
Guessing from the note included in the slide, I think there would be a similar procedure for the black cells too.
I believe the comment under the slide title comments on why they are only showing the red update.
It seems to me that the data parallel model can be implemented using ISPC. Is there any other way to implement data parallel model?
Is
reduceAddsupposed to be smart enough to keep a local versionmyDiffto avoid locks in each iteration?@Sherry. That would be up to the implementor of the data-parallel language. Hopefully they do a good job! However, notice that the presence of the high-level primitive provides the opportunity for the language implementor to do so.