Re: [E] Re: parallel aggregation

2023-04-13 Thread Alexander Saydakov
question remains how to make finalfunc understand > both states: directly from sfunc if there is no partial aggregation, and > from the combiner. > > Can a combiner receive results of another combiner? Can a combiner > output also go through serial-deserial? > > You have to r

Re: [E] Re: parallel aggregation

2023-04-12 Thread David Rowley
Not currently. However, I *can* think of reasons why that might change in the future. If we wanted to accept partially aggregated results from foreign servers and then combine them locally then, if those foreign servers did parallel aggregation, the foreign server's combined states would need t

Re: [E] Re: parallel aggregation

2023-04-12 Thread David Rowley
gh serial-deserial? You have to remember that the final func can be called without the state ever having gone through the combine func. This is what happens in non-parallel aggregation. Also, think of the case with > 2 parallel workers. There might be more than 2 states to combine for any give

Re: parallel aggregation

2023-04-12 Thread David Rowley
On Wed, 12 Apr 2023 at 22:14, Alexander Saydakov wrote: > > I have a few questions regarding aggregate functions that would be parallel > safe. > 1. do the inputs of combinefunc always go through serialfunc-deserialfunc or > they can come directly from sfunc (worker on the same machine, perhaps)

parallel aggregation

2023-04-12 Thread Alexander Saydakov
I have a few questions regarding aggregate functions that would be parallel safe. 1. do the inputs of combinefunc always go through serialfunc-deserialfunc or they can come directly from sfunc (worker on the same machine, perhaps)? 2. can the result of combinefunc ever be fed to sfunc as opposed to