Embarrassingly Parallel Time Series Analysis for Large Scale Weak Memory Systems

#1

Post by zeit_geist » Sat, Nov 28, 2015, 9:12 AM UTC

Embarrassingly Parallel Time Series Analysis for Large Scale Weak Memory Systems
arxiv.org

Re: Embarrassingly Parallel Time Series Analysis for Large Scale Weak Memory Systems

#2

Post by SFjulie1 » Sat, Nov 28, 2015, 1:24 PM UTC

Isn't this lengthy thesis about saying that doing a map reduce on consecutive data that have are hot and localized in memory is all the better that we use commutative/distributive operations. Well it was trivial.

Well, I would have prefered a first year student demonstrating the trivia that operation like sorting multidimensional vector will kick by the nature less optimization possible in CPU and GPU and that non linear operation requires reading the whole sample to have a non random quantifiable error.

Hence, map reduce performance is better for combination of distributive linear operation (like ARMA, and I still wonder about hilbertian geometry) and horrible for non linear mapped functions (median filter, nth percentile, sort, top X).

Re: Embarrassingly Parallel Time Series Analysis for Large Scale Weak Memory Systems

#3

Post by pmontra » Sat, Nov 28, 2015, 1:53 PM UTC

I skimmed through the paper. It's packed with formulas and technical details. I can't judge if it's a relevant contribution to the subject but... the title. Do they need that to be noticed? Just imagine "Embarrassingly Fast Electromagnetic Waves but not any Faster" instead of "On the Electrodynamics of Moving Bodies" (Zur Elektrodynamik bewegter Körper) http://.ca/wiki/Zur_Elektrodynamik_bewegter_Körper

Re: Embarrassingly Parallel Time Series Analysis for Large Scale Weak Memory Systems

#4

Post by sethev » Sat, Nov 28, 2015, 2:10 PM UTC

pmontra wrote: post #3

I skimmed through the paper. It's packed with formulas and technical details. I can't judge if it's a relevant contribution to the subject but... the title. Do they need that to be noticed? Just imagine "Embarrassingly Fast Electromagnetic Waves but not any Faster" instead of "On the Electrodynamics of Moving Bodies" (Zur Elektrodynamik bewegter Körper) http://.ca/wiki/Zur_Elektrodynamik_bewegter_Körper

"Embarrassingly parallel" isn't a term that they invented: https://en.wikipedia.org/wiki/Embarrassingly_parallel

Re: Embarrassingly Parallel Time Series Analysis for Large Scale Weak Memory Systems

#5

Post by pmontra » Sat, Nov 28, 2015, 2:36 PM UTC

sethev wrote: post #4

pmontra wrote: post #3

I skimmed through the paper. It's packed with formulas and technical details. I can't judge if it's a relevant contribution to the subject but... the title. Do they need that to be noticed? Just imagine "Embarrassingly Fast Electromagnetic Waves but not any Faster" instead of "On the Electrodynamics of Moving Bodies" (Zur Elektrodynamik bewegter Körper) http://.ca/wiki/Zur_Elektrodynamik_bewegter_Körper

"Embarrassingly parallel" isn't a term that they invented: https://en.wikipedia.org/wiki/Embarrassingly_parallel

Thank you and my upvote. It smelled of clickbait but it wasn't. Apologies to the authors.