Комментарии:
Thanks Simon! Do you think Photon will always be a Databricks proprietary feature or if there is any chance that they will decide to donate it to Spark OSS?
Ответитьvery well: clear, concise explanation, SIMD solved: thanks!
ОтветитьVery good, I think this is a good idea. I am special, I have great potential, this program will help me. All streamers are bad, but I'm almost brilliant, yes, I can use this, please accept a friend request.
ОтветитьNice Video! I have a question -- If photo is only exists in the thread, I suppose most of the Query Plan are remaining the same? Do we need to select physical plan base on cost models?
ОтветитьI love your videos. It's an inspiration.
And i love the plugin-ability of the databricks execution plans.
It's genious, to have an open interface for future plugins :-)
( Databricks in general consists of quite nice concepts )
And in this video i love the JNI/C++/column-vector-approach.
That's exactly how i would do it, when it comes to mass data processing.
I always had trouble with the JVM-garbage collector in this context.
And i always loved to have full control of memory in C++.
Thank you :-)
a small soft and i am sure i can help you !!
Ответитьi really apreciate your help with dowloanding this software
ОтветитьThe salaries for these c++ engineers plus the hours required to program at that level of performance probably only made sense now that the company has more revenue. Having used photos along with premium block blobs in azure it definitely paid off though and you get really fast queries in dashboards.
ОтветитьGreat talk as usual, Simon! Thanks
Ответитьwhat software do you use to record videos
Ответить