US 12,450,043 B2
Efficiently vectorized implementation of operations in a global grid indexing library
Desmond Cheong Zhi Xi, San Francisco, CA (US); and Menelaos Karavelas, San Ramon, CA (US)
Assigned to Databricks, Inc., San Francisco, CA (US)
Filed by Databricks, Inc., San Francisco, CA (US)
Filed on Nov. 3, 2023, as Appl. No. 18/501,839.
Claims priority of provisional application 63/426,341, filed on Nov. 17, 2022.
Claims priority of application No. 20220100900 (GR), filed on Nov. 4, 2022.
Prior Publication US 2024/0152338 A1, May 9, 2024
Int. Cl. G06F 8/41 (2018.01)
CPC G06F 8/452 (2013.01) 20 Claims
OG exemplary drawing
 
1. A method, comprising:
generating code for iteratively applying a geospatial function to geospatial data, wherein the geospatial function includes at least a validation portion and an evaluation portion, wherein the code includes at least a first iterative loop of performing the validation portion over the data and a second iterative loop of performing the evaluation portion over the data;
compiling the code to generate compiled code, wherein the compiled code vectorizes at least the second iterative loop;
receiving, from a client device, a request to perform one or more data processing operations on a data table, the data table including at least a plurality of instances of geospatial cell indices, wherein the one or more data processing operations includes applying the geospatial function to the plurality of instances;
compiling, by a driver node, the request into one or more tasks for execution on a set of worker nodes, wherein the set of tasks includes at least a vectorized operation based on the compiled code; and
executing the one or more tasks by at least invoking the vectorized operation on the set of worker nodes.