arrays - Sorting coordinates of point cloud in accordance with X, Y or Z value -

A for example, a series of points in 3D (X, Y, Z) Coordinate is:

& gt; & Gt; A = [1 2 0; 3 4 7; 5 6 9; 9 5; 7 8 4]

A = 1 2 0 3 4 7 5 6 9 5 7 8 4

I want to sort the matrix in terms of values ​​"y" (second column) values.

Here is the code I am using:

& gt; & Gt; Tic; [~, Public] = sort (A (,, 2)); SortedA = A (loc, :) toc;

SortedA = 9 0 5 1 2 3 4 7 5 6 9 7 8 4

Passed time ** 0.001525 ** is seconds

However, this can be very slow for a large set of data. I appreciate it if someone knows a more efficient approach.

introductory discussion

This answer will mainly talk about this How to solve a problem, a computable skilled can tap into the GPU . The problem presented in the question was the solution code -

  [~, loc] = Sort (a (,, 2)); Sorted A = A (location, :); There are basically two parts -  
 
 Select the second column, sort them and get the sorted index.  
 In the rows of the input matrix with the index sorted index   
 Now,  part 1  is an intensive count, which can be ported to  GPU  
 
 the proposed solution 
 
 Therefore, considering all but,  part 2  can be done on a  CPU  when a sequencing is done. This will be an efficient  GPU  solution - 
 
  GA = GPRRA (A (,, 2)); % // jpu [~, glock] = only the second column of the input matrix for the sort (ga); GPU SortedA =% on calculation Sort index (glow), :); Get the% // sorted index back on the CPU with `assembled`% and then use them to sort  
 
 benchmarking 
 
 Next benchmark To compare the code against the original solution to the  GPU  version, please keep in mind that since we were originally referred to as solution, the  GPU  code on different hardware Running  CPU , Benchmark results vary from system to system You can. 
 
 Here is the benchmark code - 
 
  N = 3000000; % // datasize (number of rows in input) A = rand (n, 3); % // Random Large input Disp ('------------------' with basic solution on CPU) Tic [~, folk] = sort (A (,, 2) ); Sorted A = A (location, :); TCC, Clear Sorted A Public Disp ('------------------) with the proposed solution on GPU) TIC GA = GPU (A (,, 2)); [~, Glock] = sort (ga); Sorted A = A (Collect (Glock), :); Toc  
 
 Here are the benchmark results - 
 
  ------------------ CPU time on the basic solution It has been 0.795616 seconds. ------------------ With the proposed solution of GPU, the time has passed since 0.465643 seconds.  
 
 So if you have enough  GPU , then there is more time to exclude  GPU  to sort out the related problem MattelBag offers such easy  GPU  porting solutions. 
 
 
 System Configuration 
 
  MATLAB Version: 8.3.0.532 (R2014a) Operating System: Windows 7 RAM: 3GB CPU model: Intel® Pentium Processor E 5400 (2 M cache, 2.70 GHz) GPU model: GTX 750Ti 2GB