Difference between G2 = G .* G and G2 = G * G in Matlab GPU Computing

What's the difference between G2 = G .* G and G2 = G * G in the following codes? And why I get with first code GPU Load 100% and with the second I get GPU Load and Memory Controller Load sensors both at 100% on gpu-z ?

X = rand(5000, 'double');
G = gpuArray(X);
classUnderlying(G)          % Returns 'single'

for m = 1:5000
G2 = G .* G .* G .* G;         % Performed on GPU
end

whos G2                     % Result on GPU

X = rand(5000, 'double');
G = gpuArray(X);
classUnderlying(G)          % Returns 'single'

for m = 1:5000
G2 = G * G * G * G;         % Performed on GPU
end

whos G2                     % Result on GPU

Copyright Notice：Content Author:「Aurelius」，Reproduced under the CC 4.0 BY-SA copyright license with a link to the original source and this disclaimer.
Link to original article：https://stackoverflow.com/questions/21171006/difference-between-g2-g-g-and-g2-g-g-in-matlab-gpu-computing