Add another surface mode to grdview based on the Gouraud gradient.

[joa-quim]

The recent Gouraud gradient is much more efficient than the surface method in grdview (-Qs). See these numbers:

gmt grdmath -R-5/5/-5/5 -I0.5 X Y HYPOT NEG 3 POW 10 MUL = peak.nc

time gmt grdview peak.nc -R-5/5/-5/5/-3536/0 -JX8c -JZ4c -Qg -Ct.cpt -p135/30 -Bafg -Bzafg -BWSneZ+t"Gouraud" -P > lixo1.ps

real    0m0.297s
user    0m0.026s
sys     0m0.010s

time gmt grdview peak.nc -R-5/5/-5/5/-3536/0 -JX8c -JZ4c -Qs -Ct.cpt -p135/30 -Bafg -Bzafg -BWSneZ+t"Gouraud" -P > lixo2.ps

real    0m4.527s
user    0m0.522s
sys     0m0.085s

lixo1.ps 184,005    KB
lixo2.ps 7,722,824  KB

But the result isn't exactly equal either (though most of times is visually indistinguishable). With vertical discontinuities they are different.

gmt grdmath -R0/10/0/10 -I0.2 X 5 GT 10 MUL = cliff.nc
makecpt -T0/11 > tt.cpt
gmt grdview cliff.nc -R0/10/0/10/0/10 -JX8c -JZ4c -Qs -Ctt.cpt -p260/30 -Bafg -Bzafg -BWSneZ+t"Surface" -png surf
gmt grdview cliff.nc -R0/10/0/10/0/10 -JX8c -JZ4c -Qg -Ctt.cpt -p260/30 -Bafg -Bzafg -BWSneZ+t"Gouraud" -png gou

[Esteban82]

I tried it and it works fine on Linux.
In my case is bearly better

real    0m0.636s
user    0m0.530s
sys     0m0.083s

real    0m0.819s
user    0m0.700s
sys     0m0.098s

[joa-quim]

Strange. What I posted was from WSL but on pure windows, even if only visual, the time difference was striking.