Thought someone might be interested
in the
following information.
I set up a Sun Ultra Sparc
workstation as
follows
HARDWARE:
Dual sparc 168mhz processors with
sparc
floating point processors
128Mb of memory
12Gb of Disk space
OPERATING SYSTEM:
Sun Solaris
v2.6
Common Desktop
Environment
v1.0 (Now available with Solaris v2.6)
COMPILATION COMPONENTS: (All built
specifically for
sparc-sun-solaris2.6)
gcc
v2.8.1
binutils
v2.8.1
gdb
v4.16
make
v3.76.1
libstdc++
v2.8.1
POVRAY SOURCE v3.02
(also compiled
compression tools that came with POV-Ray Source)
I built the x-povray version of the product with
little
trouble. I built a Sun Package (using pkgmk) to install the POV
product in
/opt/local/ which is the preferred location for 3rd party products for
Solaris. This makes the pkgadd and pkgrm of POV-Ray from a Sun box
effortless.
My testing of the POV engine has had zero problems
and the POV
x-povray works well with the CDE (Common Desktop Environment).
I've tested
the program with several sustained, simultaneous animation
renders. Here's
some numbers I've come up with.
A single 62 frame animation render outperformed an
Intel
(PentPro 180mhz) by 33% (Sparc was 33% faster). The Ultra-Sparc
render
reported more sampling, but I could not tell any significant difference
in the
resulting avi file.
Muliple, simultaneous animation renders (3 scenes
ranging from
48 to 62 frames) buried the Intel, but the Ultra-Sparc showed no signs
of
slowing. (I may try adding additional simultaneous renders to see
when the
Ultra slows when I get time)
Obviously, anyone would expect a
Dual Sparc to
outperform Intel. I report the results as a "Frame of
Reference". Does anyone have a suggestion where I could look
in the
POV source to get even more performance out of the Solaris v2.6
build? I
mean something simple, like a header setting or compile
option.
McNaul