Faster Radiance on Windows

Latest release include has some important updates for Windows users:

  1. Experimental multi-processing rcontrib: now rcontrib can be run with -n > 1. It’s a workaround for the lack of fork() on Windows. Do expect larger memory footprint. Stability on real world models untested.
  2. In general much better Windows ray tracing performance, should be on par with MacOS and Linux.
2 Likes

Great news! I will be testing it over the next few days. Thanks!!

This is super exciting @Taoning_Wang1! I’m trying to run rfluxmtx with n > 1 on Windows now, but it doesn’t seem to be working.

vwrays -vtv -vp -7.968 2.611 7.978 -vd 1 0 0 -vu 0 0 1 -vv 90 -vh 120 -x 600 -y 300 -pj 0.7 -c 9 -ff | rfluxmtx -n 8 -v -ffc -x 520 -y 300 -o glazing_1/g1%03d.hdr -ab 2 -ad 200 -lw 1e-2 -c 9 - glazing_1.rad -i scene.oct
rfluxmtx: running: rcontrib -fo+ -n 8 -o glazing_1/g1%03d.hdr -ab 2 -ad 200 -lw 1e-2 -x 520 -y 300 -ffc -c 9 -f klems_full.cal -p RHS=+1 -bn Nkbins -b kbin(1.22465e-16,1,-9.05942e-14,0,0,1) -m m000005 "!oconv -f -i scene.oct glazing_1.rad"

Only one thread is utilized. My Radiance version is RADIANCE 6.1a 2026-01-30 LBNL (6.1.74b430c613).

Sorry about that. Looks like I forgot to update one of the header files.

1 Like