Stream: brlcad

Topic: April 2, 1984


view this post on Zulip starseeker (Apr 03 2024 at 03:40):

Was looking at the early BRL-CAD history, and happened to notice - April 2, 1984 was the date of the very first commits in our repo history from Mike Muuss adding C code for raytracing. (Some of the files are older, but the commit dates are 40 years ago today.) So even using the most conservative starting point for counting, that makes it happy 40th birthday BRL-CAD!

view this post on Zulip Erik (Apr 21 2024 at 16:35):

I believe the code goes back to 1979, but was unmanaged (cp and tar instead of a vcs) until 1984... I think at that point, it would have been running on a vax 11/780 running maybe 4bsd (but before 43bsd, which would be a lot more recognizable to us). It's possible to build a very similar environment using simh and tracking down ancient archives of BSD, but it took a fair bit to get the performance tuning of the subsystems right to match the performance DB and report 1 'vgr' (also, sphflake was benchmarked on an o2 or o200, so it doesn't report in real vgr's)
Man, '84... my computer back then was a coleco adam running cp/m on a z80 processor with a whopping 80 kb or ram... it had dual 4 track (like, audio tape) cassette drives, but only one had a read/write head. 40x25 text on a crt :D and I was keying in basic programs from a book called "games apples play", trying to port to cp/m basic as I went, different days :grinning:

view this post on Zulip Sean (Apr 22 2024 at 05:49):

sphflake was actually baselined on a sun system if memory serves. I don't remember all the reasons why I did that, but have vague recollections of the benchmark numbers being much more stable and consistent there, and the threading model on irix was uncommon (sprocs). I believe it was also one of the oldest and slower systems I could get my hands on a the time.

view this post on Zulip Sean (Apr 22 2024 at 05:52):

getting vgr back up and running in simh would be amazing. I was even able to find copies of jhu unix (an at&t derivative with basis in bsd iirc) that were allegedly running on vgr, but it'd undoubtedly be easier to get something like modern netbsd up and running and clock-throttled until we got to vgr=1.0

view this post on Zulip starseeker (Apr 22 2024 at 16:44):

Given the modernity of the compiler we'd need for a recent build, I'd think modern *bsd of some sort would definitely be the way to go.


Last updated: Oct 09 2024 at 00:44 UTC