[chaoscope] Re: Development update
- From: "Marjo & Charles Lips" <familylips@xxxxxxxxxx>
- To: <chaoscope@xxxxxxxxxxxxx>
- Date: Tue, 8 May 2007 06:14:19 +1200
Hey, merci beaucoup for the compliment Nicolas! And thanks again for all your
work, which I did not understand at all! I am really itching to try out the new
version, can't wait to use the unlimited iterations on some attractors I found
that simply will not develop fully at max iterations; and all the other
wonderful stuff that you are putting in, all for our free use and enjoyment.
----- Original Message -----
From: Chaoscope
To: chaoscope@xxxxxxxxxxxxx
Sent: Sunday, May 06, 2007 9:11 AM
Subject: [chaoscope] Development update
Hi everyone,
first of all congratulations to Charles (or Marjo?) for the picture on
deviantart. I really like the way you seemingly stitched both attractors
together. The Moon rays in the water are also a very nice touch.
Now then, for a very much overdue development update:
I built myself a new machine, it's a 2.4Ghz Core Duo with 2Gb of RAM.
One of the first things I did once it was up and running was to update
the rendering engine and make it dual-threaded: One thread computes the
attractor equation while the other plots the result. We only get 100%
CPU usage when both tasks take roughly the same amount of cycle to
complete which is rarely the case. I haven't precisely measured the
average gain but I estimate it at 30~40%. Of course Chaoscope detects
the number of available cores or CPUs and uses the standard single
threaded render on single core CPUs. Batch Processing will keep on using
the single threaded render but will automatically launch as many renders
at the same time as there are cores/CPUs.
I've also progressed on the perspective front, it is now working for the
Solid mode which I thought would be much more of a headache. It's not
yet perfect and need more work. It's one of these features no user ever
requested whilst taking a huge amount of time to develop!
I can't remember if I've mentioned the fact I've implemented caching?
It's working but interestingly some equations won't benefit from it.
That's how costly a memory write can be on modern PCs, a simple equation
will run faster than writing twelve bytes in RAM. I expect this to
depend on the machine architecture so it will have to be profiled on a
ad hoc basis whereby caching will be enabled only if it increases render
speed. Of course caching is a very big improvement for animation of
fixed attractors (if you change a parameter you simply lose the cache)
and the slower the equation the bigger the improvement. I'm sure you'll
enjoy it Tom. ;-) I've also found out that sorting the points spatially
increases render speed further although this is only beneficial to long
animations of fixed attractors.
Kind regards,
Nicolas Desprez
======================================================
The Chaoscope mailing-list
Archives : //www.freelists.org/archives/chaoscope
Admin contact : chaoscope@xxxxxxxxxxxxxx
Web site : http://www.chaoscope.org
======================================================
Other related posts:
