Is there number of threads increase slicing speed? Actually i was try today to figure out optimal number of threads in Slic3r and make some analysis. Results are:
Difference in slicing speed between 1 & 16 threads about 4,5%
Max processor load about 20%
So i make a decision that Slic3r can’t use full power of my hardware…
Slic3r only seems to use many cores at the first couple of stages, which take up 10% of the slicing time at most. The rest of the slicing is hard-limited to one or two threads.
@Thomas_Sanladerer Seems i am not alone with this problem))
What OS are you using? did you try increasing the priority? I think the windows multi threading scheduler is the part of the reason that even MS Office is faster on Linux (with wine)
@Joseph_VanPelt I using Win7 64bit, and i am pretty sure it is not a windows problem. The Processor is almost idle when Slic3r working.
I’ll do some testing on the same hardware (I think I still have a Windows disc here somewhere) and let you know the results
@Joseph_VanPelt Thanks! i will wait for results!
Some points:
- if your cpu has 4 cores, probably you will not notice any difference from 4 threads up.
- Not all operations can be parallelized in multiple threads
- Not all operations are performed by cpu. For example during disk reading cpu is idling. If you read a lot of data from disk you can have 1trillion cpu cores and threads but you didn’t notice any improvement (just an example)
@Andrea_Fontana you are right, but STL file usualy takes to 20Mb, so i don’t think HDD has a plenty of work. Also i have 16GB of Ram and slic3r in hardest cases took about 2Gb.
As said, HDD was just an example 
The program itself can be difficult to parallelize if wasn’t written with multithread in mind. Maybe it is not (yet) full multithreaded.
Cura is faster.
I’m running full speed tests right now, comparing Slic3r, Cura, Makerware and Skeinforge. I might be biased, but the results so far look like Cura is always 7x faster then Slic3r, and sometimes 12x faster.
I was tried Cura some time ago, but in my opinion - Cura makes less quality gcode. The model printed from Cura - generated G-code looks worse than Slic3r one.
It’s possible that you simply haven’t tuned your machine properly then. Because Cura is the default software for the UltiMaker and you’ll see nobody ever saying that slic3r is better on that machine. Cura output is amazing except in a few areas, one of those being bridging (and that opinion might even be out of date by now)
I’ve not used Slic3r personally but I started out using Skeinforge and it was pathetically slow. I switched to KISSlicer which works on both Windows and Linux and its sliced everything I’ve asked in seconds.
isn’t slic3r written in mostly Perl? How about the parts that take the most time, are they also Perl? What profiling tools are there for Perl - you could find out which processing steps or algorithms actually use most of the cpu time.
As promised, I ran a quick study between environments using 10 threads. On my setup there wasn’t a huge difference, but here is the info: (jvptec dot com slash 3dprinting.html)
http://jvptec.com/3dprinting.html