H. Karsten nous apporta ses lumieres en ce 2008/01/02 07:38:
> Hi, we use PovRay for a film-documentary (underwater scenes).
> 
> The implementation of the hyper-treading in povray would speed up our rendering
> about 5GHz.
> We using this machine with 8 cores, 16 with hyper-treading:
> http://supermicro.com/products/system/4U/8044/SYS-8044T-8R.cfm
> Are there any planes to build in hyper-treading into the render process in
> PovRay?
> 
> Ps.: My job is to visualise the diving from whales, using so called "d-tag"
> data. (These are real data recorded from real wales)
> To make a complete automated process from reading the data to rendering the
> image, I've created a way to use PovRay with "bones". Means that PovRay it self
> edit the geometric data (vertices and normals) of the whale while rendering. If
> there is any interesting in it, I would like to share the trick, and the
> program I've wrote for.
> 
> 
> 
> 
> 
> 
> 
You seem to confuse multi-treading (real processors/cores) and hyper-treading 
(virtual processors on a single core). It look like you have 8 processors/cores 
(solod speed gain) each split into 2 virtual processors (marginal speed gain).
Release POV-Ray version (3.6x) use only 1 tread while rendering.
The 3.7 beta version will use any available processors/cores available, unless 
instructed otherwise.
In your case, it could mean an about eightfold acceleration. Multi-treading will 
give close to 8x acceleration and hyper-treading another 1.1 to 1.2 gain (maybe).

-- 
Alain
-------------------------------------------------
They that can give up essential liberty to obtain a little temporary safety 
deserve neither liberty nor safety.
Benjamin Franklin

Post a reply to this message