I’ve compiled blender with cycles from the latest version of the source from the
SVN repo. I used GCC 4.4.
I have installed the latest graphics driver for my nVidia 9600GT, along with the
latest version of the CUDA SDK.
However, when I change the render device from CPU to GPU, the 3D view area goes
black and the console reports that shader model 1.3 or greater is needed, whereas
only version 1.1 is installed.
Do you have any suggestions as to how to fix this?
Its because of features and technical specs being used by cylces.
If you take a look here:
The most speed benefit you gain is by using a gpu with 2.x computing capability. Because the amount of memory that can be used or the size a texture can have is doubled or tripled, etc.
As you see for example:
Number of cores for integer and floating-point arithmetic functions operations: With 1.1 to 1.3 you can only use 8 cores, with 2.x you can use 32 or even 48 cores.
Mostly your cpu is faster, if you got a gpu with 1.1 or 1.2/1.3.
this is so silly, cycle was working just fine, extremely fast and fun to work with now I’m stuck with using 2.60 and lower might as well just use the old 2.58 like I was before and just piss off cycles completely :spin: bad move blender >:C
Thanks everyone, only just saw there were additional replies. Guess I’ll just have to wait until I can get a newer GPU.
By the way, I too found that - under Windows - and with the build from graphicall.org, I could select GPU mode successfully. Should I still stick with CPU mode, even where GPU mode works? (My cpu is a Core 2 Quad Q9550 4 cores at stock speed of 2.83GHz.)