Cycles faster then Eevee ? what do i wrong

I’ve got some production material for a proto type, (cannot share at the moment),
I noticed the render time in cycles is just about 3.34 minutes and Eevee takes 5.17

I dont need cycles rendering but i do need fastest rendering.
Since i make a long animation, i thought Eevee can be faster then cycles, but how to configure Eevee for speed ?

My cycles settings are :
-Experimental CPU (got a fast CPU and older graphics cardGforce430M).
-80 samples path tracing (not squared samples).
-light treshold 0.58
-Light paths total 12

  • difuse 4, glossy 4, transp 6, transmisson 6, volume 0
  • clamping direct light 0, indirict light 10
  • filter glossy 1.0 , reflective caustics=yes refractive caustic=yes
    -Denoised (default settings).

There’s your answer, Eevee requires a decent GPU, in fact I’m surprised it runs as I thought the GPU had to be above a 600 sries.

But what would be fast settings, it does run so, maybe i use non ideal settings.

Hi. There is not much to configure regarding speed with Eevee if you are using a basic setting without using light probes or things like Volumetrics. There in Render tab you have Sampling settings in Eevee, you can try to reduce the value in Eevee for render if it is useful for you.
To do better tests you could perhaps delete your scene objects that you can not share, and just share a .blend file with settings and elements that you can share. Or maybe make tests with any of these scenes:

I experienced the following factors play a role (as well):

  • The more lights, the longer the render time.
  • More render time when using volumetrics
  • Resolution and amount of samples as mentioned
  • Complexity of shaders
  • Complexity of nodes in compositor
  • And I believe all you can switch on/off in the render tab: ambient occlussion, depth of field, motion blur, etc ?

It’s very simple: Your graphics card (GeForce 430 M) is too weak for EEVEE. Blender starts because the card supports OpenGL 4.1, but the GPU doesn’t have enough power to draw the viewport/render in acceptable time. For comparison your card has 96 CUDA cores, while a 1050 has 768. This is just one factor, the GPU, memory, bus etc are a lot slower compared to newer cards. So the only solution would be to get a newer, more powerful graphics card.

Nor should you ever share proto type stuff for production materials.
Fastest rendering? Did you consider Workbench engine? If it’s for inhouse use to show off its features, there is no point in making it look “nice”. Even if it still uses OpenGL, it should be way faster than Eevee. I have a beefy graphics card, but it still takes Eevee waaay too long to do animations, although my settings may be needlessly increased through the roof in attempts to fix issues. Whereas with Workbench I can do 2500 frame office space walkthrough animation in about half an hour.

Otherwise, same as the rest I guess; either get a proper graphics card or offload your rendering to a renderfarm service (they do Eevee?).