Corona Progressive Rendering and CPU Cores

[Kristoffer Andersen]

Hi, I apologize if this has already been adressed but I was wondering about how corona utilizes cores in progressive rendering vs bucket rendering? As far as I've understood 1 cpu core equals one bucket in bucket rendering; is it still somewhat like this with progressive rendering? How does it scale? And perhaps there's no right answer to this, but is a strong mhz CPU with few cores weaker than a weaker cpu that has many threads? I'm still using a 2600k.

OT: Tried searching the forum for answers on this topic and found none. I did however find the way the searchword turned bright white in the search results quite annoying. :P

[Ondra]

progressive rendering handles that automatically, there is no need for you to worry about anything here. Regarding the CPU performance: I recommend comparing benchmarks. Generally, it is better to have half cores with 2* higher frequency, but that is usually not possible to get. And when thinking abour real examples, 6 cores on 3.2GHZ will be much faster than 4 cores on 3.6.

[pokoy]

I tried to figure that out too, but that's not easy to answer.
With the earlier Alphas my new PC (which is approx. twice as fast as my old machine) was faster in Corona by a factor of 1.5.
With the recent daily builds however this has changed considerably and the factor is around 2.5, which means that my old PCs now render slower than they're supposed to.
That's quite weird since both numbers don't reflect the actual speed ratio which would be 2:1 in a perfect world. I guess it's best left to Ondra to answer why this happens.

[Juraj]

I tried to figure that out too, but that's not easy to answer.
With the earlier Alphas my new PC (which is approx. twice as fast as my old machine) was faster in Corona by a factor of 1.5.
With the recent daily builds however this has changed considerably and the factor is around 2.5, which means that my old PCs now render slower than they're supposed to.
That's quite weird since both numbers don't reflect the actual speed ratio which would be 2:1 in a perfect world. I guess it's best left to Ondra to answer why this happens.

Is your testing methodology consisent ?

Is the ratio consistent among synthetic benchmarks like Cinebench 11.5 ? It should reflect that quite well.

I am not sure what PCs Keymaster is testing on but perhaps he should include broader spectrum than just regular i7.