i have a new computer (mac pro) which i am running in 64-bit vista using bootcamp. It has 8-cores that also support hyper-threading. I am definitely not an expert when it comes to these things so please correct me if im wrong, but in many other applications it shows that i have 16 cores all of which are running at 2.4Ghz. Can vray use hyperthreading as this would obviously reduce render times hugely in complex renders.
If you have one core without HT running 100%, or one core with HT where each thread runs 100% - surely that is the same speed then?
Or are you seeing that of the 16 threads on your 8 HT cores, only 8 are working?
And I’m a but unsure about how HT cores work, if you have an HT core and one of the threads use 100%, does that mean the core works only 50%? Or might it not be so simple?
this is what i am unsure about. Although when i render a scene it says “setting up 8 threads”. Should this say 16 or will it make no difference at all?
I am pretty sure that the buckets work off of how many cores there are, regardless of hyper-threading. Most computers these days have either hyper-threading or hyper-transport, and I don’t think I’ve ever seen a dual core machine pop 4 buckets up during a render.
I had a look for the passes setting, but could only find “Num Phases” in light cache? not sure if this is what you meant, but changing it had no effect.
i am unsure if this is normal, however in the system properties menu above (and in speccy) it does not show at all that there is even a second core (maybe this is just because it only gives info for the first?)
When looking in device manger though, it does show 16:
I’m on a macbook, and I get one bucket per core. I’ve never seen hyperthreading yield more buckets, but then again, maybe we’ve been doing it wrong all along…
oh, another good point, if your cpu is being used 100%, who cares how many buckets there are. The cpu is being fully utilized during the render anyway.
yeah i guess this makes sense. What is the difference whether i have one core working on one render or one working on two, it will still take the same amount of time.
I am guessing that hyper-threading works well where you have two relatively small tasks that can be completed by one core, eg one that takes up 25% and another that uses 60%, but does not mean that the core can effectively “duplicate” its work at 100%. It is probably more efficient to have it dedicated to one task.