Maxwell Render | Official public forum

Hi folks,

I wondered if there is a formula or approximation which one use to predict the render time for a larger resolution.

Yes, I know that the render time depends entirely on the scene as so the number of SL to go.

I usually render out my drafts in 720x405 @ SL12. Lets say I am satisfied with the result of one of my preview and want to render out in a higher resolution 1920x1080 or even higher at the same SL level. Is there any math one can do to predict the render time?

Regards.

I don't know the answer to that but keep in mind higher resolution renders don't require the high S.L. to look as clear as the lower resolution renders. It make sit difficult to predict.

Thanks for posting ... is there an explication for that higher resolution require less SL to give a clean result?

There was another thread in which was argumented by accounting for the print DPI or viewer's distance. But is this true for current screen resolutions of 2000+ pix?

BTW: the first larger resolution render (1920x1080) I am doing now does seem to take around 10x the preview render (720x405) time at the same shading level. Considering that it is done on the main computer and I continued working on it. Dispatching the render to a separate node should things speed up somewhat.

Regards.

Bubbaloo wrote:... but keep in mind higher resolution renders don't require the high S.L. to look as clear as the lower resolution renders ...

Most of the internal work that I do takes roughly SL 14-15 for acceptable noise. SL 16 is almost always a clincher . These are at 5000-6000 pixels maximum length meant for arch-viz prints. Depending on lighting conditions I will more or less know what is going to need more samples, keep testing and you will quickly get a good indication of what you will need in the future.

I still don't have my private "renderfarm" ... but the idea is to build up a four-node system where one node is also the main node where MW Studio will run. I would expect to get "overnight" rendertimes of 8h or less @ SL 16 or maybe some higher SL.
I'll wait some time before buying the hardware, to see if Apple does a decent mac-mini refresh. Three headless mac-mini render nodes with an i5/7 inside would make a very nice figure on top of my desk

Regards.

NathanDan wrote:Most of the internal work that I do takes roughly SL 14-15 for acceptable noise. SL 16 is almost always a clincher . These are at 5000-6000 pixels maximum length meant for arch-viz prints. Depending on lighting conditions I will more or less know what is going to need more samples, keep testing and you will quickly get a good indication of what you will need in the future.

I too am waiting for a mac mini refresh i'd love to see an i7 as the high end machine. Also rack mounts are starting to surface for mini's. My only concern is the heat issues of the mini especially with an i7.

Hi Chedda,

nice to see that I am not alone on this lonesome road ... there are a lot of rumors currently that they are already under the way. I've read interesting assumptions on this topic:

1) Apples announcement to drop the xserver line in favor to a mac-mini based server

2) shortages of mac-minis in some distribution channels may indicate that Apple drives stocks down before filling them up with a refreshed model again.

I mean to position the mac-mini as xserver replacement makes very much sense. Thinking in terms of datacenter dimensions, a current mac-mini is smaller than a server blade (of course blades have other advantages). On a standard 19" rack 1U drawer you could easily get placed 6 to 9 of them ... this would correspond 18 cores with todays processor configuration (or 36 cores for I7 CPUs). If you take conventional servers you get at most about 8-12 cores within the same space. The price should be roughly the same if not cheaper. Ideal for node based computing. And reliability, well ... if your conventional 12 core server strikes you have to get down all 12 cores at once to perform the maintenance, whereas from the 18 core mac-mini configuration, you would pick only the bad one replacing it instantly and let the other 16 cores work uninterruptedly.

I think temperature isn't the point too ... when I did run a 56h render test on my mac-mini at work, the case did heat up to lets say to around 40 deg C (handwarm). If they can put an I7 into a MacBook Pro which is still slimmer, they can put it into a mac-mini as well without suffering more impact on computing power and thermal aspects.

But even a I5 would be a nice upgrade.

Of course I wouldn't squeeze them into a rack drawer ... one "designer pile" of mini-macs as render farm on the desktop or any other prominent place in your working place certainly will make a nice figure

Regards.

chedda wrote:I too am waiting for a mac mini refresh i'd love to see an i7 as the high end machine. Also rack mounts are starting to surface for mini's. My only concern is the heat issues of the mini especially with an i7.

So this just happened Caustics, the quad has a good geekbench score i'm getting interested.

Yes, my friend ... Our wishes became reality ... Sadly it still will take a couple weeks until they arrive here in Brazil. Sadly too they arrive costing us twice the $

If you get your hands on one, please let me know how it does.

Regards.

chedda wrote:So this just happened Caustics, the quad has a good geekbench score i'm getting interested.

Okay hold your fire it seems the HD3000 onboard graphics does not support open CL. I'm sure with the release of lion maxwell will be upgraded to take advantage of grand dispatch and open CL. This along with heat issues make the mini less than ideal hmmmmmmm.