Rendering Speed on New Computer...

Hi Guys,

I have a pretty complex (by my standards) model that is the interior of a restaurant. I have a attached a text with the render stats (lots of people, models, NxT materials and lighting. I expect it to take some time.

Up to now I've been using a machine to render that was Intel i7 (2 Core Duo - 4 processor) @ 2.0 GHz computer. My render times for SM, MED are about 2 hours - 3 hours. I once ran a 1280 x 720 and it took 24 hours!

I just unboxed a new machine that I got specifically for high-end video, graphics, gaming, etc.

I've got the i7 5820K processor that runs 6 cores (12 processors) @ 3.30 in Overclock mode up to 4 GHZ.

When I unboxed it I installed only SK2015, IRENDER NxT ,and my Dropbox to get to my files.

My first render of a Med size I did of my model took 2.25 hours, same as before. My large 1280 x 720 is on pass 17 and 24 Hours 9 minutes.

Am I missing a "Go FAST" configuration in the tool that tells it to use the full power of this machine? I made sure the setting didn't have the USE ONLY in the CPU settings set to a smaller number.

In the RESOURCE MONITOR of Windows 8.1 I see where each of the 12 processors is carrying a load but never more that 45 - 50% of the capacity.

I appreciate any help you can provide.

Attachments:

haddon render stats.txt, 80 KB
haddon machine specs.png, 106 KB

▶ Reply to This

Replies to This Discussion

Permalink Reply by Render Plus on June 29, 2015 at 4:19pm

I had a similar problem today - which I could not figure out - the rendering was using just 25% of my machine.

I rebooted and that fixed it. I doubt if you will be so lucky.

You can set the maximum number of threads to more than you think the machine has. Even if you are using all cores, this will sometimes speed things up a small amount, but if for some reason you are not using all cores, this could make a difference.

If it does, let us know so we can find out why we weren't using all your cores properly.

▶ Reply

Permalink Reply by Render Plus on June 29, 2015 at 4:26pm

Also, to test this better, render a very small model, for a lot passes and turn off autosave.

The idea here is to make that autosaving, or reading large textures, or something else in a larger model is not slowing down the rendering. I rendered just "sophie", (in SketchUp 2014), and was able to get up to 100%

▶ Reply

Permalink Reply by Mike Haddon on June 30, 2015 at 7:10pm

Thanks for the advice Al. I stepped back the render passes to 10 from 20 and saw no loss of lighting, shadowing, or texture depth. I also set the processor to 32. It finished in 45 minutes!

▶ Reply

Permalink Reply by Render Plus on July 1, 2015 at 4:55am

If you did both things at the same time, can you run three tests for me? - 2 passes at 24 cores, 2 passes at 48 cores and 2 passes with cores set to # of cores and let me know how many cores display when you select '# of cores.

You can make these changes and re-render from the rendering window, so you won't have to extract the geometry to do each test.

Thanks, al

▶ Reply

Permalink Reply by Mike Haddon on July 1, 2015 at 10:24am

Sure Al. I will run those and let you know the results.

▶ Reply

Permalink Reply by brooks john on December 2, 2015 at 10:03pm

QUESTION: even if your processors is at 100% running on maximum cores. increasing the core to great then what you have will make things slightly faster? is this what your are saying also ?