TurboCAD Forums

The Ultimate Resource for TurboCAD Knowledge

Register
 
Be sure your post is relevant to the current discussion.  Create a new topic within the appropriate board if you are unsure.

Tutorial - rendering of RedSDK
Read 12891 times
* October 19, 2017, 05:28:29 PM
#50
I tried my old Workstation (9 year old Dell T7500 w/Nvidia Quadro FX 4800) and it took forever. The Quality Render RedSDK GIF was 937s and the Advanced Render RedSDK Majo_HQ — 2146s. The big different, to me, is the old machine uses 4 cores/threads, while the new one uses 48 threads. I'm waiting for the "pre-processing" (in LightWorks) to go to multi-threading, not just the painting of the screen. Currently, pre-processing still uses “1” core, which really sucks and the painting of the screen uses all 48.

Logged
John R.

V17—V21, 2015, 2016, 2017
Designer, Deluxe, Expert, Basic, Platinum
RedSDK enabled
Windows 10, 64-bit


October 20, 2017, 03:32:34 AM
#51
I agree with John! Some operations as "pre-processing render" and loading/saving drawing, smesh smooth level and etc. still uses “1” core/thread of the processor. This could be to improve in the next versions TC for better work, a better experiences with more complex/bigger drawings. Now for the render is more important how "strong" is the processor! More cores/threads is better than a high level graphic, but it is only my opinion  :D

Maybe someone from development team can explain better, What is needed for faster render?

Logged
Intel Core i7-3770 CPU 3,40GHz, 16GB RAM, NVIDIA GeForce GTX 760, 120+60GB SSD, Win 7 HP 64bit

TurboCAD Pro Platinum x64
www.invent-dizajn.sk, www.facebook.com/marian.jurcak


October 20, 2017, 04:19:00 AM
#52
I asked for a preferred system configuration a few years ago, and a couple times since I became a Beta Tester.

I have yet to get a response.


Jeff

Logged
TC Pro Platinum 2017, 2016 & 2015 (all with LightWorks) & V21
System: i7-5820K @ 3.30GHz, ASRock X99 Extreme4, 16GB DDR4-2133 RAM, Gigabyte GTX 970, Samsung NVMe SSD 950 (256GB), Windows 7 Pro (64-bit) SP1


* November 04, 2017, 02:36:47 PM
#53
I'm still not convinced that the GPU is the biggest boost, at least not with this drawing. I believe that CPU core count is more important. I compared my render times of my old 660SC GPU to my new 1060SSC with 2015 P-P:

    2015  Adv Render GI_Fine_HDR   660SC  67 sec,  1060SSC 55 secs.

Now, with the 1060SSC I compared 2015 to 2017
    2015 Quality GI_Fine  57 sec    2017 Quality GI_Fine 101 sec.

I also ran the new files on 2017 with the 1060SSC
    Adv Rend Majo HQ  Full screen 1920x1080   198 sec
                             Save as PNG @ 3840x2160  1191 sec
                             Save as PNG @ 800x800  151 sec

Between Jeff and I, I have the better GPU, my 1060SSC vs his 970, but he has the better CPU, my i7-3770 (8 Threads) to his i7-5820K (12 threads). John is getting remarkable times due to his 48 threads.
The only real way to compare GPU's is to have a setup and swap out only GPU's and compare differences between gaming cards and workstation cards.


EDIT::  I should note that during the rendering process, GPU usage was at a steady 35%, CPU was at 100%.  I went into advanced render options and enable 'Use GPU', but that resulted in a rendered image that was only black and white with none of the textures applied.
« Last Edit: November 05, 2017, 07:50:49 AM by Jason B »

Logged
TC 21 Deluxe, TC 20 Platinum, TC 2015 Platinum, TC 2016 Platinum, TC 2017 Platinum
i7- 3770 Ivy Bridge 3.4 GHz, 16G Crucial Ballistic, ASRock Extreme 4, EVGA 1060 SSC, Windows 10 Pro 64 Bit