1
$\begingroup$

This is my very first post, so please forgive and correct me if I do something incorrectly. I recently purchased the new (as of May, 2018) Dell XPS 15 2-in-1 laptop which has the new Intel Core i7 8705G that has a dedicated/integrated AMD Vega M GL GPU. It is supposed to (according to benchmarks) perform in the ballpark between the GTX 1050 TI and the RX 470. When I select this GPU in the system tab of the user preferences, under the OpenCL section, I get nearly the same results (in terms of render times) when I select GPU or CPU from the render menu in the properties menu. Is there anything I am doing wrong? Also, I understand, as I am coming from a PC with a GTX 1070 that I have to adjust accordingly the Tile size in the Performance section of the render menu. What is curious to me, is that when I have it on CPU, I have to use tile sizes around 256 to 512 in order to get the best times! Which blows my mind because usually it would be 16 to 32 on a 7700K. The GPU using the same tile sizes as the CPU.

Thank you in advance to any help that I may receive.

P.S. The file that I have been using to run my tests is the AMD benchmark blend file found at http://download.amd.com/demo/RyzenGraphic_27.blend

Improve this question
$\endgroup$
4
  • 2
    $\begingroup$ just to be clear the GPU rendering does work. its just slower then you expected. $\endgroup$
    – David
    May 25 '18 at 0:50
  • $\begingroup$ Correct, when I select the GPU compute device, it does render. But much slower then expected. And when the CPU is selected, only large tile sizes lower the render times. $\endgroup$
    – Jose Quesada
    May 25 '18 at 3:22
  • $\begingroup$ I guess, because it is an integrated GPU, that it does not make a difference wether you select GPU or CPU rendering at all, except that you may find some options missing like Denoise when you don’t have GPU selected. What’s curious to me is what you say about the tile size. What happens when you enter the same tile size? Cheers, Daniel $\endgroup$
    – Daniel Hinnerkopf
    Jan 10 at 8:36
  • $\begingroup$ Can you see that the GPU is being actively (and heavily) used when you select it for rendering, @JoseQuesada? If you're using Windows 10, for example, under Task Manager, there is a simple GPU load measurement displayed (there are a number of other ways to do this, too, but Task Mgr is probably the simplest), and this should increase substantially when rendering with the GPU selected (depending on scene complexity, of course). This should let you at least determine whether the GPU is in fact significantly in use, and, if so, then you can focus on analyzing performance (or lack thereof) $\endgroup$
    – NeverConvex
    Oct 24 at 0:28
-1
$\begingroup$

Go to User Pref/System under Screencast switch none to OpenCL and choose you GC

Improve this answer
$\endgroup$
1
  • 3
    $\begingroup$ They already did that according to what is written in the question and summarized in comments above $\endgroup$
    – Mr Zak
    Aug 7 '18 at 8:58

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.