Cuda Mannager "BUG"/ more advanced settings at User Preferences Tab #46334

New Issue

diesistkeinemailadresse12345 · 2015-09-30T20:50:11+02:00

2015-09-30 20:50:11 +02:00

System Information
i7 2600k
GTX 770
GTX 770
GTX 780 Ti

Blender Version: 2.76rcShort description of error**
Hello together,
i am working with blender now since more than one and a half Jear.
now i had to buy a new PC.
So i now have a PC with 3 Gaphic Cards.
And here is my Problem:
Because of my Mainboard 2 of my Graphic Cards are only connectet at PCI-E x1 1.1 Speed, the other one is x16 2.0.
I use PCI-E Riser Cables of course.
For Final Render at Cycles this Setup works realy fine and fast.
But because of the x1 1.1 Speed at 2 of the cards, in 3D Viewport, "RENDERD" mode a single card is (at x16) is very much faster than all the three together.
This is not really a bug inside Blender, but i wish it coud be improofed, because it really slows me down at work.

So for "efficent" working i have to go to User Preferences - System Tab every time (when i switch from final render to Edit or back) and change CUDA from "GTX 770 (x2) + GTX 780 Ti" to "GTX 780 Ti" or back.
this is really annoying and costs mutch of time if you have to work fast.

So is there any possibility to add an Function like the following in future Versions of Blender?:

Instead of only Selecting CUDA Device i like to have an Button there for "Advanced" settings (maybe like the one at Auto Tile Size) witch adds more Options, for selecting your Compute device seperatly for final render and "live render" mode.
I also would very like to be able to enable or disable CUDA Devices a bit more independently with for example something like checkboxes for evey Device:

GTX 770
GTX 770
GTX 780 Ti

Maybe like this.

Best Greetings as far and i hope you could understand everything, because of my english.

**System Information** i7 2600k GTX 770 GTX 770 GTX 780 Ti **Blender Version: 2.76rc**Short description of error** Hello together, i am working with blender now since more than one and a half Jear. now i had to buy a new PC. So i now have a PC with 3 Gaphic Cards. And here is my Problem: Because of my Mainboard 2 of my Graphic Cards are only connectet at PCI-E x1 1.1 Speed, the other one is x16 2.0. I use PCI-E Riser Cables of course. For Final Render at Cycles this Setup works realy fine and fast. But because of the x1 1.1 Speed at 2 of the cards, in 3D Viewport, "RENDERD" mode a single card is (at x16) is very much faster than all the three together. This is not really a bug inside Blender, but i wish it coud be improofed, because it really slows me down at work. So for "efficent" working i have to go to User Preferences - System Tab every time (when i switch from final render to Edit or back) and change CUDA from "GTX 770 (x2) + GTX 780 Ti" to "GTX 780 Ti" or back. this is really annoying and costs mutch of time if you have to work fast. So is there any possibility to add an Function like the following in future Versions of Blender?: Instead of only Selecting CUDA Device i like to have an Button there for "Advanced" settings (maybe like the one at Auto Tile Size) witch adds more Options, for selecting your Compute device seperatly for final render and "live render" mode. I also would very like to be able to enable or disable CUDA Devices a bit more independently with for example something like checkboxes for evey Device: - [x] GTX 770 - [x] GTX 770 - [x] GTX 780 Ti Maybe like this. Best Greetings as far and i hope you could understand everything, because of my english.

diesistkeinemailadresse12345 commented

2015-09-30 20:50:11 +02:00

Changed status to: 'Open'