Jump to content

GTX770 extremely slow on 10.9


pcace
 Share

5 posts in this topic

Recommended Posts

Hi,

 

i just got a GTX 770 4GB in addition to my GTX 480 1,5GB. Both Cards run out of the Box when not using GraphicsEnabler in boot.plist. 

First i was really happy, but then i started to really use both cards (mainly Thea Render, which has a unbiased GPU/CPU engine) and i realized, that my 770 has around half the Power compared to the power it has on Win7. 

 

The results on Luxmark Win7 64 Bit are:

 

Both Cards: 2554 Points,

GTX770 1351

GTX480 1171

 

Which to me looks pretty ok, 

 

On OSX 10.9 these are the Results:

Both Cards: 1470 Points,

GTX770 689

GTX480 766

 

 

Wich to me looks looks really bad - especially the GTX770. 

Where can i look for the Problem? Is this a known Problem? 

Is there any way to change this?!

 

I am using chimera 3.0.1, and i installed OSX via [url="http://www.insanelymac.com/forum/topic/279450-why-insanelymac-does-not-support-tonymacx86/"]#####[/url], 6.3.0

 

Any idea how to fix that Problem would be awesome!!!!

 

Thanks a lot,

 

Pcace

  • Like 1
Link to comment
Share on other sites

  • 4 weeks later...

Hello all,

 

I'm finding the same thing and am hoping there's a solution. I get double the LuxMark score when I boot into Windows 7 and Unreal Editor is over 5 times faster than it is in 10.9.3.

 

I'm using the latest nVidia web driver and CUDA update and a GTX 770.

 

Any help would be much appreciated.

 

Thanks.

Link to comment
Share on other sites

  • 1 month later...

I also wondering about the same thing

I have GTX770 and it also perform better in windows

 

UPDATE :

Most likely it's the opencl driver/library on OSX is not optimized using CUDA

post-1128849-0-42877000-1408863288_thumb.png

 

GTX 770 have 8 compute unit and each compute unit have 192 CUDA core (total of 1536)

I think opencl on OSX only able to access the compute unit to do raytracing

While on windows the opencl is work using nvidia CUDA

See at the luxmark on OSX only say OpenCL 1.2

while on windows OpenCL 1.1 CUDA

Maybe that is why luxmark is slower on OSX

Link to comment
Share on other sites

  • 3 weeks later...
 Share

×
×
  • Create New...