Using integrated graphics to improve music streaming.

JasonM

JasonM

JasonM

Soldato
Joined
19 Jun 2009
Posts
3,983
This is a post about about my experience of improving music streaming on my computer.

I'm a software developer and when i'm working I always listening to music, my case it's Amazon Music HD however this I expect would apply to other streaming services such as Spotify.

My computer is an i7 8700 (non K) with 64GB ram, running 2 Quadro cards driving triple monitors, however this would apply to any cards such as Geforce or Radeon cards.

I've been listening to Amazon Music HD for about 6 months, however occasionally the streaming would stop for maybe 1/3 of a second, I contacted Amazon Music technical support but they could not resolve. I turned the Amazon Music hardware acceleration off, and this solved maybe 90% of the issues, however some stalling of playback still occurred. It was pretty obvious it was something related to decoding.

Like most people here, for many years I've been disabling the on-board graphics and just using the installed graphics cards. The issue however is graphics cards are doing many different tasks, they have to render my multi-monitors, also decoding and compute, then share tasks with my Amazon HD decoding.

I got thinking, there is a perfectly good integrated GPU that's currently disabled, could I use the IGPU for sole decoding of Amazon Music and the Quadro cards remain doing there regular tasks without interruption. Well it turns out in Windows 10 you can do exactly this, being you can set programs to use certain GPU's.

So I enabled the IGPU, and set Amazon Music as the only application to use it. And the result is Amazon Music HD has had perfect playback ever since, and all my other applications are still running on the Quadro's as before.

So if your steaming music, you may want to re-enable your IGPU and use for sole decoding of the music, you might find the streaming quality improves especially if your CPU and add in GPU's are under occasional load with other work.
 
Last edited:
Sounds like a motherboard/driver issue to me. But as long as it's working well that's all that matters.
Never read about GPUs having onboard dacs or decoding myself, but subbing to enlighten myself if such a thing exists/happens and i'm wrong ofc
 
The_Arbiter said:
Sounds like a motherboard/driver issue to me. But as long as it's working well that's all that matters.
Never read about GPUs having onboard dacs or decoding myself, but subbing to enlighten myself if such a thing exists/happens and i'm wrong ofc
Click to expand...

All modern GPU's have decode and compute ability. When I stream Amazon Music HD I can actually see the integrated GPU's decode usage increase, it then rests to 0% when I stop playback. GPU usage can be monitored in task manager.
 
JasonM said:
All modern GPU's have decode and compute ability. When I stream Amazon Music HD I can actually see the integrated GPU's decode usage increase, it then rests to 0% when I stop playback. GPU usage can be monitored in task manager.
Click to expand...

Thanks. I'm gonna take a look as i use 2 separate dacs, one hdmi and one usb. Will also have some reading to do there on that.
 
Quartz said:
Nice tip. Exactly how do you do this?
Click to expand...

First step obviously is to enable the iGPU and install drivers, I let Windows update do the driver install.

Then if you go --> Display Settings --> Graphics Settings. From here you can choose the exe to force Windows to use as priority. So in my case I found the amazon music.exe. Your add in graphics cards are still used for other activity as Windows 10 sees add in graphics cards as more powerful.

I do recommend going into power and sleep saving, and set High Performance on the Intel Graphics for the power plan your using.

Also you need to enable hardware acceleration in Spotiy or Amazon Music etc, so that program moves decoding to the GPU thats been set.

Final thing you have to do.

For some odd reason, even however all the above is done, Amazon Music was still sharing work load between my CUDA cores and iGPU. So to solve this you also have to go into NVidia Control Panel --> Manage 3D settings --> Program settings. Then set 'none' for the CUDA cores, then reboot the machine.

If you the play your streaming service, have open GPU-Z and you will see the load rise on the iGPU as you stream. You should also see it from task manage also.
 
Last edited:
The_Arbiter said:
Thanks. I'm gonna take a look as i use 2 separate dacs, one hdmi and one usb. Will also have some reading to do there on that.
Click to expand...

You will notice the advantage of this more, where you have multiple other CPU/GPU processes operating while your streaming your music. Effectively you are giving your music streaming it's own iGPU to use, the only disadvantage I see is the iGPU is sharing main memory, however it's still working out better for me.
 
Conversely I've found specifically in a multitasking environment disabling hardware acceleration gives great consistent results. My ancient 6700K has more than enough power to manage multiple browser video streams and Amazon HD all un-accelerated. Only when hardware acceleration is enabled do I start experiencing glitches...
 
a gpu should decode streaming with ease, the problem is likely something else.

IO spikes can cause cuda to slow down to a crawl during machine learning, like when you try to log the model data to a file.
the logs are 700kb files being written almost constantly.
suddenly cuda is 4x slower and my gpu usage is almost nothing when without logging it's around 70%
 
Just revisiting this old thread as it's worth it. I've been running like this now for almost a year and it's working really well. This being iGPU is handling Amazon Music HD, and other productive work my Quadro P620.

Zero issues with the Amazon Music, yet processing on the Quadro is left for all other work.

For anyone that used main GPU for productively work, and also listens to music streaming in the background, please don't ignore this post.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom