[SOLVED] Multiple Graphics card - Putting one to sleep while keeping the other active

[TrufflesG]

For about 10 years I've been hoping this would become a thing.

The reason is that when I have a $1000+ graphics card in your machine I really only want to use it when playing demanding games or processing graphics intense applications like CAD programs.

The rest of the time, I'd love to have it power down and go into hibernation and switch over to my MUCH less expensive graphics solution....all done via software.

That would save wear and tear and run time on my expensive graphics card until I actually need it. Does this exist already?

Is this a good idea? If not, why not?

 

[InvalidError]

It exists on laptops with models that use an output mux chip to switch between IGP and dGPU depending on power plan.

On the desktop, since GPU and IGP outputs are physically separate, you'd have to physically switch the display cables between IGP and dGPU to achieve something similar.

Alternately, you could output through the IGP and make as much stuff as possible use the IGP by default, though this would introduce extra latency while gaming from the dGPU having to pump frames out through the IGP for output. You could also do it the other way around using the dGPU for output but then the dGPU would never be able to power down since it needs to be always on to route the IGP's output to monitors.

With even high-end GPUs only pulling 20-30W in typical everyday desktop stuff, the amount of wear from non-gaming/3D/GPGPU work should be insignificant. If your $1000 GPU fails from this, it was destined for premature failure from the start.

 

[TrufflesG]

It exists on laptops with models that use an output mux chip to switch between IGP and dGPU depending on power plan.

On the desktop, since GPU and IGP outputs are physically separate, you'd have to physically switch the display cables between IGP and dGPU to achieve something similar.

Alternately, you could output through the IGP and make as much stuff as possible use the IGP by default, though this would introduce extra latency while gaming from the dGPU having to pump frames out through the IGP for output. You could also do it the other way around using the dGPU for output but then the dGPU would never be able to power down since it needs to be always on to route the IGP's output to monitors.

With even high-end GPUs only pulling 20-30W in typical everyday desktop stuff, the amount of wear from non-gaming/3D/GPGPU work should be insignificant. If your $1000 GPU fails from this, it was destined for premature failure from the start.

The number one reason for a lot of failures is cooling fan failure. There is always dust in the air and human skin / hair and pet hair or even rug strands that gets into the fans. The longer they are running, the more the fans degrade.

But I do see your point.

 

[InvalidError]

The number one reason for a lot of failures is cooling fan failure. There is always dust in the air and human skin / hair and pet hair or even rug strands that gets into the fans. The longer they are running, the more the fans degrade.

Dust doesn't harm fans unless you have an outrageous amount of it and don't clean it out once every now and then. A computer case with decent air filters should catch 90+% of the dust as long as you remember to clean them regularly. The air filters on my Antec 300v2 catch enough dust that I only bother deep-cleaning my case once every 2-3 years.