Runtime power management for Nvidia GPU

[doncherian]

I am trying to implement a feature with which i can issue suspend/resume to the GPU explicitly on Demand.The idea is to suspend the GPU while keeping other part of system alive and start/resume it only when it is needed to minimize the power consumption.

We have a GeForce GT 430 GPU connected to PCIe bus on Intel® Atom™ Processor D510 machine running with kernel 2.6.31-15 and nvidia driver - NVIDIA-Linux-x86-270.41.19.

After going through the presentation by Rafael J. Wysocki [http://linuxplumbersconf.org/2010/oc...untime_PM.pdf] , I understood that the device driver should provide callbacks for resume and suspend if it supports the run-time power management.
I checked the kernel interaction layer of nvidia driver and could see that those two callbacks are supported.

<NVIDIA-Linux-x86-270.41.19\kernel\nv.c>
static struct pci_driver nv_pci_driver = {
.name = "nvidia",
.id_table = nv_pci_table,
.probe = nv_kern_probe,
#if defined(NV_PM_SUPPORT_DEVICE_DRIVER_MODEL)
.suspend = nv_kern_suspend,
.resume = nv_kern_resume,
#endif
};

First of all i would like to know if it is possible to implement such a feature.If so please provide me some pointers on how to do this..

Thanks in advance,
Don