DXVA vs CUVID?

[Snappy Phoenix]

I have a GeForce GT555M with 2 GB DDR5 VRAM

I have always used CUVID in LAV as my settings but am wondering which is better?

I still think that the nVIDIA CUVID should render a better result for an nVIDIA GPU right?

Please correct me if Im wrong and I'd highly appreciate if you can tell me the differences between DXVA vs CUVID

[nevcairiel]

CUVID has the advantage that the hardware communication was directly developed by NVIDIA and therefor should (hopefully) get the values always just right.
A generic DXVA implementation may have some small issues, occasionally. In most cases, its fine, though.

DXVA might have slightly less power usage in some cases.

If CUVID works just fine for you, i would just stick with it.
On a laptop, i would try to compare power draw between the two modes, if your laptop offers precise information about current power usage.

[Snappy Phoenix]

CUVID has the advantage that the hardware communication was directly developed by NVIDIA and therefor should (hopefully) get the values always just right.
A generic DXVA implementation may have some small issues, occasionally. In most cases, its fine, though.

DXVA might have slightly less power usage in some cases.

If CUVID works just fine for you, i would just stick with it.
On a laptop, i would try to compare power draw between the two modes, if your laptop offers precise information about current power usage.

Ah I see, I guess I should just stick to LAV then

and I am not really concerned about the power because I'm always connected on AC Power

Cheers

[CruNcher]

@berryracer
There is a pretty easy rule if it works for you and you don't care about any resource usage then use what you prefer from other viewpoints then those pretty easy.
There are pros and cons of both and those are usage case specific as well there is no General way to say 1 is best the other is worse there both have their pros and cons from different viewpoints.
Though their is rarely Software who can make use of both efficiently based on these pros and cons they rather use only 1 or the other but they dont clever combine both ;)

[aufkrawall]

With CUVID my GPU stays in P0 state and starts to make noise.
Also, something of NVAPI is broken, GPU applications aren't correctly reported to the MDPS of Nvidia Inspector (fixed with Kepler but still broken with Fermi).
Annoying.

[dukey]

Speed wise there probably isn't much in it. CUVID is of course nvidia only.