Here's a pro-tip if you're using VSO Video converter on a notebook that meets the following specifications:
1. You are using a notebook with an Nvidia Optimus setup, and;
2. The Discrete Nvidia GPU supports Nvidia NVENC (Supported on Kepler and Maxwell GPUs at the moment).
3. The integrated Intel GPU is at least a Sandybridge SKU (for QuickSync support).
If that's the case, keep reading.
ConvertXtoVideo can use BOTH QuickSync & NVENC simultaneously to greatly speed up batch video transcodes!
Steps to follow:
1. Open the Nvidia Control panel, and under 3D settings, ensure that VSO ConvertXtoDVD is set to run on Integrated Graphics (Force), not autoselected.
2. Launch ConvertXtoVideo and under settings , go to Conversion, and under the Video sub-menu, navigate to the Hardware Optimizations section and confirm ( and set) the following options:
(a). Toggle to ON the Enable all H/W optimizations (ON by default, confirm that is the case).
(b). Maximum (number) of hardware decoder (DXVA2) instances in same time: Set to 4 (Very important).
(c). Now, set the hardware accelerator priority queue as follows under the Hardware/Software encoding panel:
i. Set NVIDIA NVENC to number 1.
ii. Set Intel QuickSync to number 2.
iii. Set Nvidia CUDA to number 3 (No longer important as current drivers disable NVCUVENC API support, set to deprecated. This setting is placebo unless you're on a driver prior to R337).
iv. Set Intel QuickSync software to the last option. It will show Max instances supported as Zero (0) and that is okay (as you don't have the Intel Media SDK installed, by default, with ConvertXtoVideo).
(d). Now, under the Conversion options menu, ensure that the following options are enabled (Tweak depending on your core count. Remember, maximum simultaneous conversions cannot exceed 4 due to limitations with NVENC on consumer NVIDIA GPUs. Your mileage with mobile Quadro GPUs may differ).
i. Number of used cores for decoding: Match your logical processor count. With hyper-threading enabled, multiply core count * 2. Example, for a Quad Core i7 4700HQ, this would be 8, whereas a dual core i7 640M would be 4.
ii. Do the same for number of used cores for decoding.
iii. Maximum simultaneous conversions: Fix this at 4 if you have a consumer grade Nvidia GPU on your optimus setup. If you have a Quadro SKU, this would be the number indicated by maximum concurrent encodes for NVENC by the application under Hardware Optimizations configuration sub-menu + 2 (for Intel QuickSync).
iv. Minimum free memory and muxer buffer size can be left at defaults. (safe).
v. Set process priority as per your needs.
Click OK to save your preferences, queue up videos for transcodes and enjoy your Nvidia Optimus setup at it's full potential
Now, to explain why you must associate the application with the Intel IGP despite the fact that you'll be using Nvidia's NVENC: Quicksync requires hardware decoding via DXVA to run on the Intel IGP prior to activation, else it will fail or fail over to software mode (painfully slow and unwarranted), whereas for Nvidia NVENC, the NVENC SIP block can be initialized directly via CUDA calls that do not require explicit hardware decoding on the Nvidia GPU (default on Linux and Windows XP), or via DirectX mode (optional on Windows 7 and above in supported use-cases such as in-game video capture via NVIFR in a fashion similar to Shadowplay's implementation).
Regards,
Brainiarc7.
PS: This is my first post here. Be gentle.
My handle on all forums online is Brainiarc7. Even on overclockers UK, phoronix, github, twitter and elsewhere.
You can reply for further clarification on the same.