HTPC Decoding and Rendering Benchmarks: EVR / EVR-CP

In our Ivy Bridge HTPC review, we had covered the CPU / GPU utilization during playback of various types of clips. In the Vision3D 252B review, we had graphs of CPU and GPU loading with various renderers and codecs. Unfortunately, AMD doesn't provide similar data / sensors for use with their APUs. Hence, we had to resort to power consumed at the wall along with GPU loading in the Trinity HTPC review. In order to keep benchmarking consistent across all HTPC reviews, we started adopting the Trinity HTPC review methodology starting with the review of the ASRock Vision HT.

The tables below present the results of running our HTPC rendering benchmark samples through various decoders when using the Enhanced Video Renderer / Enhanced Video Renderer (Custom Presenter) (EVR / EVR-CP). Entries in bold indicate that there were dropped frames which indicate that the unit wasn't up to the task for those types of streams. Fortunately, none of the streams presented any problem to the system and there were no dropped frames. The recorded values include the GPU loading and power consumed by the system at the wall when playing back the streams using MPC-HC v1.6.5.6366 and LAV Filters 0.54.

Enhanced Video Renderer (EVR)

The Enhanced Video Renderer is the default renderer made available by Windows 8. It is a lean renderer in terms of usage of system resources since most of the aspects are offloaded to the GPU drivers directly. EVR is mostly used in conjunction with native DXVA2 decoding.

LAV Video Decoder (DXVA2 Native) + EVR
Stream GPU Usage % Power Consumption
     
480i60 MPEG-2 24.05 35.04
576i50 H.264 21.38 36.06
720p60 H.264 26.13 36.6
1080i60 H.264 28.9 39.95
1080i60 MPEG-2 28.19 37.06
1080i60 VC-1 31.23 45.57
1080p60 H.264 30.11 37.09

The GPU is not taxed much by the EVR despite hardware decoding also taking place. Deinterlacing and other post processing aspects were left at the default settings in the Intel HD Graphics Control Panel (and these are applicable when EVR is chosen as the renderer)

Enhanced Video Renderer - Custom Presenter (EVR-CP)

EVR-CP is the default renderer used by MPC-HC. It is usually used in conjunction with MPC-HC's video decoders, some of which are DXVA-enabled. However, for our tests, we used the DXVA2 mode provided by the LAV Video Decoder.

LAV Video Decoder (DXVA2 Native) + EVR-CP
Stream GPU Usage % Power Consumption
     
480i60 MPEG-2 26.69 38.78
576i50 H.264 24.43 37.88
720p60 H.264 32.76 40.4
1080i60 H.264 40.16 42.02
1080i60 MPEG-2 39.75 41.62
1080i60 VC-1 40.99 48.45
1080p60 H.264 41.33 42

In addition to DXVA2 Native, we also used the QuickSync decoder developed by Eric Gur (an Intel applications engineer) and made available to the open source community. It makes use of the specialized decoder blocks available as part of the QuickSync engine in the GPU.

LAV Video Decoder (QuickSync / DXVA2 Copy-Back) + EVR-CP
Stream GPU Usage % Power Consumption
     
480i60 MPEG-2 27.16 38.42
576i50 H.264 25.26 38.05
720p60 H.264 36.84 41.6
1080i60 H.264 44.2 43.41
1080i60 MPEG-2 44.32 43.02
1080i60 VC-1 43.56 43.26
1080p60 H.264 48.28 45.13

In general, using the QuickSync decoder results in a higher power consumption because the decoded frames are copied back to the DRAM before being sent to the renderer. Using native DXVA decoding, the frames are directly passed to the renderer without the copy-back step. The odd-man out in the power numbers is the interlaced VC-1 clip, where QuickSync decoding is around 5W more efficient compared to 'native DXVA2'. This is because there is currently no support in the open source native DXVA2 decoders for interlaced VC-1, and hence,  it is done in software [Clarification: This restriction is only on Intel GPUs. On both AMD and NVIDIA cards, DXVA2 native decode acceleration is supported for all VC-1 streams]. On the other hand, the QuickSync decoder is able to handle it with the VC-1 bitstream decoder in the GPU.

 

Refresh Rate Handling HTPC Decoding and Rendering Benchmarks: madVR
Comments Locked

138 Comments

View All Comments

  • lexluthermiester - Monday, January 21, 2013 - link

    I've run many tests doing the 8 VS 7 comparisons. Not only have you left out many of the comparing scores for 7, but your power usage numbers don't seem right.

    I have a Kill-A-Watt power usage adapter that plugs right into the wall, which measures total power running through it. Last night I ran tests using Netflix and 8 vs 7 on two different notebooks. The first is a Gateway P-7811FX[P9700 2.8 ghz C2D, GF 9800m GTS] and the second a Toshiba L655-S5150[i5-480m 2.66ghz, Intel HD]. The numbers for the Gateway were 67.3[8], 64.4[8 app] and 66.1[7]. The Toshiba's numbers were a bit better, 50.9[8], 47.1[8 app] and 50.8[7].

    Now I could go one for days about game benchmarks and compatibility between 7 & 8. But real issue here is the your power numbers don't seem realistic. The Netflix app does have a small advantage, but nothing as dramatic as what you are showing. The difference between 8 VS 7 numbers are statistically insignificant. I ran battery tests as well. And those numbers are also statistically insignificant.

    I'm not willing to accept your results as valid and quantifiable, unless you declare the testing equipment used and OS config setup.
  • lexluthermiester - Monday, January 21, 2013 - link

    And before anyone states the obvious, yes the CPU's in each of those notebook were upgraded from their factory offerings. Both have 4GB of ram. The gateway has a WD 500gb 7200rpm HDD and the Toshiba has a Seagate 320GB 7200rpm. But remember these notebooks were not compared to each other, only themselves using one OS vs the other.
  • ganeshts - Monday, January 21, 2013 - link

    I stand by my numbers, and the tests were repeated multiple times to confirm this. Our configuration of the testbed itself is described in the first page of the review.

    For power measurement, we use the UFO Power Center from Visible Energy with a custom power measurement script described here:

    http://www.anandtech.com/show/6413/visible-energy-...

    I see that both the CPUs you used have old Intel HD Graphics (yes, Clarkdale and Arrandale were released when Intel HD Graphics wasn't that great). I think a lot of the advantage for the app version has to do with very good hardware decode acceleration (improved GPU and drivers).
  • Gigaplex - Monday, January 21, 2013 - link

    You're claiming that minimal power savings with old architecture CPUs on notebooks (were the batteries plugged in?) as measured by you invalidates Ganesh's measurements on a desktop platform? Just... no.
  • Galatian - Monday, January 21, 2013 - link

    Does the Windows 8 GUI actually bring something on the table for a couch potato like myself? I just retired my old gaming rig which now serves as a HTPC and I kinda tasted blood with Steam in Big Picture mode. I'm planning on running a power efficient yet graphically powerful HTPC once Haswell (or perhaps a new AMD A10) is out, but I would really like for Windows 8 to be completely controllable by an XBox controller? Has Microsoft actually included support for that or was their only new feature touch support?
  • lexluthermiester - Monday, January 21, 2013 - link

    You can control Windows 7 with a 360 controller? Didn't know that. Is this something natively supported? Or is it a hack? Wait... I'm on the net I'll look it up... Very cool if it works though.
  • Galatian - Monday, January 21, 2013 - link

    Never said it did...not sure were you red this in my post??? I was just curious because for me the tile based start menu of Windows 8 would seem to be a perfect fit for a XBox controller support, hence my question. At least this would be one place in my house were Windows 8 might be useful. If not I'll jus stick to Windows 7
  • Gigaplex - Monday, January 21, 2013 - link

    If your TV is 720p rather than 1080p, then absolutely not. "Metro" apps don't run at 1280x720, an error message pops up telling you to change the screen resolution whenever you try to launch one.

    And no, sadly an Xbox controller doesn't work with "Metro".
  • coolhund - Tuesday, January 22, 2013 - link

    LOL seriously?
    I mean I know that the minimum is 1024x768, but hell, everyone knows 720p is often used and thus I thought they were smart enough to allow that resolution, even if its not quite as high in the vertical. The stupidity of MS never ceases to amaze me...
  • powerarmour - Tuesday, January 22, 2013 - link

    Unfortunately the Metro/Modern UI is neither remote friendly, nor controller friendly.

    Waste of time for a HTPC tbh.

Log in

Don't have an account? Sign up now