The lack of a standardized HTPC GPU evaluation methodology always puts us in a quandary when covering the low end / integrated GPUs. Towards this, I had a long discussion with Andrew Van Til, Mathias Rauen and Hendrik Leppkes, all popular open source multimedia software developers. The methodology we developed is presented below.

The first step is to ensure that all the post processing steps work as expected. HQV benchmarking gives us an idea. Once the cards' post processed videos pass visual inspection, we need to gather an idea of how much time is left for the GPU to do further post processing activities. These may include specialized scaling algorithms, bit-depth etc. as implemented by custom MPC-HC shaders / renderers like madVR.

Deinterlacing and cadence detection are aspects which affect almost all HTPC users. Other aspects such as denoising, edge sharpening, dynamic contrast enhancement etc. are not needed in the mainstream HTPC user's usage scenario. Most mainstream videos being watched are either from a Blu-ray source or re-encoded offline or TV shows which need deinterlacing (if they are in 480i / 1080i format).
 

Denoising OFF Denoising ON
Under what circumstances would a GPU run out of steam for such post processing?

The intent of the benchmark is to first disable all post processing and check how fast the decoder can pump out decoded frames. In the typical scenario, we expect post processing to take more time than the decoding. Identifying the stage which decides the throughput of the decoded frames can give us an idea of whether we can put in more post processing steps. This is similar to a pipeline whose operating frequency is decided by the slowest stage. We then enable post processing steps one by one and see how the throughput is affected.

DXVAChecker enables us to measure the performance of the DXVA decoders. We use a standard set of 1080p / 1080i H264 / MPEG-2 and VC-1 clips. We also have 1080p DIVX / XVID and MS-MPEG4 clips. Cyberlink PowerDVD 11, Arcsoft Total Media Theater 5 and MPC-HC video decoders were registered under DirectShow. DXVA Checker was used to identify which codecs could take advantage of DXVA2 and capable of rendering under EVR for the sample clips. An interesting aspect to note was that none of the codecs could process 1080i VC-1 or the MPEG-4 clips with DXVA2.

Note that the results in the next section list all the cards being tested. However, the 6450s and GT 520 shouldn't really be considered with seriousness because of the issues pointed out in the previous sections.

Denoising Performance and ESVP on the 6450s DXVA Benchmarking
Comments Locked

70 Comments

View All Comments

  • qwertymac93 - Monday, June 13, 2011 - link

    What the heck are you talking about?
  • velis - Monday, June 13, 2011 - link

    A great review. Provides all the answers one could wish for and even gives some further hints.
    I sure hope you have something like this lined up for llano.

    If I may suggest a couple or three things:
    Perhaps you should also mention reclock - it will solve most 23.976 and similar problems... It's not like many will detect that the video is running 1/24000th faster. Plus it's insanely easy to use.
    I understand you couldn't just post full blown images for space problems, but those thumbnails require too much work too. Is it possible to display a popup of sorts when one mouse-overs those thumbnails?
    Also a vertical line showing 60FPS in those DXVA tests would be great :)
  • ganeshts - Tuesday, June 14, 2011 - link

    I will pass on your request(s) to the person in charge of the graphing engine :)
  • Salfalot - Monday, June 13, 2011 - link

    What might have been a nice option is to see what sound levels the cards produced. Even it was only for the GT430 and the HD6570. I know that the decibels can differ between manufacturers but it would have been nice!
    For the rest a very nice detailed review between HTPC cards. I was deciding which card to buy so this helped a great deal! I was only looking between the HD6450 and the HD6570 but the GT430 is a better option than the HD6450.
  • nevcairiel - Monday, June 13, 2011 - link

    HDMI Audio is purely digital, there is no diference based on what card you use.

    It depends on the audio decoder, and your receiver at the other end of the HDMI link, the HDMI sound card on those cards does not change the audio.
  • Salfalot - Monday, June 13, 2011 - link

    I think I did not use the right word, as I meant the levels of decibel the fan of the cards produce and not the audio too and through speakers.
    All reviewed cards have a fan on them and since most of the HTPC setups are in the living room it would have been nice to know which of the cards are most silent.
  • ganeshts - Monday, June 13, 2011 - link

    Though we considered cards with fans in this review, we made it a point to note that the same configuration (GPU model + DRAM bus width + operating frequencies) can be obtained with passive cooling from other vendors.

    For example, the 6570 has a passively cooled model from HIS with the same config and Zotac has a passively cooled 430 too. Other vendors have also demonstrated passively cooled models in Computex.
  • cjs150 - Monday, June 13, 2011 - link

    Firstly, a truly informative article. Very high quality.

    The fact that none of AMD, Intel and Nvidia can lock onto to the correct frame rates is unforgiveable. It is not as though these frame rates have changed over the last 6 months. It should not be necessary to be an advanced HTPC user and delve into custom creation of frame rates.

    I really hope that the representatives of AMD, Intel and NVidia are hanging their heads in shame at such basic errors - sadly I doubt they care.
  • Grasso789 - Monday, January 28, 2013 - link

    The mistake is rather with Microsoft. Video playback speed should be adapted to the refresh rate of the grafx card. There is a software called Reclock doing that. Then, for example 23,996 Hz can be run with a monitor refresh rate of n times 24 Hz. (The same with audio, because bit-perfect transmission only works with synchronization.) In the end and for most sources, the RAMDAC needed only (multiples of) 24, 25 and 30 Hz. In any system, one of its parts should be the clock master, while the other parts serve.
  • casteve - Monday, June 13, 2011 - link

    Excellent review, Ganesh! Your HTPC insight/reviews have been missed.

Log in

Don't have an account? Sign up now