We have been using our media streamer test suite for the last few reviews in the HTPC / media streamer space. In v1.0.1 of the test suite used in the WDTV Live Plus review, we added DVD and Blu-Ray folder structures to the mix. In v1.0.2, we have added two Xvid clips (one with Q-Pel and another with GMC (Global Motion Compensation) with 3 warp points) and a MKV with forced subtitles. Recently, we added Blu-Ray and DVD folder structures to the mix in v1.0.3.

For the Zino 410, we used v1.0.3 of the media streamer test suite. Weights have been assigned to each stream based on how frequently one might encounter each type of file in a home theater setting.

We now have 50 streams in our test suite and a maximum possible score of 358. Presenting a table listing every stream would make this section messy. Instead, moving forward, we will only indicate streams which have playback issues. In addition to that, for HTPCs, we will also indicate all the different softwares / codecs used to enable playback of all the test streams.

The Zino 410 HD HTPC scored 328 out of a maximum of 358 in our test suite. Not all files were played back from the same software. It was quite difficult to find the correct splitter / filter / renderer combinations for some of the test streams. The following softwares were used to process our media streamer test suite:

  1. Cyberlink PowerDVD 10.0.2025.52
  2. Total Media Theater 5
  3. MPC-HC with ffdshow Audio Decoder
  4. VLC 1.1.4
  5. Monogram GraphStudio with MPC-HC Matroska Splitter (for 024-1080p24.x264.DTS-MA.7Ch.mkv, which had minor stutters with the default MPC-HC combination used to playback other streams)
  6. Monogram GraphStudio with Sage Mpeg Demux Splitter (for 042-1080p24.VC1.TrueHD.AC3, which the latest MPC-HC M2TS splitter couldn't split properly. VLC could play this back without bitstreaming, but GraphStudio, with the Sage splitter was able to properly deliver the TrueHD soundtrack to ffdshow for bitstreaming).

Let us take a brief look at the test streams which had issues:

  1. 1080p60 8 reference frame H264 encode stuttered badly, and played back at a fraction of the expected frame rate. Full software decoding wasn't of much help either.
  2. 1080p24 VP8 in WebM couldn't be decoded at full frame rate  in VLC using the CPU. (Appearance of multi-threaded VP8 decoders in VLC might help in the future)
  3. 1080p24 RMVB couldn't be decoded at full frame rate in VLC using the CPU.
  4. Quality of 1080i deinterlacing (H264 and VC1) in our deinterlacing test streams was not as good as the result from some of our other testbeds. MPEG2 software deinterlacing in VLC (Yadif2) was excellent without too much CPU usage.

The graph below shows how the Zino 410 stacks up against other SFF HTPCs with respect to media / codec compatibility. While it is obviously much better than the low cost Atom / Fusion solutions, it sadly falls behind the Core 100 by a little bit.

AnandTech Media Streamer Test Suite

However, some video enthusiasts might still prefer the Mobility 5450 based Zino 410 to the Core 100.

Refresh Rate Handling

Many HTPC purists are concerned about the lack of support for proper display refresh rates. Intel's hardware bug in the Arrandales and Clarkdales (as well as Sandy Bridge) makes this one area in which ATI's efforts stand out. nVidia drifts the refresh rate back and forth to achieve 23.976 Hz when averaged over time, ATI maintains and locks the refresh rate perfectly. MPC-HC's statistics OSD indicates that the display correctly refreshes at 23.976 Hz with the 23 Hz setting for the Zino 410.

Reference screenshots for the Core 100 and Vision 3D are provided below:

Gaming with the ATI Mobility 5450 Network Streaming
Comments Locked

69 Comments

View All Comments

  • silverblue - Saturday, February 19, 2011 - link

    ...would having two differently-sized SO-DIMMs have on system performance?

    I'd be tempted to take replace that 4GB module with a 2GB one just to see what happens. 6GB of RAM just doesn't compute. :)

    We seem to be getting a decent number of Dell-AMD systems lately... I only hope they take up Brazos with the same level of enthusiasm, because even if it did result in a small drop in performance, this review would've been largely the same in terms of gaming and video playback/quality, albeit with a much smaller footprint. Also, in that scenario, dual channel wouldn't matter as Fusion doesn't support it.
  • fabarati - Saturday, February 19, 2011 - link

    If AMD has anything like Intels asynchronous dual channel, the first 4 GB will perform like dual channel, whilst the remaining 2 GB will perform like single channel.
  • Taft12 - Saturday, February 19, 2011 - link

    Dual-channel memory was a scam from the beginning that has somehow survived to this day to make PC buyers think they needed to buy more memory than needed.

    Have you seen the benchmarks? ~1-2% benefit AT MOST for anything that's not a synthetic memory bandwidth test, regardless of platform.

    http://www.tomshardware.com/reviews/PARALLEL-PROCE...
  • fabarati - Saturday, February 19, 2011 - link

    Didn't it help back in the P4 and/or P-M days? And how did it do in early Athlon 64 days?

    But yeah, Core Duo and newer doesn't really benefit from Dual channel, one
  • silverblue - Saturday, February 19, 2011 - link

    Theoretically, dual channel would help APUs as they're bandwidth-limited.

    I suppose you're right about standard usage though, even raising memory clocks doesn't make for a sizeable performance advantage.
  • asmoma - Saturday, February 19, 2011 - link

    Starcraft 2 is not a synthetic memory test, and it does benefit from going from 2 to three channels, just read some performance reviews of Starcraft 2.
  • DanNeely - Saturday, February 19, 2011 - link

    The einstien@home applications benefitted from a 3rd channel on i7 quad cores; and the 2nd channel on C2 quads as well; high performance server farms/clusters/super computers are a significant segment of the market. On the i7-quad, E@H had a 66% speedup from the 2nd channel, and a 5% gain from the 3rd. Sandybridge gave a similar speedup from the 2nd channel. I can't find the thread with the C2Quad results, but IIRC they were a 10-25% speedup.

    http://einstein.phys.uwm.edu/forum_thread.php?id=8...

    I don't have benchmarks handy, but I suspect a heavily loaded DB server would also benefit from the extra memory bandwidth because the queries would result in a psudorandom memory access pattern that would limit the ability of the cache controller to prefetch most of the data being requested.
  • jeremyshaw - Friday, March 4, 2011 - link

    that actually had to do with uncore clocks, not memory channels.
  • vailr - Saturday, February 19, 2011 - link

    How would a Mac Mini compare?
    Could a Mac Mini be retrofitted with a Blu-Ray drive (since Blu-Ray is available factory installed) and then run as a Windows HTPC?
  • tipoo - Saturday, February 19, 2011 - link

    Why on earth would you buy a Mini to run as a Windows HTPC? You'd pay more for less performance.

Log in

Don't have an account? Sign up now