More > Media Center 17

SACD ISO buffering issues

(1/4) > >>

andrewt:
I'm testing out playing SACD ISOs and MC is having to pause to buffer approx. every 10-20 seconds or so (stereo and MCH). I have the iso on an external drive which I thought might be the bottleneck, so I copied it to my C drive (which is a SSD) and it's still the same. Any thoughts? Here is my setup:

ISO downsampled by MC to 24/96 (max supported by my soundcard)
ASIO to PreSonus FireStudio Project soundcard (24/96, 8 output channels)
Vista, 3 (or so) GB RAM, AMD Turion 64 dual-core processor

I tried downsampling to 44.1 and that didn't change anything. I can play 5-channel 24/96 FLACs just fine. I also notice that MC sets my soundcard buffer to 8192 ms (which I assume is the largest). I don't notice the CPU or RAM being stressed during playback or buffering.

Matt:

--- Quote from: andrewt on July 11, 2012, 09:18:16 pm ---I'm testing out playing SACD ISOs and MC is having to pause to buffer approx. every 10-20 seconds or so (stereo and MCH). I have the iso on an external drive which I thought might be the bottleneck, so I copied it to my C drive (which is a SSD) and it's still the same. Any thoughts? Here is my setup:

ISO downsampled by MC to 24/96 (max supported by my soundcard)
ASIO to PreSonus FireStudio Project soundcard (24/96, 8 output channels)
Vista, 3 (or so) GB RAM, AMD Turion 64 dual-core processor

I tried downsampling to 44.1 and that didn't change anything. I can play 5-channel 24/96 FLACs just fine. I also notice that MC sets my soundcard buffer to 8192 ms (which I assume is the largest). I don't notice the CPU or RAM being stressed during playback or buffering.

--- End quote ---

It's most likely a CPU issue.  SACD ISO files that use DST compression and more than two channels require a lot of horsepower to decode.

What is your JRMark (Help > Benchmark)?

andrewt:
Here's my benchmark while running Firefox and the usual background stuff:

--- Code: ---=== Running Benchmarks (please do not interrupt) ===

Running 'Math' benchmark...
    Single-threaded integer math... 13.465 seconds
    Single-threaded floating point math... 5.473 seconds
    Multi-threaded integer math... 13.740 seconds
    Multi-threaded mixed math... 5.693 seconds
Score: 495

Running 'Image' benchmark...
    Image creation / destruction... 3.668 seconds
    Flood filling... 1.654 seconds
    Direct copying... 3.473 seconds
    Small renders... 9.098 seconds
    Bilinear rendering... 8.875 seconds
    Bicubic rendering... 6.743 seconds
Score: 657

Running 'Database' benchmark...
    Create database... 1.124 seconds
    Populate database... 4.180 seconds
    Save database... 0.506 seconds
    Reload database... 0.151 seconds
    Search database... 5.029 seconds
    Sort database... 3.405 seconds
    Group database... 2.141 seconds
Score: 1300

JRMark (version 17.0.180): 817
--- End code ---

Not sure how good that is. My computer is pretty old, though--CPU is 1.8GHz dual-core. However, I don't really see the CPU being stressed; while playing and buffering a MCH track, the CPU stays in the 60% range according to the Sidebar gadget. Although it could be another CPU issue that I'm not aware of.

Edit/sidenote: Just wanted to thank you, Matt (and Jim and everyone else) for making such a kick-a** product and responding directly to user feedback and questions; you're the only company I know of that does that. MC does everything I want and more. I just got done recommending MC to a friend who wants to play some FLAC files, and I recommend MC to anyone not attached to the seemingly ubiquitous iUmbilicalCord

Matt:

--- Quote from: andrewt on July 11, 2012, 10:17:58 pm ---JRMark (version 17.0.180): 817

Not sure how good that is. My computer is pretty old, though--CPU is 1.8GHz dual-core. However, I don't really see the CPU being stressed; while playing and buffering a MCH track, the CPU stays in the 60% range according to the Sidebar gadget. Although it could be another CPU issue that I'm not aware of.

--- End quote ---

Decoding the SACD is mostly single-threaded, which is why you're seeing 60% usage of a dual-core.  That likely means it's running one core at full speed, and it's not quite enough.

An 817 JRMark may just not be able to handle real-time DST decompression across multiple channels with resampling.  To compare, a medium range computer scores around 2000 and a fast computer will score over 4000.

andrewt:
Gotcha... time to look forward to the day when I can afford a new computer... Thanks Matt!

Navigation

[0] Message Index

[#] Next page