Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[jacobacci]

Hi everyone
When I play mch SACD ISOs, MC plays for a few seconds and then starts to re-buffer. Then it plays again for a few seconds and re-buffers again. 2ch tracks on the ISO play fine.

My HTPC is a Gigabyte BXBT 1900 with a j1900 Celeron processor, 8G of RAM + SSD.
The ISO is on a SanDisk Extreme 64G USB3 Stick (245MB/s read speed) attached to the USB3 port. While playing, the processor is at 45% usage, so it does not seem to be the bottleneck.

Any idea what I need to do to make mch tracks on SACD ISOs play smoothly?

[Hendrik]

Multi-Channel SACD is very CPU intensive to play, that system may just not be powerful enough.
You could try using a lower sample rate for output (in DSP Output Format), this can sometimes help to reduce CPU usage.

[blgentry]

Presumably you're decoding the SACDs into PCM?  I.E. your DAC is not capable of mutli-channel DSD is it?  Decoding DSD to PCM is intensive.  I know you said the CPU is under 50%, but that might not tell the whole story.  That might be an aggregate of several cores, which means that one core is completely pegged while the other is idle (for example).

I think you need something like a JMark of 1400 or so for 2 channel SACD decoding.  I'm not sure on multi-channel SACD.  What is your JMark score?  (Help > Benchmark > Run Benchmark).

Brian.

[jamezua]

Do you have Prebuffering 6 seconds and Bitstreaming YES (DSD) ?  ?

[jacobacci]

Thanks, guys. I think the answers have resolved the apparently contradictory low CPU usage.
My system has a jMark score of 1272, barely enough for 2ch SACD decoding, which works as intended. However mch SACD decoding appears to be beyond the j1900 CPU.
Brian's remark that the conversion may be saturating one of the four cores seems plausible to me.
So I will have to get a better NUC if I want to do on the fly mch SACD decoding. Thinking of getting an i5 skylake NUC. Hope that will work. Passmark of the i5 6260u is 4272 vs.  1890 for the Celeron j1900. So I think the i5 NUC should do the job.

[blgentry]

Don't take my word.  If you're inclined to believe the numbers, do some research before buying new hardware.  This was just the number that came to my head.  I'm not confident in that number.  I might be right.  I might not be. 

Brian.

[jacobacci]

I did a bit of checking before coming to the decision:
j1900, passmark 1890: decodes SACD stereo, but rebuffers heavily with SACD mch, the USB3 stick not being the bottleneck
i5 3320m (Thinkpad T430s), passmark 3399: decodes both SACD stereo and mch without issues
i5 6260u (NUC6i5syk), passmark 4272 is bit more powerful than the i5 in the T430s, so I guess it should work.

The jMark score of the j1900 would also indicate that it can barely do 2ch SACD.

[kamimeguro]

Hi, Just saw this thread on SACD ISO.

I have also an issue with a dedicated laptop (Lenovo X220, Win 10/64, i7-2620M, 8GB), JR Mark 3091 (21030) ~ 3136 (21027).
All files are on a USB3 RAID 5, but I did copy one ISO on the C: drive and got roughly the same results:

On SACD ISO Multi Channels:
After the initial buffering and CPU intensive work (100% 12 ~ 15 sec) the memory usage is stable at 40% (3+GB) and less than 10% CPU or disk usage but I get every 30 seconds or so a heavy (100%+) CPU usage during 12 ~15 sec, leading to sound stuttering...

On SACD ISO Stereo (from same ISO or different):
After the initial buffering and CPU intensive work (100% 12 ~ 15 sec) the memory usage is stable at 40% (3+GB) and less than 10% CPU or disk usage usage but I get every 100 seconds or so a heavy (100%+) CPU usage during 9 ~12 sec, leading to sound stuttering...

On Windows task manager, only Media Center is the only application going up to 100%+

I tried to remove all effects (output format = source number of channels and no other DSP) beside the PCM conversion to 24/176.4 for all channels, went back to 027 but got the same.
I also tried DFF and DSF (5 or 6 Channels) files and they play fine without reaching more than CPU 30% (even with output channels to 5.1 and room correction on)

I am not listening regularly to SACD ISO, so cannot tell when this appeared, but this looks directly related to the buffering and 'decoding' of the ISO.

Any idea on what is happening? Any advice will be worth a try!

Kami