Print

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

[Iving]

Evaluating JR MC 28 as potential new user.
I play flac music. Don't do anything else.
Problem: intermittent pause for 2-3 secs at consistent point just before end of music track.
Very probably contingent on playing from memory - which I value for clear SQ delta.
Cannot alleviate with settings inc Device buffer / PreBuffering.
Playing flac on Optane PCIe AIC > upsample 176.4 in Sox > decoded file in memory > ASIO > Dante Virtual Soundcard
JRiver specific - doesn't happen in e.g. fb2k.
Any ideas please?
Thank you

[JimH]

Simplify by avoiding any upsampling and memory playback.  If that works, add back, one step at a time.

[Iving]

Simplify by avoiding any upsampling and memory playback.  If that works, add back, one step at a time.

Thanks Jim,

As I say - I already am confident that the problem is contingent on playing from memory. And the SQ delta is very significant.

If I can't upsample x 4 (or to 176.4) or play from memory the value of JRiver would be hugely reduced to me - and I'd likely not purchase. But never mind me and my disappointment! Why would a User not be able to do these simple things? My PC is more than spec. for it. No DSP or anything.

I'd be very surprised if this isn't a known problem.

[Matt]

It could be hitting the drive to load the whole file.  If the drive is slow, it could be a problem.

Please try the other kind of memory playback or no memory playback.

[Iving]

It could be hitting the drive to load the whole file.  If the drive is slow, it could be a problem.

Please try the other kind of memory playback or no memory playback.

As said the drive is an Optane PCIe AIC - doesn't get much faster.

As said I value memory playback for SQ - and why would it not work?

[Iving]

Please try the other kind of memory playback or no memory playback.

What is "the other kind of memory playback" please?

[Matt]

Load the full file or the decoded file.  Loading decoded works in a thread.  Loading the full file blocks.

[Iving]

Load the full file or the decoded file.  Loading decoded works in a thread.  Loading the full file blocks.

Right ...

The play from memory options I can see are:
- file not decoded
- album not decoded
- file decoded

I don't think "album" - I play individual flac tracks.
I favour file decoded because that's all the work done > RAM > playback. That's the whole point of playing from memory.
File not decoded would leave upsampling to do after loading into memory - defeats object of playing from memory?

[Dawgincontrol]

I have this intermittently. 

It seems related to changing from different bitdepth/bitrates between tracks in a playlist (ie 192 or 88khz to 44.1khz).  Happens with about 5 seconds left in the track or thereabouts.

[Iving]

I have this intermittently. 

It seems related to changing from different bitdepth/bitrates between tracks in a playlist (ie 192 or 88khz to 44.1khz).  Happens with about 5 seconds left in the track or thereabouts.

Sounds like exact same problem.

But my files are all 44.1 flac without exception.

I haven't found a remedy in settings.

I would need to resolve to buy JRiver.

Unpredictable pausing at end of track is too much of a defect in playback experience.

So frustrating because JRiver play from memory is winning on SQ.

[zybex]

You should use 'File decoded' as it's supposed to be non-blocking, according to Matt.

Try configuring your antivirus to exclude FLAC files (or temporarily disable it for testing). It may be causing a freeze when the file is accessed while it checks it. Same for any other security software you may have.

Also make sure you have the latest Optane drivers.

[Iving]

You should use 'File decoded' as it's supposed to be non-blocking.

Try configuring your antivirus to exclude FLAC files (or temporarily disable it for testing). It may be causing a pause when the file is accessed while it checks it. Same for any other security software you may have.

Also make sure you have the latest Optane drivers.

Hi -

I *am* using file decoded as explained. This is the mode I value and want.

When you say "non-blocking" you mean thread?

I am offline with Defender disabled - no antivirus.

Drivers are all good.

The problem is JRiver specific.

I do think that the solution is one of:

1. Someone who understands JRiver under the hood/bonnet is able to argue which settings should remedy to the best possible extent. We try that and it works. But I say - I have played with a lot of settings - certainly all the buffer ones. But what is the *argument* based on how JRiver works in flac > upsample to 176.4 in Sox (no DSP) > play from memory > ASIO.

2. Settings can't fix it but some other code solution can be found.

3. Live with the defect having paid the subscription.

4. Don't buy.

I really want to subscribe because I like the SQ of JRiver in play from memory mode.

[zybex]

Yes, Matt says that mode loads the file on a separate thread, meaning that the main thread that is playing the current file is not blocked.

When you use Windows explorer to access the files, is there any delay on showing the folder contents? What about when you copy a file to some other device (network/usb/disk)?

The non-connected might be a factor... could you check to see if it also happens when the PC is connected to network/internet?

</ideas>

[Iving]

Yes, Matt says that mode loads the file on a separate thread, meaning that the main thread that is playing the current file is not blocked.

When you use Windows explorer to access the files, is there any delay on showing the folder contents? What about when you copy a file to some other device (network/usb/disk)?

The non-connected might be a factor... could you check to see if it also happens when the PC is connected to network/internet?

</ideas>

Windows explorer delay? No.

There are no noticeable performance issues on this PC which does nothing except play music.

I have gone online for a few seconds a few times just to register software. Nothing to report. Plug in the cable. Register. Instant. Off.

My playback is ASIO > Dante RedNet - so that is a network - but it is offline. I have had no Dante/RedNet issues of this kind except the one I am reporting now with JRiver.

I am fairly sure JRiver doesn't do this when not in play from memory mode. That has to be a very strong clue.

[zybex]

OK. I meant having the PC online while playing a file to reproduce the issue. The reasoning being, perhaps MC is trying to check for an updated CODEC or something like that and is blocking until the DNS/ping check times out (this is just conjecture).

I am fairly sure JRiver doesn't do this when not in play from memory mode. That has to be a very strong clue.

Perhaps it does, but then the freeze would happen between tracks where you wouldn't notice it.

[Iving]

OK. I meant having the PC online while playing a file to reproduce the issue. The reasoning being, perhaps MC is trying to check for an updated CODEC or something like that and is blocking until the DNS/ping check times out (this is just conjecture).

If I accepted that that was a plausible explanation, I'd have to go online (anathema to me - this is an *offline* audiophile machine) with my unprotected PC - and wait indefinitely/over a long period - watching all the time at the end of every track - to log this intermittent problem.

What could MC be looking for online towards the end of every track to play the next 44.1 flac? That doesn't seem plausible enough to warrant the online experiment. Why would MC be designed to have to check online every time it played a Redbook rip.

This looks like it is reading the next file in prebuffering - readying itself to play the next file from memory. But the prebuffer settings do not help.

[Iving]

Perhaps it does, but then the freeze would happen between tracks where you wouldn't notice it.

Between tracks - gapless - MC is phoning in to get a CODEC for 44.1 flac and I am not noticing?

[zybex]

Yes, you would need to do that to rule out the possibility. I understand your reluctance to do that, though as long as there's a router in the way and you're not browsing the web, it's not that dangerous.

I'm just throwing possibilities. I don't know how MC is coded, but I am a developer so I do know some possibilities. Nowadays it's normal for all kinds of apps to go online for a number of reasons, so this would not be so uncommon. MC is using some lib to decode FLAC; often these are 3rd party libs, not developed in house (again, I don't know if this is the case here). So it's possible that the lib itself has this functionality, and each time it's invoked to decode a file, it tries to do that.

MC itself downloads components/decoders as it needs them. So it's not unreasonable to think that it may check if there's a new version of the FLAC decoder each time this decoder is invoked.

I think you found a valid bug, and all this is just to try to find the root cause so that JRiver can fix it...

[Iving]

Yes, you would need to do that to rule out the possibility. I understand your reluctance to do that, though as long as there's a router in the way and you're not browsing the web, it's not that dangerous.

I'm just throwing possibilities. I don't know how MC is coded, but I am a developer so I do know some possibilities. Nowadays it's normal for all kinds of apps to go online for a number of reasons, so this would not be so uncommon. MC is using some lib to decode FLAC; often these are 3rd party libs, not developed in house (again, I don't know if this is the case here). So it's possible that the lib itself has this functionality, and each time it's invoked to decode a file, it tries to do that.

MC itself downloads components/decoders as it needs them. So it's not unreasonable to think that it may check if there's a new version of the FLAC decoder each time this decoder is invoked.

If this is really true, then JRiver is designed for online use only - and I cannot use it.

Why would MC not have flac decoders onboard. Seems beyond sense.

[zybex]

Between tracks - gapless - MC is phoning in to get a CODEC for 44.1 flac and I am not noticing?

Gapless, of course you would notice. You seem a bit angry here, but I'm just throwing ideas to try to find the root cause of your problem, which is apparently not shared by most users.

[zybex]

Why would MC not have flac decoders onboard. Seems beyond sense.

... it does have them, doesn't it? Or else it wouldn't play at all. I mentioned 'updates' several times (edit: ok, actually just once)

[Iving]

Gapless, of course you would notice. You seem a bit angry here, but I'm just throwing ideas to try to find the root cause of your problem, which is apparently not shared by most users.

Please - not angry. No drama if you don't mind.

I would like to understand the cause of this effect. I'd like a solution. I want to buy JRiver.

[zybex]

Great
The point of trying to find a root cause is that, you'll find, JRiver team is very responsive to fix issues as they're found. If the cause is found, they'll have a fix in a few days for you.

The problem is usually in convincing them that it's not some user/environment issue... and since this problem is not common, that's the default assumption unfortunately. I agree with you that the main Playback thread should not EVER freeze.

[Iving]

Great
The point of trying to find a root cause is that, you'll find, JRiver team is very responsive to fix issues as they're found. If the cause is found, they'll have a fix in a few days for you.

The problem is usually in convincing them that it's not some user/environment issue... and since this problem is not common, that's the default assumption unfortunately. I agree with you that the main Playback thread should not EVER freeze.

Thank you for your responses. I am not going to try going online unprotected for an extended period. But I will carry out further diagnostic tests of my own - and report back if I find anything useful.

[AGAWA]

SOX does upsampling on the fly on decoded file loaded to memory ? Or decoded and upsampled file is load to memory ready to be played? How does it really work
Is this problem happening always in the same time measured from the beginning of the track?
Just trying to influence your thinking.

[Iving]

SOX does upsampling on the fly on decoded file loaded to memory ? Or decoded and upsampled file is load to memory ready to be played? How does it really work
Is this problem happening always in the same time measured from the beginning of the track?
Just trying to influence your thinking.

As said, the intermittent pause is always just moments before the end of a track. Remarkably consistent. So it does seem as though there is a stall in prep. for playing next. As said, problem may happen only when playing from memory. Decoded file in memory setting should mean that MC is reading flac / upsampling in Sox / new thread / sending to RAM vs. non memory playback > ASIO.

If this decoded memory option is decoded file (thread) vs. undecoded file (block) I cannot see/understand "SOX does upsampling on the fly on decoded file loaded to memory ?". Sox must have done its work already.

No DSP involved btw.

[DJLegba]

As said, the intermittent pause is always just moments before the end of a track. Remarkably consistent. So it does seem as though there is a stall in prep. for playing next.

You have probably tested this, but is there a pause when you play only a single track with nothing else queued?

[AGAWA]

SOX - this is exactly the question: when SOX is doing the job?
On the fly when file is being played or beforehand, when loading memory?
My guess it that this is on the fly during playback. But I may be wrong.

[zybex]

Also, did you try without upsampling, just to check if that's the source of the issue?

SoX is just an algorithm for upsampling. I think the actual flow is:
- load FLAC into memory - let's say, 60MB for a 5 minute track
- decode FLAC in memory - needs around 100MB for a 5 minute, 44.1 x 32 bits x 2 channels file
- upsample to 176Khz in memory - buffer now needs 400 MB of RAM (or as AGAWA says, perhaps this is only done in realtime/playback)

These memory needs might increase for more channels - up to 1.2 GB usage could be expected for 6 channels, which could trigger windows to do some pagefile swapping. Even more if your tracks are 30 minutes instead of 5 minutes.
Since you have an Optane drive, I assume your PC also has loads of RAM so I don't expect this to be a problem.

[Iving]

trigger windows to do some pagefile swapping

Is a pagefile *required* for JR MC do you know?

[Iving]

You have probably tested this, but is there a pause when you play only a single track with nothing else queued?

Unlikely to occur. Problem is intermittent.

Please explain why you ask?

[AGAWA]

just checked on Mac Pro , Catalina, MC28
44,1 /24 FLAC files.

No problems, but 5 seconds before end of each track , in playback window (top of the screen with track info) there is a blink - and "Waiting" is displayed for a moment.

[Iving]

SOX - this is exactly the question: when SOX is doing the job?
On the fly when file is being played or beforehand, when loading memory?
My guess it that this is on the fly during playback. But I may be wrong.

It would be great to have an authoritative answer.

[Iving]

Since you have an Optane drive, I assume your PC also has loads of RAM so I don't expect this to be a problem.

Correct - RAM is not an issue and RAM usage is low as a % at all times.

[Iving]

just check on Mac Pro , Catalina, MC28
44,1 /24 FLAC files.

No problems, but 5 seconds before end of each track , in playback window (top of the screen with track info) there is a blink - and "Waiting" is displayed for a moment.

Very interesting!
What exactly is this "waiting"?

[zybex]

Is a pagefile *required* for JR MC do you know?

No, pagefile is only required if all your loaded applications+system require more memory that what is available.

[DJLegba]

Unlikely to occur. Problem is intermittent.

Please explain why you ask?

I ask because you said "So it does seem as though there is a stall in prep. for playing next." I thought you may have tested your hypothesis.

[Iving]

I ask because you said "So it does seem as though there is a stall in prep. for playing next." I thought you may have tested your hypothesis.

OK
I'm new to MC so please bear with me.
This would require Playing Now to be empty except for the single playing track, right? In other words no prep. for ensuing track required?
Most of the time the pause is not going to happen because the problem is intermittent.
I can't give a ratio as I have to be very free of ordinary obligations to sit and wait for many tracks to pass and log occurrence rate!
It did happen with Bonnie Tyler "Lost In France" - so I played it again thinking it might be her. Not so unfortunately.

[JimH]

Iving,
You may not be angry, but you're debating a lot of the valid suggestions being made.  It would help your cause if you'd just test as advised.

The pause seems to occur when the next file is being read.  Reading the file slowly would explain it.  I think you've said that's not possible.  Windows Defender is running on Win10 and has to be configured.  Reading from a NAS can be slow.  Etc.  Decoding or encoding could explain it.

That you only go online a few seconds doesn't give you much protection from the outside world.  Computers are pretty quick.  Your router would not normally allow inbound access to your machine unless you've set up port forwarding.

FLAC decoding doesn't require Internet access.

[DJLegba]

OK
I'm new to MC so please bear with me.
This would require Playing Now to be empty except for the single playing track, right? In other words no prep. for ensuing track required?
Most of the time the pause is not going to happen because the problem is intermittent.
I can't give a ratio as I have to be very free of ordinary obligations to sit and wait for many tracks to pass and log occurrence rate!
It did happen with Bonnie Tyler "Lost In France" - so I played it again thinking it might be her. Not so unfortunately.

This is helpful. Does your drive go to sleep? It may not make sense that waiting for a drive to wake from sleep could affect playback of a file already loaded into memory, but if you can disable sleep mode it's a simple thing to test.

[zybex]

An Optane drive is a very fast SSD. It wakes up in microseconds.
https://www.anandtech.com/show/12136/the-intel-optane-ssd-900p-480gb-review/8

However, if there's another C: drive which is not an SSD, it could cause that issue with spin up time.

[Iving]

Iving,
You may not be angry, but you're debating a lot of the valid suggestions being made.

I promise I am not angry. Please don't expect me not to "debate".

The pause seems to occur when the next file is being read.

Exactly so!

I am anticipating that we can identify what "waiting" means and take things from there if settings (prebuffer etc) cannot resolve.

Reading the file slowly would explain it.  I think you've said that's not possible. 

I have a fast Optane PCI-e drive. The drivers are good / up to date. It behaves perfectly in all respects. fb2k play is fine. I hope to move on to JRiver.

Reading from a NAS can be slow.  Etc.  Decoding or encoding could explain it.

I don't use a NAS. My flac are on the Optane drive.

That you only go online a few seconds doesn't give you much protection from the outside world.  Computers are pretty quick.  Your router would not normally allow inbound access to your machine unless you've set up port forwarding.

I don't need protection from the outside world. I am an offline user.

FLAC decoding doesn't require Internet access.

Good to know.

[Iving]

This is helpful. Does your drive go to sleep? It may not make sense that waiting for a drive to wake from sleep could affect playback of a file already loaded into memory, but if you can disable sleep mode it's a simple thing to test.

My drive does not go to sleep ever. Power settings make sure of this. I assure this in different ways. Anyway - flac is playing or the machine is off.

[Iving]

An Optane drive is a very fast SSD. It wakes up in microseconds.
https://www.anandtech.com/show/12136/the-intel-optane-ssd-900p-480gb-review/8

However, if there's another C: drive which is not an SSD, it could cause that issue with spin up time.

There is only one drive. The Optane PCI-e AIC. It is 900P 480 Gb. It has the o/s Windows 10 Pro and my flac files. Nothing else except Programs inc. JRiver!

[Dawgincontrol]

As I too have had the problem, let it be known I have tried all memory playbacks including none.  I have my drives (yes I have multiple) set to not sleep.  I have tried both SOX and not using SOX.  I increased my memory thinking that may be the problem.  Sooner or later the pause/wait happened again. 

It's there.  Like I stated before, mine seems to have a relation to different bitdepths/bitrates.  It only happens when playing tracks in a playlist with combinations, but not all the time.  I seamlessly go from 2 channels to 5 and 6 channel and back with no problem, but 2 channels with different bitdepths are where it happens.  It's difficult to pinpoint.

One more thing.  I do have to convert 88,200HZ to 96,000 or 44,100 to play through my receiver/dac.  If memory serves me correctly (no pun intended), that is where mine mostly "hangs".

I don't think the gentleman is angry; just frustrated.

[Iving]

As I too have had the problem, let it be known I have tried all memory playbacks including none.  I have my drives (yes I have multiple) set to not sleep.  I have tried both SOX and not using SOX.  I increased my memory thinking that may be the problem.  Sooner or later the pause/wait happened again. 

It's there.  Like I stated before, mine seems to have a relation to different bitdepths/bitrates.  It only happens when playing tracks in a playlist with combinations, but not all the time.  I seamlessly go from 2 channels to 5 and 6 channel and back with no problem, but 2 channels with different bitdepths are where it happens.  It's difficult to pinpoint.

I don't think the gentleman is angry; just frustrated.

Thank you for this.

It's quite correct I am not angry! No drama please!

So the problem exists without play from memory too. I haven't been able to confirm as the problem is intermittent and my focus is on trying to get play from memory working.

My flac are 44.1 without exception.

It's a most curious thing. If it weren't for this anomaly I'd have paid my subscription by now.

[Matt]

I've used loading decoded files since we added the feature.

It gives me a warm fuzzy that my APE files are in memory exactly like a WAV file.  There simply could never be an audible difference.

I haven't had delays switching tracks, so I'm not much help.

[Iving]

I've used loading decoded files since we added the feature.

It gives me a warm fuzzy that my APE files are in memory exactly like a WAV file.  There simply could never be an audible difference.

I haven't had delays switching tracks, so I'm not much help.

Am I reading correctly - you are denying a SQ delta (to your own ears anyway) between playing decoded in memory and not playing from memory?

sorry lol

warm fuzzy is about right - the sound is fuller, deeper, bass not as distinct, digital edge mitigated, but high volume confirms no distortion - esp. compared with fb2k.

Great product. Just stalling mysteriously!

[Matt]

Audiophiles can sometimes argue that the weight of decoding an APE (or FLAC) file somehow makes the SQ worse, even if the bits are the same.

But with that option enabled, the file is fully decoded in a second.  So then there's just really no difference in APE, WAV, FLAC.

I'm not a believer that it really matters, but I like it for the sake of MC being the best audiophile player!

[Iving]

Audiophiles can sometimes argue that the weight of decoding an APE (or FLAC) file somehow makes the SQ worse, even if the bits are the same.

But with that option enabled, the file is fully decoded in a second.  So then there's just really no difference in APE, WAV, FLAC.

I'm not a believer that it really matters, but I like it for the sake of MC being the best audiophile player!

Audiophiles argue all the time about all sorts of things!

But there is a clear SQ elevation to me in play from memory. That's why I'm persisting. If I A/B with fb2k [easy to do], fb2k has slightly more detail/edge - but is "emptier" - and cannot win vs. JRiver at high volumes - distortion gives it away. JRiver without play from memory is between the two.

Arguably JRiver play from memory is not strictly audiophile! But it sounds great. And it's not distorted. What more could you want?

Plays all the way through track-to-track? :-)