Build 76 still skips

[Roj]

Now testing on my work machine - a totally differnt box. This brings it to three dirfferent boxes, three different sound cards, three different motherboards, three different sound cards, three differnt NICs.

Do the following:

Play something in the player - I don't care what decoder you use.

Start Windows Explorer.

Right click in the right pane and choose New / Folder. Do it again. And again. And again. Sooner or later (and my guess is that it's sooner), the player will replay the last fraction of a second of the song it's currently playing.

"That, Mr. Anderson, is the sound of inevitability... " :)

So, sorry Paul but she ain't cooked yet...

[Sheepeh]

I can't replicate this...

AthlonXP 2600+
1GB ram
Soundblaster Live! 5.1 Digital
Build B76
Directsound X-Fader

even the crossfader works exactly as it should, it's seamless MP3 playback, and a nice fading effect that works very well, for my system it's ready to go final from the looks of it.

What soundcard are you using? Could it be something in your system can't cope with it? Maybe your hard drive is too full/slow/fragged?

[krazyd]

I also can't replicate this.

[Roj]

I also can't replicate this.Have either of you turned of all (and I do mean ALL) fading stuff? I have all fading turned off, all crossfading turned off, the output set to my soundcard (in the three cases it was a M-Audio Revo, an Audigy and a Santa Cruz) and adjust volume system wide turned off.

[Rex_Mundi_Incarnit]

Me neither, you sure your harddrive is not in PIO?

[Roj]

Me neither, you sure your harddrive is not in PIO?Seriously? :)

On all three machines? In one case it's SATA RAID, in another a Promise controller and in the third a mobo connector. NONE are in PIO mode. Besides, if they were were, I'd get scratching noises in the sound stream.

I just replicated it twice in quick succession on this machine (Audigy / MSI KT4V-L / Athlon XP 2000+ / Western Digital 40Gb BB / XP Pro SP-1) and all I did was go to my root directory and create five new folders as fast as I could one after another in Explorer.

[Roj]

For the record, the system I first discovered it on:

Athlon XP 2600+ Barton
512Mb DDR333
80Gb / 120Gb Western Digital JB
Radeon 9000 Pro
M-Audio Revolution 7.1
Windows XP Pro / SP-1
Diskeeper 8 Pro defrags automatically in Smart Set-It-And-Forget-It mode

Slow and fragged she ain't.

QCD 4.11 didn't do this on ANY of my systems, not even my old K6-III 450 w/384Mb PC100.

[Tokelil]

Just did a lot of copying (5 different folders) while doing some other stuff and it did skip for me. My buffer is reduced to 2000ms with the crossfader plug-in though. (All fade turned off) Doesn't happen offen though. Maybe once or twice a week I experience a skip.

[Roj]

Just did a lot of copying (5 different folders) while doing some other stuff and it did skip for me. My buffer is reduced to 2000ms with the crossfader plug-in though. (All fade turned off) Doesn't happen offen though. Maybe once or twice a week I experience a skip.Oddly enough it never happens with massive I/O on my systems (and I do mean **massive** - try un-RARing a 4.5GB ISO image from 90 RAR files while using PAR2 to repair another RAR set - both your processor and hard drive are pegged).

However I do have a question:

Does turning off all the fading and the crossfader have any affect on buffering, even if I don't touch the settings and leave them at default?

[Paul]

The only way the directsound buffer would loop is if QCD is being starved of all cpu. There are high-priority threads checking the DS buffer position so if it runs out playback stops. The fact that it loops means Windows itself is not scheduling QCD. Now, I can't reproduce this bug, and even if I could, I don't think it's QCD's bug to begin with.

[Roj]

The only way the directsound buffer would loop is if QCD is being starved of all cpu. There are high-priority threads checking the DS buffer position so if it runs out playback stops. The fact that it loops means Windows itself is not scheduling QCD. Now, I can't reproduce this bug, and even if I could, I don't think it's QCD's bug to begin with.
But in something as ubiquitous as EXPLORER? And why would 4.11 not do this? What changed?

[Roj]

The only way the directsound buffer would loop is if QCD is being starved of all cpu. There are high-priority threads checking the DS buffer position so if it runs out playback stops. The fact that it loops means Windows itself is not scheduling QCD. Now, I can't reproduce this bug, and even if I could, I don't think it's QCD's bug to begin with.Look at the attached picture. It's a crop of a screen shot of Windows Explorer, specifically the right pane (I use a tree display in the left pane). The edge of that left pane with the folder tree can just barely be seen. See those new folders at the bottom of the right pane list? I created those one after the other just now and the player skipped three times. Just from that. I'm playing OGG files encoded at Q8. There's nothing else running on the system which is an Athlon XP 2000+. This is what we're talking about. If I can replicate this on three machines I can't understand why someone else can't.

[Paul]

What does your CPU go up to when creating those folders?

[krazyd]

On XP2600, 512MB RAM, XP Home I had Firefox, Word and QCD open and with the settings you suggested on the crossfader (all fading off). Created 10 new folders in explorer as fast as I could without QCD missing a beat. Player and Input plugin (MP4) priorities set on 'normal'.

[Roj]

What does your CPU go up to when creating those folders?That was one helluva good call Paul.:) See the attached image - sorry it's blurry. I left the Right Click / New dialog open and listened to the song loop and loop and loop and loop while the processor stayed pegged at 100%...

In short, this is at least partially an Explorer bug as you likely envisioned because even when my processor is pegged at 100% when doing a PAR2 recovery for an hour at a time, the skipping does not occur.

Is there anything you can do to work around this?

[Roj]

On XP2600, 512MB RAM, XP Home I had Firefox, Word and QCD open and with the settings you suggested on the crossfader (all fading off). Created 10 new folders in explorer as fast as I could without QCD missing a beat. Player and Input plugin (MP4) priorities set on 'normal'.
Another reason I see this easily where others may not is that I have all the cutesy-poo animation and fading crap of the XP interface turned OFF. My menu display delay is similarly shortened to almost zero with TweakUI. That makes things happen immediately and may cause the situation to arise much more readily.

[Paul]

That was one helluva good call Paul.:) See the attached image - sorry it's blurry. I left the Right Click / New dialog open and listened to the song loop and loop and loop and loop while the processor stayed pegged at 100%...

In short, this is at least partially an Explorer bug as you likely envisioned because even when my processor is pegged at 100% when doing a PAR2 recovery for an hour at a time, the skipping does not occur.

Is there anything you can do to work around this?
I still don't see it as being related to QCD. Why would Explorer use any cpu at all to hold a menu open? There is another issue at work here.

[Roj]

I still don't see it as being related to QCD. Why would Explorer use any cpu at all to hold a menu open? There is another issue at work here.I'm at home now and testing on my primary system, the Athlon XP 2600+. I have narrowed it down to essentially one type of activity: creation of a new document using the right click button in Explorer. Essentially, if I go into the New menu after right clicking, CPU utilization goes up (but does not peak) and if I select Folder and create one, that's it - it's skip-to-my-lou time.

And that's the ONLY time it happens.

Seriously.

Too frikkin' weird.

What do you know about what happens when the context menu to create a new document is selected?

...and WHY (and I do believe this is absolutely key) does 4.11 NOT exhibit this behavior?

[hedge]

I'm at home now and testing on my primary system, the Athlon XP 2600+. I have narrowed it down to essentially one type of activity: creation of a new document using the right click button in Explorer. Essentially, if I go into the New menu after right clicking, CPU utilization goes up (but does not peak) and if I select Folder and create one, that's it - it's skip-to-my-lou time.

And that's the ONLY time it happens.

Seriously.

Too frikkin' weird.

What do you know about what happens when the context menu to create a new document is selected?

...and WHY (and I do believe this is absolutely key) does 4.11 NOT exhibit this behavior?
Just a total guess, but maybe its got something to do with one click items in explorer? Because i cannot reproduce this bug on a athlon 2700, and my cpu usage on right click > new only hits about 25%...
I can remember something like happening when i was using my mum's old computer, a p3 733mhz comp., but right click > new would have hit that a lot harder, and that was using winamp 2.7 or something cruddy.

[acozz]

Roj, try messing with your buffer size. In 4.11 the default buffer was 7000ms and in the new builds it's 2000ms.

[Roj]

Just a total guess, but maybe its got something to do with one click items in explorer? Because i cannot reproduce this bug on a athlon 2700, and my cpu usage on right click > new only hits about 25%...
I can remember something like happening when i was using my mum's old computer, a p3 733mhz comp., but right click > new would have hit that a lot harder, and that was using winamp 2.7 or something cruddy.
I do have single-click turned on...

[Roj]

Roj, try messing with your buffer size. In 4.11 the default buffer was 7000ms and in the new builds it's 2000ms.
Wow! That's quite a difference. I never mess with those defaults so I had not noticed. Thanks for the "head's up" - I'll give it s shot and report back.

[Roj]

Roj, try messing with your buffer size. In 4.11 the default buffer was 7000ms and in the new builds it's 2000ms.
acozz, yer a god!

Problem solved - many molto muchas gracias!

The moral of the story:

"Thou Shalt Not Be Penny Wise And Pound Foolish - RAM Is Cheap!!!"

More explicitly:

"Change The Default Back!"

Thanks to all who helped - my QCD experience is once again complete bliss.

[Sheepeh]

=) Good to see yer all sorted out mate!

[matty28carter]

All fixed here too!!