Beta testers needed for fancy new waveforms.

General Mixxx discussion.... discuss

Moderator: garth

Beta testers needed for fancy new waveforms.

Postby rryan » Thu Mar 08, 2012 3:33 pm

Hey Mixxxers,

We're hard at work on getting 1.11.0 ready to roll. One of the new features will be a totally rewritten, 3-band colored waveform. Since this is a pretty big change we want to be careful to get the kinks worked out early. If you're feeling brave, we would love your feedback:

A new beta is available, check here the Mixxx 1.11.0 beta1 thread for more details.

Download beta builds here:
Mac OS X (10.5 32/64-bit)
Windows 64-bit
Windows 32-bit

Notes:
  • Check Preferences -> Interface to change between waveform type and FPS
  • GLSL Waveform Renderer *does not work right now*

When giving feedback, please report:
  • Your graphics card make and model
  • Your OS (e.g. Windows 7 64-bit, Mac OS X 10.6.8)
  • What waveform renderer and FPS you are using (e.g. GL or Filtered) -- check in Preferences -> Interface
  • The CPU usage you saw while using the beta
  • The CPU usage you see using Mixxx 1.10.0

Please file a bug here if you run into issues: https://bugs.launchpad.net/mixxx/+filebug and please mention that you are using the waveform-2.0 beta.

Thanks for the help!
RJ
The Mixxx Manual, Wiki and FAQ are the best place to start for documentation and support.
Please report any bugs you find to our Bug Tracker.
rryan
Mixxx Developer
 
Posts: 1335
Joined: Thu Mar 26, 2009 12:14 am
Location: California

Re: Beta testers needed for fancy new waveforms.

Postby mbochon » Thu Mar 08, 2012 5:48 pm

Hi Ryan,

I would like to test it on Windows XP (and Ubuntu if needed) but for now I can only build Mixxx on Linux. Is there a compiled version online I could download ?

Regards,
Maxime
mbochon
 
Posts: 4
Joined: Thu Mar 08, 2012 12:48 pm

Re: Beta testers needed for fancy new waveforms.

Postby rryan » Thu Mar 08, 2012 5:53 pm

Hi mbochon,

Did the Windows 32-bit or 64-bit installers I linked above not work?

RJ
The Mixxx Manual, Wiki and FAQ are the best place to start for documentation and support.
Please report any bugs you find to our Bug Tracker.
rryan
Mixxx Developer
 
Posts: 1335
Joined: Thu Mar 26, 2009 12:14 am
Location: California

Re: Beta testers needed for fancy new waveforms.

Postby Irwin Cespedes » Thu Mar 08, 2012 6:13 pm

Woa! Those waveforms look gorgeous, although, they bring some heat to the CPU. Let me give you what I'm using:

GPU: Intel® 82945G Express Chipset
OS: Windows XP SP3
Waveform: Filtered [email protected]
CPU Usage, Mixxx Beta: 50% (Intel Core 2 [email protected])
CPU Usage, Mixxx 1.10.0: Less than 12%. Some peaks to 25% when the BPM is discovered, but then is stabilized to 10%~12%

One thing I have noticed is that those waveforms tend to speed up the CPU fan due to the high amount of processing. One example, the first test I have done the departure Temperature was 55ºC, and then, after a couple of songs rendered it climbed up to 70ºC. I'm going to test at night, mainly because is too hot in here at noon :-)

Aside of that, wwaveforms look OK and EQ killing works as it suppose to do. Btw, Is there a BZR brach for testing on GNU/Linux?

Way to go!
I am also known as altiplΛne, this is my music

¡Pura Vida! Image
User avatar
Irwin Cespedes
 
Posts: 170
Joined: Mon Feb 01, 2010 8:59 pm
Location: Puntarenas, Costa Rica

Re: Beta testers needed for fancy new waveforms.

Postby rryan » Thu Mar 08, 2012 6:54 pm

Thanks Irwin :)

The branch is here: https://code.launchpad.net/~mixxxcontri ... eforme-2.0

Yes -- I've noticed high CPU usage on the GL waveform as well. We haven't dug in yet and started profiling it to try and speed up the slow parts. Hopefully there is low-hanging fruit.
The Mixxx Manual, Wiki and FAQ are the best place to start for documentation and support.
Please report any bugs you find to our Bug Tracker.
rryan
Mixxx Developer
 
Posts: 1335
Joined: Thu Mar 26, 2009 12:14 am
Location: California

Re: Beta testers needed for fancy new waveforms.

Postby Irwin Cespedes » Fri Mar 09, 2012 2:23 am

As promised, a second long shot test with a screenshot.

Image

After 1 hour mixing, everything works OK. Same stuff as in morning. Although, I have noticed when a song has too much treble or mids the waveform gets a little bit stop, as if the FPS goes down. It has the same behaviour of a game when have too much to render at same time and creates a visual gap. I'm still on my Windows Box, so as soon as I have time to log in to my pengüin Box, I'll post again :-)

¡Pura vida!
I am also known as altiplΛne, this is my music

¡Pura Vida! Image
User avatar
Irwin Cespedes
 
Posts: 170
Joined: Mon Feb 01, 2010 8:59 pm
Location: Puntarenas, Costa Rica

Re: Beta testers needed for fancy new waveforms.

Postby mbochon » Fri Mar 09, 2012 6:00 pm

Oops I missed the links :D probably not a good idea to browse the forum from my mobile in public transport...

I had access to a Windows 7 desktop computer. Its features are displayed in the attachment (*features.png).

mixxx111a_mbo_win7_features.png
Features
mixxx111a_mbo_win7_features.png (109.03 KiB) Viewed 14363 times


Here are my feedbacks:

- When switching from one mode to another, the mixxx interface behind the preference dialog window flashes (white filled rectangle) before displaying the waveform again (this is just an aesthetic remark).

- The 'Filtered (GL)' rendering method seems not to be position-independent, leading to some unpleasing blinking. This phenomenon is due to stains on the rendered surface which change for each time position, as shown on the 3 dedicated attachments (*pos_dep_stains.png).

mixxx111a_mbo_win7_pos_dep_stains.png
Position Dependent Stain
mixxx111a_mbo_win7_pos_dep_stains.png (208.08 KiB) Viewed 14363 times


- There seems to be a slow and steady memory leak. I ran both instances of Mixxx (1.11 Alpha having the 'Filtered (GL)' activated), then loaded the same tracks without playing them, and let them run for a few hours with no interaction at all. The memory usage and processor time of 1.11 rised up slowly to approximately 1GB/1h45 while 1.10 stayed calm, as shown on attachment (*memory_leak.png).

mixxx111a_mbo_win7_memory_leak.png
Memory leak
mixxx111a_mbo_win7_memory_leak.png (33.03 KiB) Viewed 14363 times


I'll do some other testing later if possible.
mbochon
 
Posts: 4
Joined: Thu Mar 08, 2012 12:48 pm

Re: Beta testers needed for fancy new waveforms.

Postby mbochon » Fri Mar 09, 2012 9:54 pm

Windows XP SP3 / Intel Core 2 Duo / GeForce GT240M 1GB
  • Mixxx 1.10 : CPU=52%
  • Mixxx 1.11a Filtered @ 33Hz : CPU=6%, FPS=32
  • Mixxx 1.11a Filtered @ 99Hz : CPU=14%, FPS=62
  • Mixxx 1.11a Filtered(GL) @ 33Hz : CPU=7%, FPS=32
  • Mixxx 1.11a Filtered(GL) @ 99Hz : CPU=18%, FPS=62
Remarks:
  • the frame rate parameter in the preference window is not saved
  • the blinking surface issue is the same as on my Windows 7 test
mbochon
 
Posts: 4
Joined: Thu Mar 08, 2012 12:48 pm

Re: Beta testers needed for fancy new waveforms.

Postby rryan » Fri Mar 09, 2012 10:17 pm

Whoa -- those are some nice performance numbers compared with 1.10 mbochon. How many tracks did you have loaded (if any) in either of those cases?

Also, thanks a lot for your details infodump in your previous post :)
The Mixxx Manual, Wiki and FAQ are the best place to start for documentation and support.
Please report any bugs you find to our Bug Tracker.
rryan
Mixxx Developer
 
Posts: 1335
Joined: Thu Mar 26, 2009 12:14 am
Location: California

Re: Beta testers needed for fancy new waveforms.

Postby mbochon » Sat Mar 10, 2012 11:43 am

You're welcome.
On all the previous CPU tests, Mixxx was only playing one track. And where I get 52% using Windows, I get 26% using Ubuntu (however, the measurement methods may be different from an OS to another).

Now about the waveform display itself : it would be great to be able to choose between really different ones, depending on our personal preferences or on the visual information fitting our music best. If I can spend some on this, I'll investigate this since it must have already been done on the wiki or on a launchpad blueprint. They are so many ways to explore : temporal amplitude, loundness, perceived loundness, panning information, spectrum, timbre, noise, spectral centroid, peak-to-rms ratio, and most importantly, how to get a well-tuned auto-adaptative combination of a bunch of these parameters.

Maxime
mbochon
 
Posts: 4
Joined: Thu Mar 08, 2012 12:48 pm

Next

Return to General Discussion

Who is online

Users browsing this forum: No registered users and 3 guests