| french | english |
 |
|
| Sign in | |
| 20-05-08 / 17:15 : EastWest PLAY : deceiving on PowerPC ? (cjed) |
| Since the abandon of Kompakt licence for EastWest samples libraries in profit of their own PLAY samples player (not easy to reference such a generic name in Google... without counting the tech support that isn't public anymore - through private messages -, that limits the found occurences when doing searches and then reduces the visibility of informations and problems about PLAY), many users of PowerPC macs (including Dual G5 ranging from 2.0 to 2,7 Ghz) complain about poor to inacceptable performances. Since PLAY was promoted as allowing to get rid of some relative Kontakt performances issues (K2 is in fact said to be far less efficient than EXS24) - and avoid the Kontakt 2 Player licence price on the EastWest side -, it seems PLAY development was done completely with cross platform considerations, without any handling of OSX particularities or specific optimizations . Then it is clearly stated that performances are better on Intel mac than on PowerPC, despite the Apple vector APIs (Accelerate framework) being able to manage transparently (same code) PPC Altivec and Intel SSE : Soundsonline thread : How to optimize an audio application for PPC & Intel macs. ADC - audiocommunity : Developing an Audio Unit Plug-in / 4. Optimize Performance Moreover PLAY suffers from an unpredictable bug that makes it unusable with some sounds : audio cuts appear without any reason (low polyphony used, low CPU usage in OSX activity monitor, whatever single G4 or powerful dual G5) and completely randomly. The CPU ressources meter in PLAY is also reporting wrong values (audio cuts appear sometimes even with only 20% CPU usage displayed by the software - and with convolution reverberation off). About the later, when activated on less powerful macs, the reverb very quickly produces harsh audio overloads (it can lead to making the OSX MIDI Server not responding anymore, requiring a reboot) and the PLAY CPU resources meter doesn't display accurate value (a little higher than with the reverb off, where in fact the OSX activity monitor shows huge jump). It seems the PLAY audio engine doesn't feed as fast as needed the CoreAudio buffer (while the CPU isn't even overloaded), what leads to audio cuts or full audio overloads, as by design CoreAudio has to ensure a min latency. Setting the latency higher (larger buffer) doesn't fix the problem however. Soundsonline thread : Tip to improve PLAY perf on PowerPC macs. Some useful explanations from an Apple CoreAudio team member : CoreAudio and overload messages Globally PLAY seems to have been developed with basic Intel code to work on Windows and OSX. For example the user interface (GUI) uses the Qt cross-platform framework, and when displaying the main Player window of PLAY the CPU usage raises 50% more (ok with Browser window displayed). There is also a not nice white flashing when switching from one of these windows to the other (probably no double buffering when Qt draws the window content). |
| Comments |
I noticed that if I switch the audio frequency from one value to another (for example 44,1Khz, then 48, and then back 44,1) in the audio settings (standalone PLAY or AU host) after having loaded each PLAY patch (MOR in my tests, not tested in VOP but it should also work), the audio cuts artefacts are gone and even the Telecaster Lead guitar of MOR is playable with long arpeggios and sustains (and that with streaming on !). I saw this advice in some forum about something else (another software that i don't remember). |
| Write a comment |