Specs: Macbook
OS X 10.6.5
2.16 Intel Core 2 Duo
2 GB 667 DDR2 SDRAM
I put together a live set to perform last week, which consisted of rendered patterns in separate tracks for some fun live-style performing. In addition to this, I had two tracks devoted to pre-rendered tunes to dj (16 bit 410000 wav, converted to flac when saving the set).
The pattern-length elements performed excellent, but when reaching the segments of the set that included complete rendered flacs (averaging around 3 minutes in length per track) the tunes would sporadically skip. I’ve tested again and am still experiencing this. the skips are ranging from a few milliseconds to roughly 1/2 second) and happen in bursts lasting up to 30 seconds. I’m not running any other software, no vsts, and only the native renoise compressor.
I’ve tried:
turning autoseek off
turning off PDC
turning off GUI effects
and turning off auto save
the cpu remains below 5% the entire set. if this isn’t a bug, is there any reason why the tracks are skipping?
I’ve tested on the firewire 410, and my integrated audio.
Is your macbook configured to work on full performance or are there some CPU efficiency and idling tools spoiling your performance?
I don’t know how it is doing RAM buffering, but if it also does so when you have started your Macbook from scratch and loaded your project in a fresh started environment, i guess it may be looking for the power management options on your Mac book and try to set any eco or low profile mode to full performance and test further.
Not running any performance tools. Turned off power management (put hd to sleep when possible and brightness reduction). Snow Leopard no longer has adjustable CPU preferences. Testing now and the crackling isn’t happening, so it seems fixed at the moment.
These are some fairly light adjustments however, so this could be a confound. I’m speculating it’s related to ram allocation, though with my specs, it’s bizarre that renoise would show any semblance of a heart-palpitation while playing a single wav or flac.
It depends if you have many samples loaded in memory and they all require large blocks of memory, if these samples are split and segmented across your RAM you can get this effects easily and it doesn’t matter what platform you have, RAM fragmentation can occur and it turns a lot clearer when you have large files loaded into ram.
This should cause less problems with 64-bit though, but that can only be tested properly in Linux.
Could this be a future option? (i.e. ram allocation settings with large wavs); an old pc with a whopping 128mb of ram can easily make its way through a playlist of wavs.
ok. thanks for clearing that up guys. with the addition of autoseek, DFD seems pretty important for those using renoise for live performance.
my dream has been to use renoise exclusively (for both performance and composition). i performed in SF last week and had lots of crackles near the middle of my set containing pre-renders. i was terrified, so I cut my set short by jumping to the short live performance segments.
Before DFD, if I cut up my tracks into pattern-length segments, the issue shouldn’t occur?
Mushen: How big are the files in total. Do they fit (in theory) in the RAM?
DFD indeed would be great to have here, but well, this is nothing we can fix away now for 2.7.
The project will be quite huge, but can you maybe nevertheless share it with us, so that we could do some more tests and at least try to help you finding some workarounds?
The wav sizes ranged from around 30-55mb, and the XRNS contained 6 wavs, 1 mp3. Summed, 272.6mb though the project compressed to roughly 200mb.
With only Renoise active, looping an empty pattern in an empty xrns, the OSX native activity monitor shows that I have slightly above 1GB free (the rest dedicated to the OS, hardware, etc). To my understanding, they should fit in the RAM with plenty of headroom. Though when I load and begin the XRNS, my free RAM drops to 80 MB; freshly loaded, playing the first tune, with only the Renoise-native compressor on the master track.
Yes, but stuff is not compressed in memory. Also mp3 material gets expanded to full blown 32-bit wave files.
If all that stuff is approx. 200 mb compressed, it could expand to five times its size which is roughly near the 1GB you speak of.
But that all depends on the native bitsize, amount of channels and frequency that your samples are using.
