u-he

swindus · Post by **swindus** » Wed May 20, 2015 6:41 am

Is there a Zebra HZ version for LInux?

Beamboom · Post by **Beamboom** » Thu May 21, 2015 7:03 am

I stumbled across an issue yesterday. It was experienced in FilterscapeVA on Linux but I would guess it applies to more of the VSTs, I also assume it will be the same across all platforms:

It does not handle handle well (or at all) the case if a preset file is renamed/moved/deleted while being active. If I physically (in the file system) move or delete preset files while the VST is active, the preset is still listed in the preset lists and the plugin will totally hang if I try to browse through any of the (no longer existing) presets. On my Linux/Jack/Bitwig stack it even seemed to kill the Bitwig audio engine.
I think there should be an exception handling in the case of a file no longer accessible (for whatever reason).

Reason I found it was that I was adding some downloaded presets, and later moved them to their own directory instead. A not too unrealistic scenario.

psycha0s · Post by **psycha0s** » Thu May 21, 2015 8:19 am

swindus wrote:Is there a Zebra HZ version for LInux?

Yes, Zebra HZ is ported to Linux. If you own it, try to PM abique.

Lump · Post by **Lump** » Thu May 21, 2015 5:19 pm

I don't know why ? but with Zebra for example, the LINUX sound is better than PC and OS-X with the same equipment !

I think this is a very subjective pov.

On my computer, the same software combination (BWS and U-He VSTs / Zebra, Diva and Hive) sound less better because of the poor performance of the Linux audio system. Often crackles and glitches. Same hardware, same DAW, same VSTs - but with OSX this combo runs 10 times smoother then with Ubuntu /KXstudio as OS. Yes I know I have to compile my own RT kernel, then everything should be fine.

I've tried to use Linux since 1999 for music production and one day it will work for me. But I don't see to "compile my own RT kernel" or make other improvements to use a Linux OS for music. Thats the job of the distributors, but they still didn't recognize the minority of music production system users, they still focused in office and server.

But still I think, a good step, the U-He VSTs for Linux. Maybe in 2019, after 20 years, Linux distributors will eventually support us and I can use OOTB more then 2 instances of Diva in divine mode.

BTW I've purchased a license for Hive lately...

psycha0s · Post by **psycha0s** » Thu May 21, 2015 6:15 pm

I don't use RT kernel and I don't do any crazy tricks for LL audio. My system runs with low buffers (128) without any cracks using pure ALSA. I can easily run several Diva instances (great + multicore mode) simultaneously. Moreover, I can use several Windows plugins (Dune2, FabFilter/Voxengo plugins) through Airwave VST bridge without noticeable performance issues. So Linux is pretty usable for music production today.

PS: I have a pretty fast system (i7 2600 @ 3.9 GHz, 16GB RAM, SSD). I've changed the CPU governor to "performance" and that's all I did.

Beamboom · Post by **Beamboom** » Thu May 21, 2015 7:27 pm

Lump wrote:On my computer, the same software combination (BWS and U-He VSTs / Zebra, Diva and Hive) sound less better because of the poor performance of the Linux audio system. Often crackles and glitches. Same hardware, same DAW, same VSTs - but with OSX this combo runs 10 times smoother then with Ubuntu /KXstudio as OS. Yes I know I have to compile my own RT kernel, then everything should be fine.

Sorry but that's just nonsense - you're not a Linux user. "Compile your own kernel"? Hello, the 90s called and wanted their Linux setup challenges back.

Here's how you do it if you use Ubuntu (the supported distro for Bitwig) or really any debian-derived distro:

1.: "apt-get install linux-lowlatency jack"
2.: Add your user to the group "audio"
3.: Reboot into lowlatency kernel, tell Bitwig to use Jack and voila: Done. Low latency, great performance environment is before you.

Or, of course, install one of the distros tweaked for media production tasks, like Ubuntu Studio and have all this set up from the start.

I mean, seriously, this above is *one* Google search away...

Lump · Post by **Lump** » Thu May 21, 2015 8:32 pm

you're not a Linux user

Wrong. I'm a very experienced old linux user, young padawan.

Here's how you do it if you use Ubuntu

Yes I know this, got this advices a few pages before in this thread (studiodave and others). But it doesn't help so much. The performance is still poor in comparison to OSX on the same iMac 2010, i7, 16 GB Ram. Lot of glitches and crackles (with ALSA - BWS or Renoise) and thousands of xruns if I use JACK. It is no fun, it is a cramp.

Maybe I need another computer - maybe the iMac has too special specs for Linux .

I don't use RT kernel and I don't do any crazy tricks for LL audio. My system runs with low buffers (128) without any cracks using pure ALSA. I can easily run several Diva instances (great + multicore mode) simultaneously.

Tell me about your system specs! I want this, too!

psycha0s · Post by **psycha0s** » Thu May 21, 2015 9:03 pm

Lump wrote:
phantom-one wrote:I don't use RT kernel and I don't do any crazy tricks for LL audio. My system runs with low buffers (128) without any cracks using pure ALSA. I can easily run several Diva instances (great + multicore mode) simultaneously.
Tell me about your system specs! I want this, too!

I'm using Arch Linux with a stock kernel + ALSA (I've already written my hardware configuration in the previous message).
Several months ago I had a mediocre audio performance in the Bitwig Studio. During the investigation of this issue, I found that my CPU doesn't activate the TurboBoost. After adding a change of the CPU governor on boot my performance issues are gone:

Code: Select all

sudo cpupower frequency-set -g performance

PS: To be honest, some "heavy" presets could eat a half of Bitwig's DSP meter. But this is an exceptional case. In general, I don't have noticeable audio issues.

glokraw · Post by **glokraw** » Fri May 22, 2015 9:09 am

Lump wrote: Often crackles and glitches.
BTW I've purchased a license for Hive lately...

If macs let you get into the bios, or can shut down hardware componants by command,
shut down the wireless, network, webcam, anything not musical, double check
the irq sharing, power management, cpu throttles, and watch htop run for awhile,
to make sure hal and some daemonic utilities aren't on the loose.

My ancient bare naked linux audio setup has a /usr/bin of a whopping 254 meg,
my practicing theory, is that diverse freeware coders don't always have the time
or inclination to adequately test their works in conjunction with one another,
so the fewer the apps, the fewer the bugs, fewer strange noises in the night.

Hive is truly the bees knees of vsts. Another ace for u-he! (

)

hihu · Post by **hihu** » Fri May 22, 2015 11:38 am

Getting x runs does not mean it's not working well. I also though when using Jack the first time that I have to avoid all x runs.

But better go for audible errors.
When you don't hear crackling sound everything is okay.

My Linux system is (subjective) also not on par with the windows installation on the same computer. But I think this is just a driver issue because my sound card is very old (PCI) and I have to emulate it in an PCI express port. So maybe this fact leads to unwanted x runs and performance issues.

Maybe some day I will get an external USB audio interface. Internal cards are not the trend anymore.

Lump · Post by **Lump** » Fri May 22, 2015 1:25 pm

If macs let you get into the bios

Apple Computers never had something like a BIOS

No way for this PC tweaks

Seemed so, if I want a good running Linux computer I have to bite in the sour apple and buy a PC.

I think about a used HP Elitebook 8560w with an Intel i7-2820QM, a better chance then with my old iMac: https://beradrian.wordpress.com/2012/06 ... -on-8560w/

But I think this is just a driver issue because my sound card is very old (PCI) and I have to emulate it in an PCI express port

Hmm, I always thought (and made the experience), a traditional tower PC with an old envy24 PCI should work fine for Linux audio? I mean, this was the root of ALSA (the envy24 driver) at the end of the last century. Is pci support on Linux obsolete now?

psycha0s · Post by **psycha0s** » Fri May 22, 2015 1:34 pm

Lump wrote:Hmm, I always thought (and made the experience), a traditional tower PC with an old envy24 PCI should work fine for Linux audio? I mean, this was the root of ALSA (the envy24 driver) at the end of the last century. Is pci support on Linux obsolete now?

Yes it should, the PCI bus isn't obsolete (actually the PCI-E is used nowadays). I'm using ASUS Xonar Essence II, this is a plain PCI-E card. Maybe it's not intended for audio production but it sounds great and works well in Linux.

hihu · Post by **hihu** » Fri May 22, 2015 1:40 pm

Yeah it's the envy24. But from the m audio audiophile 192. It's another chipset where the envy mixer does not work.

The problem is not that Linux does not support PCI but the new mainboards hardly ever support it. Those newer mainboards only have PCI express and normal PCI slots are connected via kind of bridge.

That does not matter in case of usb PCI cards or FireWire cards and also Lan cards. But when latency matters it does not work that good.

Best value of latency is 256ms on my system. And that's not that low. But for me it's okay.
But you are right. I also like to have all thing internal in my tower instead of hundreds of cables on my desk.
But nowadays everything is connected external. From hard disk, sound-card t video editing cards.

psycha0s · Post by **psycha0s** » Fri May 22, 2015 1:51 pm

Actually, my audio card has PCI-E->PCI bridge onboard. I think that all PCI-E cards have such chip.
Are you sure you have 256 ms of latency? Perhaps you mean buffer size in audio frames (sample count * channel count)?

hihu · Post by **hihu** » Fri May 22, 2015 2:01 pm

No my Card is PCI and I have a PCI slot but this slot is bridged to an internal PCI express lane. For compatibility issues. Because newer Intel chipsets don't support PCI anymore.

Yeah I mean buffer size. Didn't measure the latency exactly. But with same buffer size on Linux and windows I can load more instances of hive and satin in windows at the same time. That's what I found out.

u-he

KVR Audio

Linux public beta (4408)