Hi everyone,
I’m using the Waves LV1 Classic console and I’d like to ask: What is the recommended maximum server load percentage to maintain 100% audio stability without any risk of dropouts, especially when running at higher sample rates?
I’m currently running at 96 kHz, with about 6 plugins/effects plus 5 instances of F6, and I’m seeing around 64% server load.
Everything is stable so far, but I’ve noticed that sometimes the letter “A” turns red, and the yellow peak line goes above the green bar, which makes me a bit concerned about potential instability or performance limits.
I’d like to know what the safe operating threshold is to avoid any issues during live use.
Any tips, recommendations, or shared experiences would be greatly appreciated!
I’d also love to understand this better, especially the server letter turning red.
I’m running at 48kHz with a pile of plugins[1] and haven’t noticed any problems in the DSP between 60-80% load, but I’m also not sure what it does if it misses a deadline? Silence for a few samples? Unprocessed audio? I haven’t particularly tried, but I haven’t managed to push the load beyond 80%.
The lowest I’ve ever gotten the load was when I was only using the integrated inputs, so I wonder if the SoundGrid network also consumes load? I wasn’t expecting it, but probably there’s just something I don’t understand going on.
Also, I’m pretty sure that bypassing a plugin or disabling a rack (the On/Off button at the top) doesn’t reduce load (seems you have to actually disable the plugin), but would love to have that confirmed.
Generally I have 6-8 vocal channels with matching chains (5-6 plugins each), couple of guitars/keys/drums with simple processing, and then a few layers of bus processing and FX channels. ↩︎
You should strive to stay on max %85 avg. The yellow line you mentioned is peak usage - some plugins, especially IR Reverbs and RTA’s are more ‘peaky’. You can experiment with changing the server network buffer or using different plugins. However as long as these peaks are minor they will usually not cause dropouts. I’d say %64 average is a good place to be.
Yesterday, i was playing with mix via “virtual soundcheck”, and i have experienced few screen freezes. Audio keep passing thru, but plugin screen just tears for 3 sec during this time A letter (server) turned RED. Yet load was ±65% but peaks are realy High..
about peaks - as you said problem might be 2 IR Verbs + F6 with RTA.. i will try to turn that to standart f6 (without RTA)
and also i will play with network buffer, i think right now im set in between (17ms i think.. 3rd fastest option if i remeber correctly)
The Server DSP meter turns red - this indicates a buffer underrun, which may very well be an audible dropout. In that case, look into your plugin selection, server network buffer, and mixer engine mode (depending on your application, perhaps DSP optimized will work better for you). Additional servers can also help take the load off.
The server does have some reserved capacity for handling network transport, which is not configurable by the end user. Obviously, it is totally optimized for plugin processing, and that’s not something you should worry about IMO.
Regarding the drop, if a server in group A (which is in charge of running the mixer + processing plugins) disconnects, no audio. If a redundant server is assigned, there will be a switch - it will take a few seconds. In racks assigned to servers in group B/C/D (only doing plugin processing), if the server disconnects, the rack will be bypassed until the server is back or a redundant server is there to pick up.
However, often in cases where a bad cable is used, network errors might not result in a complete disconnection, causing audible audio issues. Gotta use proper cables from a known brand (:
You are correct about bypassing vs. disabling. Only disabling plugins will actually take the load off the server.
Great to know! I can’t say I’ve noticed it, but since I guess it’s only happening when I have a lot of channels it’s easier for a short drop to be lost in the mix. I’ll have to do some stress tests and try to record the outputs
This is also really helpful, though hopefully I’m a little way away from needing additional servers. Might look into renting some for certain gigs, I guess.
So I didn’t go as far as recording direct outputs to see if there are any drops, but I’ve just been running at about 75% reported load with the bar constantly red and didn’t notice any audio glitching at all. I wonder if there’s any way to narrow down which plugin is facing the buffer underrun? Though I guess it could be a different one for each packet, so that’s probably not helpful.
I did go through disabling plugins to find which ones contributed most to the load, and I found that Scheps Parallel Particles on my main bus was worth about 5% (which is fine, I feel like I’m cheating using that one live ), but strangely the instance on my drum bus[1] didn’t.
More disappointingly, the GEQ Modern also contributed 5%. The GEQ Classic doesn’t have the same load, so I’ll probably switch over, but I like having this on my main output for room tuning, so it’s a shame it’s so intensive. Maybe one day I’ll get to work with PA that has its own EQ…
For people who haven’t tried it, this made a huge improvement to the load. Didn’t test how much it hurts the live latency, but since I’m usually running in-ears off the same desk it’s probably not a good choice for me (didn’t seem to affect any of the latency displays, unlike changing the latency comp. or network buffer settings). If you’ve got a separate desk for monitors, it’s probably worth considering. I think I’d probably pull out an old desk to do monitors before adding more DSP to my LV1.
I mostly get electric drums these days, and this is one of the best plugins for spicing them up! ↩︎
At this point, everything was massively distorted, but it sounded okay up until the letter A went red. The peak bar being red, or the meter bar being red, or even the load being slightly over 100 didn’t seem to inevitably cause buffer underruns, but I wouldn’t want to be running that close.
Still, I’m less worried about running at 75% now, so it was a worthwhile experiment. (I’m also a little worried about GEQ Modern - turning the RTA off doesn’t make any difference, so I guess there’s something about it that doesn’t like LV1?)
), but strangely the instance on my drum bus