• SONAR
  • Strange issue regarding soft synth at low latency. (p.2)
2017/01/31 16:35:17
Sanderxpander
If problems go away at 64 samples I would just stick with that, should be well beneath noticeable latency. Laptops can be finicky with extremely low latency. I have similar specs to yours, just a faster CPU and an RME card and my laptop's Touchpad creates relatively high DPC latency. It could be the midi/USB driver in your case. It may even be USB 1, forcing the USB 3 port into legacy mode or whatever.

I'd be happy you run stable at 64 samples, that's better than many people do and doesn't give you any problems for tracking or live performance.
2017/01/31 16:37:25
Sanderxpander
brundlefly
gprokap
PCIe SSD, what laptop in 2017 still has a 5400 RPM drive?
 
CPU usage is BARELY above 1% -  disk also shows no activity.



Sorry; didn't read carefully enough. I'm not sure what else might be going on, then, but would still check DPC latency. I can play handfuls of sustained chords on TruePianos and Dim Pro at 32 - or even 16 - samples here so I know it's not an inherent limitation of SONAR or the synths you're using.
 
 

16 samples, really? Wow, what interface is that?
2017/01/31 16:42:19
brundlefly
It's in my signature: MOTU 2408/PCIe-424.
2017/01/31 17:09:06
abacab
gprokap
Using an Asus Zenbook, Sonar plat, Focusrite Clarret Thunderbolt interface.  Windows 10, 16 GB RAM, PCIe SSD drive, 4 core 2.6 GHZ I7 with Hyper thrreading.
 
I can play a guitar/bass through an amp sim no issue at all with a 32 sample buffer.   Mostly TH3 and some other cheapware ones.
 
when running a synth through MIDI (USB controller, not using the MIDI in on the Focusrite if the polyphony of the synth goes above 8ish notes or so it kinda freaks out, snap crackle and poping noises.   Not using greedy synth either, DM pro, True Pianos, Korg legacy stuff, all low-power synths.
 
The CPU meters in SONAR barely show any CPU activity. No other apps on the PC, AV turned off makes no difference, same with turbning off blue Tooth, wireless, and pretty much every Windows Service that isn't needed.
 
All works fine at a 64 sample buffer, which is playable and I can live with this if this is how it is.
 
But what throws me if is live instruments work fine at the 32 sample buffer - midi ones don't.   Are there any optimizations I might have missed or settings in SONAR that would help? 




The answer is adjust the buffer up until you stop having problems.  Then that is where you should run them.  The breakups are occurring because your CPU is not keeping the audio buffer full at all times.  All computers are different, and it is not just a matter of the interface type.
2017/01/31 17:12:52
Sanderxpander
brundlefly
It's in my signature: MOTU 2408/PCIe-424.

Ah sorry, I'm usually on my phone on here and the mobile version doesn't show signatures. That's pretty amazing performance, what's your total roundtrip with that? Lowest I've managed on my FireFace UCX (without its excellent direct monitoring) is 2.9ms, which I already consider very good (especially since softsynth latency is even less than that).
2017/01/31 17:59:33
abacab
This video has the very best explanation I have seen on real-time audio performance, and why it is not CPU dependent.  At 8:15 in the video you will see a great graphic simulated animation of the buffer.
 
CPU Performance vs. Real-Time Performance in Digital Audio Workstations (DAW)
 

2017/01/31 18:02:37
abacab
This is a good quick overview of the trade-offs between buffer size and plugins.
 
A Lesson on Buffer Size - Berklee
 

2017/01/31 18:29:22
brundlefly
Sanderxpander
That's pretty amazing performance, what's your total roundtrip with that? Lowest I've managed on my FireFace UCX (without its excellent direct monitoring) is 2.9ms



121 samples  = 2.5ms at 48kHz. That's including 44 samples of A/D/A conversion latency not reported to SONAR by the driver. If you're going by what's reported in SONAR,  it's likely missing some 'hidden' hardware latency.
 
You can measure the actual RTL with this free utility:
 
     http://centrance.com/downloads/ltu/
 
 
 
 
2017/01/31 18:41:51
Sanderxpander
The 2.9 is a real world figure too. That's at 44.1KHz btw. I do know RME uses really low latency converters though, 14 samples for AD and only 7 for AD. USB as opposed to PCIe makes it slower though.
2017/01/31 19:00:38
brundlefly
Ah, I wasn't paying attention and thought that was the OP's RTL with the Focusrite. I know RME's performance is a cut above. Pretty amazing for USB.
© 2026 APG vNext Commercial Version 5.1

Use My Existing Forum Account

Use My Social Media Account