From: toby@tobiah.org   
      
   So I found a utility called 'RTL' that measures true latency by   
   actually timing the round trip from input to output through a cable.   
   Mind a little blown.  Buffer size is only tangentially related   
   to hard latency figures.  The Presonus, using a 256 sample buffer   
   scores around 12ms, same as the Scarlett with a 64 sample buffer!   
   The best figure I could get out of the Scarlett with any buffer   
   size (safe mode off) was about 10ms.  The Presonus can do 3.855ms   
   at 16 samples, although as I mentioned, the best stable buffer   
   size I can use with the Presonus is 128, but that yields around 7ms.   
      
   There are always so many layers to peel through with this stuff.   
      
      
      
      
   On 1/25/2026 4:25 PM, Tobiah wrote:   
   > I've never been able to get reliable low-latency audio under windows.   
   > I've spent many hours looking through lists of tweaks, like power profile   
   > settings and countless other magic incantations that never seemed to   
   > make a difference.  I've also never gotten LatencyMon to run for very   
   > long without the red flag.   
   >   
   > My largest need for the low latency is for playing sampled pianos   
   > and other instruments.  I've been using a PreSonus Studio 18|10,   
   > and the smallest buffer size I can use without clicks is 256 samples   
   > at 44100Hz.  Any smaller than that, and it mostly works, but I get   
   > little clicks periodically.  Now, some would say that 256 samples   
   > is small enough, but I can really tell the difference when playing   
   > a digital piano; I can feel the difference between 128 and 64 as   
   > well.  After that the change is imperceptible.   
   >   
   > So I came across a Scarlett 2i2 and tried it out and found that   
   > I could use a 64 sample buffer size (this is using Reaper) and it   
   > seems rock solid.  This got me to thinking maybe Focusrite does a better   
   > job with drivers, so I started shopping for a more capable   
   > Scarlett.   
   >   
   > Then I thought that the Presonus, having so many channels, is probably   
   > taking most of the USB bandwidth, and the lowly 2 in, 2 out Focusrite   
   > is may be enjoying an unfair advantage.   
   >   
   > Now, I'm on a ThinkPad at the moment, so I get that I may not be   
   > able to expect as much as from a well-built desktop.  I went through   
   > dozens of tweaks though, and all I got to show for it was the laptop   
   > fan running more often, and louder.   
   >   
   > Having spent so long Googling and performing low-latency tweaks, I   
   > didn't know where to start backing out the changes, so I went for a   
   > fresh Windows install, which was badly needed anyway.   
   >   
   > The Scarlett still dutifully handles a 64 sample buffer.  I'd   
   > spring for a 18i16 4th gen, but I still have my concern that the   
   > 2i2 may only be performing better because of the low number of   
   > channels.   
   >   
   > Another question:  if the large number of channels over usb 2.0   
   > is indeed a concern, would I do better with a USB-C interface?   
   > What if that cable also carries the monitor signal?   
   >   
   >   
   > Thanks   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)