Performance Issue: Audio Hiccups

(Jon Heal) #1

My patches tend not to be gigantic, but the one performance issue I experience regularly is what I call audio hiccups where the audio through the monitors periodically stops for a fraction of a second. It’s quite jarring, but interestingly, if I have recorded a segment that included these hiccups, the audio dropouts are not recorded—the sound is continuous—but the effect of the dropout can be heard, e.g. the repeating pattern of a delay changes at that point.

Anyway, I am on on older i7 running the latest Windows 10, but I am using the CPU’s integrated HD4000 graphics. If I were to install a separate graphics card, would that possibly improve my situation? My options are a little limited because I would need a low-profile card to fit in the case. Also, I am not thrilled with spending more that $50-60 for a card for this older computer.

Would something as meager as this be worth the effort?

(Tuborgman) #2

I’ve experienced the same behaviour on new 2018 mac book pro. Tried to increase latency and it seems that there are no more hiccups - but need more testing.

(Jon Heal) #3

Yes, increasing the latency helps the dropout/hiccup problem, but makes it problematic to keep things lined up with the external devices I am syncing with.

(Jim T) #4

graphics shouldn’t cause stuttering, they should just get slower frame rates. however it does sometimes seem to affect things. especially on macs, not sure about integrated graphics.

maybe try a dev build of v1 or the unofficial, unsupported experiments build from and drop the refresh rate. if that helps, it might be worth getting a better graphics card to help improve things.

I’ve got a 1060 and it’s absolutely fine.

(Jon Heal) #5

On Passmark, a $50 card (i.e. GX-710) is marginally better performing than the HD4000 built into the i7, but just barely. It looks like you need to spend closer to $100 to get marked improvement. But if the lack of video power does not lead to dropouts, then … I’m not sure it’s worth bothering with.

Is it dangerous to run Rack.exe at a higher priority in Task Manager?

(Jim T) #6

it’s fine to run at higher priority. there’s a module that does this, from impromptu I think.

(Paul Piko) #7

That’s Squinky Labs Thread Booster

(Jon Heal) #8

I tried raising Rack.exe’s priority to High in Task Manager. It did not impact this problem at all. I am curious as to what exactly is causing the dropouts. CPU utilization for Rack stays below 20%. I am using a PreSounus iOne audio interface set to 24-bit 44.1K and the lowest latency setting. Using ASIO or WASAPI seems to make no difference,

(Skrylar) #9

I would be curious to know if the problems persist under JACK. I have no data on JACK ASIO Router (windows) or JACK vs. Core Audio (mac), but a computer piled in xruns under ALSA experienced zero stutters under JACK and the block sizes were identical.

The GPU portions can/do steal compute, and Intel has made a few statements that they are not interested in optimizing OpenGL performance. When using the integrated on a newish i7, sysprof on linux (full system profiler) shows about 20% of CPU use across the system completely unaccounted for and another 10-15% are in visible parts of MESA (doing GL calls?)