Re:DPC Latency Checker: Worst culprits?
2012/04/18 10:59:20
(permalink)
Don't worry about it.
Deferred Procedure Calls are primarily hardware interrupt handlers, meaning they are tied to hardware devices (e.g. disk controllers, video adapters, network interfaces). Reducing DPC-based latency therefore can only be accomplished by disabling adapters or replacing them with more efficient cards. There aren't many adapters in there that you don't need for audio, so your options for improving efficiency are few.
Your 100us spikes may not look ideal, but 100us is a very brief interval (less than 5 samples at 44.1 KHz) and isn't likely to cause dropouts (if it does, it's not your only problem). The system is either fast enough or it's not - there is no advantage to being faster than necessary.
But if you really want to get those DPC spikes down, you may be able to. The first step is to determine which device is causing the largest spikes. That in itself is not easy. Most people determine it through trial and error, but even the most systematic methods are still indirect supposition. Once you've identified the culprit, you're faced with what to do about it - disable or replace. If it's a critical component (e.g. video adapter) you'll spend significant money just to find out if brand x offers lower latency than brand z. It's not information they put on the box.

All else is in doubt, so this is the truth I cling to.
My Stuff