Re: resource monitor add-on or some other way?


Nolan Darilek
 

I'm a bit hesitant to wade into this, but I too am a bit suspicious by the jump directly to HD issues.

Long shot, but you're not running Firefox or Thunderbird by chance, are you?

I've been having some odd, regular slowdowns recently. My PC is fairly high-power--I7, 8 cores, 32 GB of RAM, a few years old but not under-powered. Windows installation is only a few months old, but previously it ran Linux and barely broke a sweat. Under Windows, I got about a day or two of use before I often had to hard powercycle the machine. Task Manager didn't give me much that was reliable. Antimalware Service was often high on the performance list, but not reliably. I couldn't trace it to a specific action or change--my CPU fan would eventually go crazy, and my entire system would slowly go unresponsive.

Eventually I tracked it to Firefox. Killing Firefox always replaced the need for a reboot. Tried debugging Firefox further--disabling some extensions, using Nightly, etc. Nothing reliably fixed it. I'm also using a few layers of ad blocking, and while the blocking itself may have slowed things down a bit, I imagine the code that was blocked and didn't run probably cancelled out the extra effort the blockers made filtering it out. So it likely wasn't any page in particular, and none reliably caused this behavior either.

Now I'm running Brave, and the problems are mostly gone. I do, however, experience slowdowns when working in Thunderbird, particularly when opening a message. Yes, I've seen the Thunderbird speedup tips recently, but I don't feel like those helped, particularly since things worked fine under Linux.

Starting to wonder if there's some Gecko NVDA regression, or maybe a Gecko regression with the underlying a11y APIs. Haven't tried diagnosing it further, but as a developer that's kind of what this feels like. Wondering if you might be hitting this too. Are you running FF, and if so, does killing it get your performance back? I guess this could be non-a11y-related, or maybe an issue with crappy drivers, but if this regression were more mainstream then I can't imagine it wouldn't be high priority and fixed. That's why I suspect something in the a11y stack.

Join nvda@nvda.groups.io to automatically receive all group messages.