High Disk Usage caused by Gadget Reporting

sorbic-acid

New Member
I happened to have task manager open this evening and noticed that I was seeing a see-saw pattern of writes on my primary disk. Resource monitor indicates it's the registry being written to. I was able to isolate this down to HWiNFO64's Gadget reporting.

I use HWiNFO in conjunction with Rainmeter plugins, specifically the Modern Gadgets to report various things for my system (GPU clock, memory usage, etc). These two applications launch on login, so they are running all of the time.

I know that HWiNFO is inherently writing these values to the registry and then the aformentioned Rainmeter gadget is scraping them back out, but the disk usage doesn't line up with the amount of data that is actually being written. I am seeing 20+ megabytes written every two seconds whenever Gadget reporting is enabled.

To further complicate things, the aggressive disk activity isn't consistent. In addition to HWiNFO64 and Rainmeter running on startup, I use HardDiskSentinel which basically sits in the background and periodically scrapes SMART values for my drives, so I have some historical data pertaining to disk writes.

1742000300261.png

I would expect this chart to be flatter, but it is clearly skyrocketing at various times and I am pretty sure it's related to this.

Adjusting the polling period helps (it was previously set to 1000ms, I bumped to 2000ms), as does simply turning off some of the individual settings, but this just seems to stretch the writes over a longer period.

Am I mistakenly seeing this behavior for the first time or is there something amiss here?
 
Last edited:
20MB+ is really a lot. Do you maybe have Debug Mode enabled in HWiNFO? That would write a lot of data to disk.
 
Then it's clear, it's this feature. But the amount of data written shouldn't be that high. How many sensor values have you configured to be exported?
 
I've historically enabled them all and then just plucked the values out that I actually needed for the respective gadgets. I've done this for as long as I can remember.

For a test I disabled them all and whittled the list down to 39 values I actually care about. I also slowed down the polling interval (1000ms -> 2000ms) so it isn't checking the values so often.

I'm seeing "normal" usage now.

The only thing I can think of that is causing this behavior is that one of those now-disabled devices is intermittently reporting a value that is inordinately large and that value is getting written to the registry repeatedly when reporting is enabled.

Is there any sanitizing of the gadget values that ultimately get written to the registry?
 
I have verified this now and HWiNFO correctly writes only values enabled for reporting and there isn't any bug. Moreover, it's optimized so that if the value didn't change from previous one, it's not written again.
So the only way to reduce the writes is to limit the number of sensors for reporting only to those necessary.
 
Back
Top