Question Red Power Reporting Deviation Minimum In-Game, Should I Be Worried?

[The Guy From The Thing]

hi, I've read up on the Power Reporting Deviation reading in HWInfo before, and have mostly ignored it when the value drops into the red because it always does that while idle. in game, it usually hovers between the 92-120% range, but yesterday, after getting off one of my games, I saw the minimum PRD was 82%, which is in the red

this didn't seem to happen when I closed the game, or when I opened it. additionally, the values never go red while I'm in a discord call on idle, so it wouldn't have done this just because the load was removed since I was calling a friend at the time. this seems to have happened while playing the game

it does say that the value only matters under 'full CPU load at stock settings', so that would means games don't report properly either, right? I ran Cinebench multi-core a couple of times, and the PRD is pretty much always 94-96%; currently, it's at 98-100% while doing a 30 minute benchmark. I also know CPU-Z stress test/benchmark isn't the greatest but I did that as well, and it reported roughly the same 94-96% PRD reading.

my CPU is a 5800x and my motherboard is an Asus B550-F; I also have Eco Mode 65w enabled, so that limits the PPT and whatnot, so I'm not sure if that completely negates the Power Reporting Deviation reading, since I'm not technically using stock settings anymore. should I be concerned? and does the red value under load indicate that the CPU or Motherboard is dying?

 

[drea.drechsler]

hi, I've read up on the Power Reporting Deviation reading in HWInfo before, and have mostly ignored it when the value drops into the red because it always does that while idle. in game, it usually hovers between the 92-120% range, but yesterday, after getting off one of my games, I saw the minimum PRD was 82%, which is in the red

this didn't seem to happen when I closed the game, or when I opened it. additionally, the values never go red while I'm in a discord call on idle, so it wouldn't have done this just because the load was removed since I was calling a friend at the time. this seems to have happened while playing the game

it does say that the value only matters under 'full CPU load at stock settings', so that would means games don't report properly either, right? I ran Cinebench multi-core a couple of times, and the PRD is pretty much always 94-96%; currently, it's at 98-100% while doing a 30 minute benchmark. I also know CPU-Z stress test/benchmark isn't the greatest but I did that as well, and it reported roughly the same 94-96% PRD reading.

my CPU is a 5800x and my motherboard is an Asus B550-F; I also have Eco Mode 65w enabled, so that limits the PPT and whatnot, so I'm not sure if that completely negates the Power Reporting Deviation reading, since I'm not technically using stock settings anymore. should I be concerned? and does the red value under load indicate that the CPU or Motherboard is dying?

I'm pretty sure using EcoMode would indeed make the PRD reading invalid.

From what I gathered reading the explanation and Martin's posts on his HWInfo forum, it's pretty much only an estimate of what the deviation might be. It sounds kind of like it looks at the power draw being reported when it's at a proper full load he's modeled (Cinebench, apparently) and then compares to the power draw a similar processor set up in full stock configuration (on an instrumented motherboard would be ideal but I can't say for certain) should be consuming in that same workload...the difference is the deviation.

So if you use any load other than the recommended (Cinebench) any comparisons aren't going to be valid; if you set it up different (ecomodes, custom PBO settings, undervolt, overvolt, overclock, underclock) it will also not be a valid comparison.

That may be oversimplified or otherwise inaccurate, but it's the best I've understood it.

 

[The Guy From The Thing]

I'm pretty sure using EcoMode would indeed make the PRD reading invalid.

From what I gathered reading the explanation and Martin's posts on his HWInfo forum, it's pretty much only an estimate of what the deviation might be. It sounds kind of like it looks at the power draw being reported when it's at a proper full load he's modeled (Cinebench, apparently) and then compares to the power draw a similar processor set up in full stock configuration should be consuming in that same workload...the difference is the deviation.

So if you use any load other than the recommended (Cinebench) any comparisons aren't going to be valid; if you set it up different (ecomodes, custom PBO settings, undervolt, overvolt, overclock, underclock) it will also not be a valid comparison.

That may be oversimplified or otherwise inaccurate, but it's the best I've understood it.

I see, thank you. that would make sense, that's kind of how I've understood it as well, I just struggle to wrap my head around it