Overclock.net - An Overclocking Community - Reply to Topic
Thread: [OFFICIAL] HWiNFO/32/64 Thread Reply to Thread
Title:
Message:

Register Now

In order to be able to post messages on the Overclock.net - An Overclocking Community forums, you must first register.
Please enter your desired user name, your email address and other required details in the form below.
User Name:
If you do not want to register, fill this field only and the name will be used as user name for your post.
Password
Please enter a password for your user account. Note that passwords are case-sensitive.
Password:
Confirm Password:
Email Address
Please enter a valid email address for yourself.
Email Address:

Log-in


  Additional Options
Miscellaneous Options

  Topic Review (Newest First)
02-16-2020 06:11 AM
sdmf74 Is there a guide or instructions on how to use the multiple column feature? I just figured out how to open multiple colums finally lol but Im wondering if there are more options
02-07-2020 08:45 AM
Timur Born The short 5 second burst (and corresponding noise) was based on the Apple iCloud service running (as part of the Photo service). Hwinfo still does not refresh as fast as set in preferences unless I mess with the Windows timer.
02-06-2020 01:34 AM
Timur Born I disabled monitoring off everything except the CPU entries, same result. HWiNFO refreshes slower than what I set in preferences except for a short burst every 5 seconds. Once I change the Windows timer resolution via TimerTool it works properly. The tighter the timer resolution (lower value in ms) the faster HWiNFO's refresh rate.
02-05-2020 01:23 PM
Mumak HWiNFO uses HPET only in some cases to measure the BCLK, but in your case (on Intel Skylake and later CPUs) it's not used at all.
02-05-2020 12:34 PM
Timur Born Disabling all C-states does not increase HWinfo's polling rate ("Disable Idle" in power profile, which also disabled C1 non E). On the other hand, changing the Windows timer resolution even works when the power saver profile is used at a measly x8 idle CPU multiplier.
02-05-2020 12:18 PM
Timur Born I use to set SMART to 1000 cycles, the 5 seconds also happen when Hwinfo is not running. So something else in my system is active every 5 seconds.

The "Profiling Time" column reads 0 for most entries, some headlines show higher values even when every sub-entry is 0. CPU xyz: Enhanced is one such headline that shows 1 ms, but all sub-entries show 0. Likely a rounding thing.

Sometimes the GB sensor IR35201 increases to 149 ms, with VR Loop 1 at 117 ms and IIN and IOUT in the two digit range. This happens about every 8 seconds. Changing the Windows timer resolution to 0.5 ms increases the time from 8 seconds to about 38 seconds. I think this is connected to polling cycle, C-states and power profiles. Dimm0 temp occasionally acts up, 1-3 seem to stay at 0, likely based on the same source. These things only every few seconds, though, while HWinfo keeps polling too slow.

GPU headline is 7-10 ms, with GPU load at 3-5 ms and everything else at 0. The Intel PCH and two other GB sensors show 1 in the headline and 0 in the sub-entries. The NVidia driver currently is set to maximum performance (no downclocking for power-saving), so it will not get any faster than that.

And changing the Windows timer resolution does not change the normal "Profiling Time" measurements, but HWiNFO still increases its polling towards what was really set in its preferences.

I also tried HPET on/off in Hwinfo's settings, but it made no difference. Does this really specifically use the HPET or does it use QPC (in my case invariant TRC at 10 MHz)? What does HWinfo use when HPET is disabled?
02-05-2020 11:45 AM
Mumak Isn't that 5 second noise related to disk SMART polling in case you have a different rate set for it?
If the real polling rate of HWiNFO is significantly longer than the rate set, enable the "Profiling Time" column and check whether some sensor isn't taking too long to read. If there's a sensor taking excessive time (100s of ms), disabling it should improve the rate.
02-05-2020 10:52 AM
Timur Born
Quote: Originally Posted by Mumak View Post
Efficiency is just a simple result of: PowerOut / PowerIn * 100 [%].
So if the formula for calculating one of the power values is different, Efficiency will be as well.
Well, I would have hoped that two of these three variables were known/measured. If only one is measured, one is derived from an algorithm and the third is then calculated then the measurement is not fully useful.

Quote:
This is an interesting observation I wasn't aware of. Probably just another proof of the 'quality' of Corsair software...
To be fair, the VRM noise is a result of polling the USB interrupt endpoint in a burst. It's mostly based on C-state / current changes in the VRM and as such only one of many noises.

Turns out that HWiNFO's USB polling also creates its own noise signature, but higher in frequency and more easily masked by other noise. I really wish modern mainboards were less noisy, especially because that noise carries over ground lines to (professional) audio equipment via protective ground (PG) loop.

Curiously my current noise vs. PSU vs. CPU current measurement also finally pointed me to what I always found odd/wrong about HWiNFO's polling rate. In short: I always noticed that polling rates below 1000 ms were longer than what was set in preferences (minimum 100 ms). During USB polling noise tests I noticed that I get one high pitched noise every 5 seconds, similar in tone to HFiNFO's USB polling noise. Then I noticed that whenever this occurs HWiNFO flickers quicker with faster polling rates, which I always kind of noticed, but never looked closer.

While I do not know what is especially happening on my system every 5 seconds I did notice that I can increase HWiNFO's *real* polling rate by decreasing the Windows timer resolution via TimerTool. The lower I set it (minimum 0.5 ms) the faster the polling and the less the 5s interval matters.

Things are a bit more complicated, though, because I noticed that setting 5 ms via TimerTool results in a faster HWiNFO polling rate than just measuring 5 ms being set by Firefox when a Youtube video is started (used to be 1 ms in the past, but they "fixed" it after my report to keep laptops from draining their batteries).

https://github.com/tebjan/TimerTool/...ster/README.md

Quote:
There were guys (i.e. the author of SIV) who spent a great amount of time with Corsair stuff trying to figure out how it works and even giving Corsair suggestions how to improve things. But this wasn't neither accepted nor appreciated at all.
Like HWiNFO I find SIV to be a great tool. I just used it to determine that the Corsair USB 2 endpoint is using interrupt mode.

Did you try to reach Johnny Guru directly (if he still is with Corsair's PSU division)? Or maybe reach out to Channel Well Technology, the real producer of least the HX750i (that I am currently testing)?
02-05-2020 07:35 AM
Mumak
Quote: Originally Posted by Timur Born View Post
What about the "efficiency" reading in HWiNFO, is that based on a measurement or just likely some formula based on the current load state? In iCUE Power In x Efficiency = Power OUT (or maybe Power IN / Efficiency = Power OUT). Compared to my wall measurements (+-1% +1 W average tolerance) I'd say that the "(Sum)" is closer at least.
Efficiency is just a simple result of: PowerOut / PowerIn * 100 [%].
So if the formula for calculating one of the power values is different, Efficiency will be as well.

Quote: Originally Posted by Timur Born View Post
There is another drawback: Every time the Corsair service polls the PSU my mainboard's (GB Aorus Master) VRM section creates a specific noise pattern. This does not happen when HWiNFO polls the PSU, even at faster rates than 1/s.
This is an interesting observation I wasn't aware of. Probably just another proof of the 'quality' of Corsair software...

There were guys (i.e. the author of SIV) who spent a great amount of time with Corsair stuff trying to figure out how it works and even giving Corsair suggestions how to improve things. But this wasn't neither accepted nor appreciated at all.
02-05-2020 07:09 AM
Timur Born I understand, thanks for the information. What about the "efficiency" reading in HWiNFO, is that based on a measurement or just likely some formula based on the current load state? In iCUE Power In x Efficiency = Power OUT (or maybe Power IN / Efficiency = Power OUT). Compared to my wall measurements (+-1% +1 W average tolerance) I'd say that the "(Sum)" is closer at least.

As mentioned before: While HWinfo can poll the PSU on its own Corsair's iCUE needs the a special Corsair service to run. Said service does not cause high CPU load (below 1% average), but it still measures as one of the highest average CPU loads of all "idle" services on my system.

There is another drawback: Every time the Corsair service polls the PSU my mainboard's (GB Aorus Master) VRM section creates a specific noise pattern. This does not happen when HWiNFO polls the PSU, even at faster rates than 1/s.
This thread has more than 10 replies. Click here to review the whole thread.

Posting Rules  
You may post new threads
You may post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off