hx850i shows 965W with a 3080ti

[jasonc_01]

So I just installed a 3080ti ftw3 hybrid last night, I'm running an hx850i and through icue I'm seeing as high as 956w output. No ocp or shutdowns however I am starting to look at getting a 1200w. Not sure how accurate icue is, so I've also reached out to jonnyguru  as well.

 

Just some food for thought. 

Edited July 25, 2021 by GDRRiley
Changed title

[MultiGamerClub]

6 hours ago, jasonc_01 said:

So I just installed a 3080ti ftw3 hybrid last night, I'm running an hx850i and through icue I'm seeing as high as 956w output. No ocp or shutdowns however I am starting to look at getting a 1200w. Not sure how accurate icue is, so I've also reached out to jonnyguru  as well.

 

Just some food for thought. 

Never got mine to power on, it's just laying on my desk now until my rm1000x arrives.

 

Wouldn't trust icue too much, not sure I 100% trust hwinfo some times too, but it gives an average of what your using so..

 

956w over the 850 you have? Wouldnt it shut off before going above what it's rated for? I know 3000 series cards can spike or peak but that doesn't sound right to me.

[SavageNeo]

47 minutes ago, MultiGamerClub said:

Wouldnt it shut off before going above what it's rated for?

no sicne it is such a short spike.

 

also the thing that shuts the psu when t draws too much power is OPP or OCP.

[jasonc_01]

5 hours ago, MultiGamerClub said:

Never got mine to power on, it's just laying on my desk now until my rm1000x arrives.

 

Wouldn't trust icue too much, not sure I 100% trust hwinfo some times too, but it gives an average of what your using so..

 

956w over the 850 you have? Wouldnt it shut off before going above what it's rated for? I know 3000 series cards can spike or peak but that doesn't sound right to me.

This run peaked at 1008 watts.

[Jeppes]

17 minutes ago, jasonc_01 said:

This run peaked at 1008 watts.

Maybe if you have 2x 3080ti.

[jasonc_01]

1 hour ago, Jeppes said:

Maybe if you have 2x 3080ti.

It's a single 3080ti and that was last night. I'm not sure however how accurate icue smart link is, but it should be accurate. 

[Jeppes]

3 minutes ago, jasonc_01 said:

It's a single 3080ti and that was last night. I'm not sure however how accurate icue smart link is, but it should be accurate. 

If I had to take a guess, that 1008W is there to show max value. What your picture shows is something like 600W.

[Guest]

11 hours ago, MultiGamerClub said:

Never got mine to power on, it's just laying on my desk now until my rm1000x arrives.

 

Wouldn't trust icue too much, not sure I 100% trust hwinfo some times too, but it gives an average of what your using so..

 

956w over the 850 you have? Wouldnt it shut off before going above what it's rated for? I know 3000 series cards can spike or peak but that doesn't sound right to me.

He has an HX850i.  The "i" means it has a serial interface that communicates real time data from the PSU to the motherboard.  So while I wouldn't normally trust software because in 99% of cases a PSU does NOT have any communication with the motherboard, a PSU that is supported by iCUE is quite different.

 

[GDRRiley]

==thread split==

do not take over someone else thread

[--SID--]

6 hours ago, jasonc_01 said:

This run peaked at 1008 watts.

No. The y-axis has a maximum of 1008w. There are 8 parts of ~125w on the y-axis. Maximum load is on top of part 5 what means ~625w.

[jasonc_01]

6 minutes ago, --SID-- said:

No. The y-axis has a maximum of 1008w. There are 8 parts of ~125w on the y-axis. Maximum load is on top of part 5 what means ~625w.

You are incorrect, 1008w was the wattage peak not the y axis scale. I have other screens shots at 992w and 984w, they are peak numbers.

[--SID--]

2 minutes ago, jasonc_01 said:

I have other screens shots at 992w and 984w, they are peak numbers

Show them

[jasonc_01]

10 minutes ago, --SID-- said:

Show them

 

[--SID--]

Where does the white chart hit the 984w, 992w and 1008w line? All 3 charts hit not higher than about 600w

[seon123]

13 minutes ago, jasonc_01 said:

 

-snip-

 

 

It seems like you have issues reading graphs. You're confusing the maximum value of the y-axis with the maximum reading. Here's a resource to help you

https://www.wikihow.com/Read-Graphs

[Lurick]

1 hour ago, jasonc_01 said:

-snip-

Not sure if confused but it literally says the current output of the PSU with the yellow tab, show the yellow tab saying 1008w.

[Guest]

56 minutes ago, --SID-- said:

Where does the white chart hit the 984w, 992w and 1008w line? All 3 charts hit not higher than about 600w

The scale changes as the load changes.  If the power out increases to 992W, for example, the scale of the graph increases to 992W.  

 

The best way to "capture" this is to use the logging function.

 

But again.... we're taking OP's thread WAAAAAAAAAAAAAAAAAAAAAAY off topic.  A mod really needs to come in here and split this off into a separate thread.

 

[--SID--]

2 minutes ago, jonnyGURU said:

The scale changes as the load changes.  If the power out increases to 992W, for example, the scale of the graph increases to 992W.  

 

The best way to "capture" this is to use the logging function.

 

But again.... we're taking OP's thread WAAAAAAAAAAAAAAAAAAAAAAY off topic.  A mod really needs to come in here and split this off into a separate thread.

 

It is split

2 hours ago, GDRRiley said:

==thread split==

do not take over someone else thread

 

[GDRRiley]

==fixed title==

[akio123008]

 

1 hour ago, seon123 said:

It seems like you have issues reading graphs.

I don't think the OP is to blaim for misreading graphs here:

 

The "max 1008W" is really confusing; the fact that it says "max" implies that it's the peak measured value, but its placement nex to the graph implies that it's the graph's max value.

 

But no sensible graph would use such a weird number as its maximum, then it'd be something like 1000 or 1100.

 

Also the word "max" is never used on the axis of a graph, normally it'd just say "1008" next to that line, and "0" at the bottom. (and there'd be a label like "power in watts" or whatever).

 

Then, to make things more confusing:

1 hour ago, jonnyGURU said:

The scale changes as the load changes.  If the power out increases to 992W, for example, the scale of the graph increases to 992W.  

So he did hit 1008W? How would it otherwise have rescaled to 1008? But why is that not visible in the graph? Is it too long ago? And why would you rescale to a stupid number like that?

 

If I could give the developer of that software some tips:

 

- rescale the graph to a nice round number, for example the nearest multiple of 100 or so. If the consumption goes to 990, rescale to 1000. For 1023, rescale to 1100. This would  distinguish it from a measured peak value (at least in most cases)

 

- remove the word "max" from there, it has no business being in on the scale of an axis and makes things more confusing. The number being at the top of the graph says enough.

 

 

 

 

 

 

[Guest]

30 minutes ago, akio123008 said:

 

I don't think the OP is to blaim for misreading graphs here:

 

The "max 1008W" is really confusing; the fact that it says "max" implies that it's the peak measured value, but its placement nex to the graph implies that it's the graph's max value.

 

But no sensible graph would use such a weird number as its maximum, then it'd be something like 1000 or 1100.

 

Also the word "max" is never used on the axis of a graph, normally it'd just say "1008" next to that line, and "0" at the bottom. (and there'd be a label like "power in watts" or whatever).

 

Then, to make things more confusing:

So he did hit 1008W? How would it otherwise have rescaled to 1008? But why is that not visible in the graph? Is it too long ago? And why would you rescale to a stupid number like that?

 

If I could give the developer of that software some tips:

 

- rescale the graph to a nice round number, for example the nearest multiple of 100 or so. If the consumption goes to 990, rescale to 1000. For 1023, rescale to 1100. This would  distinguish it from a measured peak value (at least in most cases)

 

- remove the word "max" from there, it has no business being in on the scale of an axis and makes things more confusing. The number being at the top of the graph says enough.

 

 

 

 

 

 

It's labeled max because it's the max.  It IS the peak measured value. When the PSU hits that higher number, the graph resizes.  So if the load after "1008W" is 504W", then you'll only see the line go half way up the grid.

 

The reason you don't see the 1008W on the graph is it's before the screenshot was taken.  After five minutes, the graph will reset to whatever the highest load was within the last five minute time period.  That's why I suggested using logging.  It will capture the 1008W and everything before and after it.  The graph is just used for "real time".

 

 

[electropical]

2 minutes ago, jonnyGURU said:

It's labeled max because it's the max.  It IS the peak measured value. When the PSU hits that higher number, the graph resizes.  So if the load after "1008W" is 504W", then you'll only see the line go half way up the grid.

 

The reason you don't see the 1008W on the graph is it's before the screenshot was taken.  After five minutes, the graph will reset to whatever the highest load was within the last five minute time period.  That's why I suggested using logging.  It will capture the 1008W and everything before and after it.  The graph is just used for "real time".

 

 

Can the "i" PSUs catch transient spikes like an oscilloscope would do?

[Guest]

6 hours ago, electropical said:

Can the "i" PSUs catch transient spikes like an oscilloscope would do?

It's "luck of the draw".  The sample rate is still slower than an oscilloscope, but it's fast enough to where it might actually catch some spikes if they align correctly and that's what the OP is seeing.

 

[Juular]

7 hours ago, jonnyGURU said:

It IS the peak measured value.

Talk about bad UI/UX. That's not what i expect to be the top value at the Y axis of the graph. But sadly that's not the only usability issue with iCUE. Especially the latest redesigned version, it's just horrible all around ...

PS: At least i don't need to shut my Corsair Harpoon manually anymore if the PC is switched off / sleeping so it doesn't shine the rainbow-puke RGB on me, yay! It took Corsair code-monkeys 3 years to fix that.

17 minutes ago, jonnyGURU said:

but it's fast enough to where it might actually catch some spikes if they align correctly and that's what the OP is seeing.

So it catches the spikes and resizes the scale accordingly but doesn't actually render them on the graph ? Why resize the scale then ?

[Guest]

12 minutes ago, Juular said:

So it catches the spikes and resizes the scale accordingly but doesn't actually render them on the graph ? Why resize the scale then ?

It does.  OP captured the screenshot afterwards.

 

So, in iCUE, you can set the size of the "window".  If within the "window" a spike is caught, the graph expands to that spike size.  What it looks like OP did was run iCUE in the background while doing a benchmark.  The spike occured, the graph resized.  Then he exited the benchmark, the spike was gone, but the graph was still resized to the measured spike.