Question about 'Add to Tray'

Hi there,

I've a minor problem and I'd like to ask if there's a solution. When I want a specific monitoring on tray, its default behavior is set 'Only show notifications' and thus, I've to re-set each one individually on 'Show icon and notifications'. How can I make them all being shown initially?

Seemingly at least, the only way is to enable 'Always show all icons and notifications on the taskbar', but I lose the hiding-feature generally.


Also, I don't know why, but I don't get solid GPU utilization, like the CPU's... On games I don't have problem, but on lighter tasks, like browsing with Hardware Acceleration On, it doesn't work. It's like the Hardware Polling Period ( or Scan Interval ) is set to 10 seconds while in the meantime remains at 0%... Do you think it's driver's fault ( AMD 12.6 )? Lately I face the same problem on Afterburner ( currently 2.2.3 ). To get the GPU utilization, the card must be working 'heavily'. On Process Explorer, though, GPU utilization works flawlessly, no matter if the card is being utilized 1% or 100%.

As you see, English isn't my language. I hope you got the main points... Thank you!
 
I'm sorry, but I don't have an answer for the 1st question, it's a Windows 'feature'. Maybe somebody else will answer this..

About GPU Utilization - I don't think this is a matter of polling period, because it's calculated by the GPU. But you can still try to change the HWiNFO sensor polling period (from default 2 secs) to see if that changes anything. Which GPU do you have ?
 
Yes, I tried to lower the number, but nothing works, unfortunately. I've the Gigabyte 7950 Windforce.

The GPU utilization should work as good as on Process Explorer, normally, right? The fact is that on Process Explorer it's working flawlessly, as I said, which means if I had a general problem, it shouldn't work on that program either. Mysterious...

Anyway, thank you!
 
Theoretically it should work similar to Process Explorer, but in reality there's no reason for that. As I already said, this usage is calculated by the GPU itself. Unfortunately AMD doesn't disclose anything about this - what exactly does it mean, what parameters are taken into account, nor how accurate it should be. They don't even say that the GPU is capable of such a feature ;)
 
Strange then, the fact that on HWiNFO and Afterburner doesn't work as well as on Process Explorer.

By the way, the program 'Core Temp' has a feature on "Windows 7 Taskbar" section which says: "Do not hide taskbar button when minimized" which seemingly does what I asked on the first issue... Anyway, thanks, again.
 
Sorry, there's probably some misunderstanding here.. I meant that the PRECISION of the GPU load is not as perfect. So I'm not sure which feature exactly are you comparing to Process Explorer.
 
Please, take a look... I drew the parts I'm talking about:

http://i.minus.com/i8B1Kq7LmFNax.png

Those zeros on HWiNFO do not stay 0% for ever... They work some time and the 'Max' will be set. As I said, if I open a game which utilizes heavily the GPU, all of them ( min/max/current ) will work fine, but for light tasks, they rarely work just for a random moment.

http://ie.microsoft.com/testdrive/Performance/FishBowl/

On this same test, if I increase the fishes to 2000, the GPU Utilization works as it should... This highlights that it doesn't work on light tasks. Maybe when the clocks are on idle mode ( 300Mhz ) HWiNFO cannot show the GPU utilization?

Meanwhile, all other monitorings for GPU ( Thermal Diode/Core Clock/Memory Clock etc ) working normally.
 
Thanks, I see now exactly what you mean.
First, I didn't know that Process Explorer was able to report GPU utilization numbers.
Second, I'm quite sure that Process Explorer reads this kind of information using some undocumented/hidden functions in the Direct3D core. Moreover it seems to me (also according to some discussions on Sysinternals forum) that these numbers are not always exact. I also think that if you would run a non-DirectX application utilizing the GPU might not be reflected correctly.
HWiNFO on the other hand uses a different approach - reads the utilization direct from the GPU internal registers.
It's hard to judge which method provides correct numbers, maybe each of them but under a certain scenario.
Unfortunately since the method Process Explorer uses is not documented and hidden, I'm not able to evaluate/implement it.

I can only check if I can somehow improve the method used by HWiNFO. Please attach (or send me) the HWiNFO Debug File from a situation where HWiNFO reports 0% and there is a light GPU task active. I'll analyze the data and check if I can improve reporting.
 
Deo Domuique said:
Please, take a look... I drew the parts I'm talking about:

http://i.minus.com/i8B1Kq7LmFNax.png

Those zeros on HWiNFO do not stay 0% for ever... They work some time and the 'Max' will be set. As I said, if I open a game which utilizes heavily the GPU, all of them ( min/max/current ) will work fine, but for light tasks, they rarely work just for a random moment.

http://ie.microsoft.com/testdrive/Performance/FishBowl/

On this same test, if I increase the fishes to 2000, the GPU Utilization works as it should... This highlights that it doesn't work on light tasks. Maybe when the clocks are on idle mode ( 300Mhz ) HWiNFO cannot show the GPU utilization?

Meanwhile, all other monitorings for GPU ( Thermal Diode/Core Clock/Memory Clock etc ) working normally.

Hi Guys,

I don't know it this will help but AMD has an utility called System Monitor. It has a graph on it for those momentary usage presentations on the AMD ATI graphics card only. I am running my AMD in an augmentation mode to my Intel Mother Board Graphics. I am running Sandy Bridge. When there is absolutely no load on the AMD System Monitor shows it obviously in idle. When I turn on Display Port to my same monitor(I am usually using DVI from Intel output on my multi video input Asus Monitor) with simple wallpaper on it, coming from the AMD ATI 6870 Graphics Card, System Monitor shows that minimal activity effectively on the graph and in momentary % value in real time. One can also see a number of other diagnostics.

Enjoy Gentlemen,

Crysta
 
Thanks for the suggestion, Crysta.
Have you also checked HWiNFO in parallel if it reports the same utilization?
 
Martin said:
Thanks for the suggestion, Crysta.
Have you also checked HWiNFO in parallel if it reports the same utilization?

Martin,

in low utilization mine always seem to report 0%. Utilization happens to quickly I suspect.

Crysta
 
Martin said:
A side-by-side screenshot would be nice ;-)

o_O :( Martin,

When I was looking at the NEW Gadget Setup, I found more info. It is part of the Intel CPU 0 suite of values and simple out of order. It is above CPU Package in the Sensor report, and I moved it there in the Monitor Gadget. It is at the very bottom in the .png I sent you.

:p
 
Back
Top