Q9650 using 8.91watts & 8.66amps ??

[BTRY B 529th FA BN]

?

 

[jtisgeek]

looks like it's in c1 or what ever put it under load and check it again.

 

[BTRY B 529th FA BN]

Under load

 

[Lonyo]

What's the question?

 

[BTRY B 529th FA BN]

I'm just curious if those CPU watt and amp numbers are right... I thought the Q9650 had a thermal design power of 95w.

 

[Cogman]

I'm just curious if those CPU watt and amp numbers are right... I thought the Q9650 had a thermal design power of 95w.

It seems a bit low to me, but then again, it could be related to your stressing method.

besides having a CPU pegged at 100% utilization, the actual instructions you are using have just as much affect on power usage as the clock speed does

If you are just doing a nop loop, its quite possible for that to be the power draw.

 

[veri745]

Doubt it's very accurate, but it might be a decent estimate.

You'd probably have to find an artificial load like OCCT or linpack to get close to TDP. Not sure what you were running there.

 

[BTRY B 529th FA BN]

Under load

I was running LinX and got these results.

 

[Martimus]

P=I*V. They probably just used that to calculate the wattage.

P= 8.66A * 1.03V = 8.92W

That is within 0.01 of what the program states, so it is within rounding error. I would say that the program uses that equation to find power.

 

[Cogman]

P=I*V. They probably just used that to calculate the wattage.

P= 8.66A * 1.03V = 8.92W

That is within 0.01 of what the program states, so it is within rounding error. I would say that the program uses that equation to find power.

The question I have is how it is calculating the current. AFAIK, no motherboard has current reporting sensors.

 

[BTRY B 529th FA BN]

The question I have is how it is calculating the current. AFAIK, no motherboard has current reporting sensors.

Good question.

 

[RussianSensation]

But you are only running the processor at 9x300. That's not stock speeds. Therefore, you were able to lower the required voltage to run at only 2.7ghz. Set your BIOS settings to default and see what nominal voltage is required to run at stock speeds of 3.0ghz. Don't undervolt either. I imagine the processor will need 1.20V or higher for those speeds and you should see rising wattage.

Finally, the software monitoring program may not actually be reporting the correct numbers. Just to give you an idea, my HW Monitor reports 95W at full load for 1.31V Core i7 @ 3.9ghz. This is mostly likely incorrect though, and seriously understated: http://www.xbitlabs.com/articles/cpu/display/power-consumption-overclocking_10.html#sect0

 

[BTRY B 529th FA BN]

9x333 @ idle & @ load

 

[RussianSensation]

Ok just think about it...Your idle wattage is near dual core atom load of 8-9W. Your load wattage at 3.0ghz is 39 watts. It doesn't take a scientiest to realize that the software reported numbers are incorrect. I pointed this out already by showing you that my i7 @ 3.9ghz shows up at 95W, but total system power consumption with the graphics card at idle would be > 250W as measured by xbitlabs.

 

[TridenT]

You do know that these software programs often report inaccurate measurements, right?

 

[BTRY B 529th FA BN]

I know it doesn't look right. That's why I was curious about it. I'm not trying to brag or anything.

 

[BTRY B 529th FA BN]

Yes, I know.

Has anyone of you opened up HWmonitor and looked to see what yours says? EDIT: I'm looking to find a conclusion of proof that it is wrong. I do agree it doesn't look right, i'm just trying to get to the bottom of this.

 

[PottedMeat]

The question I have is how it is calculating the current. AFAIK, no motherboard has current reporting sensors.

If that screenshot is right, the ASP0800 ( I guess Analog Devices sold the IP to On Semiconductor ) does do current sensing and reporting. The datasheet says it has sense amplifier that can read the voltage across a current sense resistor or the inductor ( probably the case, cheaper ).

Unfortunately the accuracy could be anything, I would guess not really implemented.

 

[EarthwormJim]

I've noticed that the current sensing on my x3210 (65nm core 2 quad) was reasonable when I was at stock speeds and stock voltage. The second I changed my FSB, accuracy disappeared.

At load it was saying I was using 80 W under LinX at stock speeds (2.13ghz). When overclocked to 3.8ghz, it was saying I was using 30 W under load...

 

[Gillbot]

http://forums.anandtech.com/showthread.php?t=257485

 

[EarthwormJim]

http://forums.anandtech.com/showthread.php?t=257485

That won't tell you current draw by the CPU. You'd have to put the multi-meter in series with the CPU's rails, there's really no way to do that.

I suppose if you could find a sense resistor to use on the motherboard, you could measure the voltage across it.

 

[PottedMeat]

That won't tell you current draw by the CPU. You'd have to put the multi-meter in series with the CPU's rails, there's really no way to do that.

I suppose if you could find a sense resistor to use on the motherboard, you could measure the voltage across it.

i doubt they went with the resistor, here are the monitor chips options:

Output inductor DCR sensing without a thermistor for
lowest cost.

Output inductor DCR sensing with a thermistor for
improved accuracy with tracking of inductor temperature.

Sense resistors for highest accuracy measurements.

you'd have to either know the exact model of the inductor or pluck it off and measure its resistance and use that correct the reported values.

that's an interesting chip - it gives options for V/I/Power readouts:

0x8B R 0x00 READ_ VOUT 2 Readback output voltage. Voltage is read back in VID Mode
0x8C R 0x00 READ_IOUT 2 Readback output current. Current is read back in Linear Mode
(Amps)
0x96 R 0x00 READ_POUT 2 Readback Output Power, read back in Linear Mode in W’s.

 

[EarthwormJim]

i doubt they went with the resistor, here are the monitor chips options:



you'd have to either know the exact model of the inductor or pluck it off and measure its resistance and use that correct the reported values.

that's an interesting chip - it gives options for V/I/Power readouts:

Honestly makes sense that it has those options. It dynamically adjusts current and voltage for correction purposes, so on some level it must detect them anyway.

 

[hans007]

I'm just curious if those CPU watt and amp numbers are right... I thought the Q9650 had a thermal design power of 95w.

tdp doesnt mean how much power your cpu uses.

TDP is a guideline for heatsinks so a heatsink must be able to cool 95W. intel makes a lot of 95W tdp cpus theyare all different speeds and voltages.

think of it like tire ratings. cars that have z-rated tires on them do not all go 160mph.

 

[BTRY B 529th FA BN]

tdp doesnt mean how much power your cpu uses.

TDP is a guideline for heatsinks so a heatsink must be able to cool 95W. intel makes a lot of 95W tdp cpus theyare all different speeds and voltages.

think of it like tire ratings. cars that have z-rated tires on them do not all go 160mph.

Ah, ok, that explains things a little :thumbsup: