Radeon RX 480 GPU Core Current

[Swarlos]

I'm running hwi_589_3520 on Windows 7 64-Bit. Radeon RX 480 GPU.

I ran a Unigene Valley Benchmark a little while ago and when I was going over my temperatures and power draw I saw something that didn't look right to me, but I'm far from an expert on electrical stuff.

I have a screen shot showing max GPU Core Current at 119.094 A.  If I'm understanding that correctly there was a peak draw of 119 Amps?  I'm hoping a sensor threw out a garbage number somewhere because that doesn't look right to me.

https://pasteboard.co/HGCwDHM.jpg

And yes, I'm running an older system with a Core i5-2500k so the PCIe v2 limitation is correct.

Thanks for your help.

 

[TomWoB]

Hi Swarlos,

run the Benchmark again with Logging switched on, but "Polling Period" should be at least 1000ms (otherwise needed ressources for logging are too high, especially during a benchmark):

[attachment=3055]

In the logs it's much easier to check if such a high value was a "spike" (invalid) measurement or not ... e.g. I saw spike values of temperatures underneath "absolute zero point"

 

[Swarlos]

Hi Swarlos,

run the Benchmark again with Logging switched on, but "Polling Period" should be at least 1000ms (otherwise needed ressources for logging are too high, especially during a benchmark):



In the logs it's much easier to check if such a high value was a "spike" (invalid) measurement or not ... e.g. I saw spike values of temperatures underneath "absolute zero point"

I just opened up the settings to increase the polling period and it's already at 2000ms.

I'll try running it again to see if I can duplicate the problem.

Update:

Ran it again, got similar high max amperage readings.

https://pasteboard.co/HGD2Fpc.jpg

 

[TomWoB]

2000ms is fine ... can you upload a logifle of the benchmark?

 

[Swarlos]

I haven't used the logging feature yet, but I think this is it.

http://www.filedropper.com/benchmarklog

 

[TomWoB]

Hi Swarlos, yes this is it!

[attachment=3056]

If you have a look at the "GPU Core Power" (up to 150W) you can see that the 120 ampere make sense by a voltage of 1.1-1.2V, because
Power = Voltage * Current

But these 120 ampere are "internal the graphic card", not from power supply to graphic card (which is powered by 12V). Hmmm ... your 12V drop down is a little bit high 12.20 -> 11.83 V ... thin cables ?

 

[Swarlos]

Hi Swarlos, yes this is it!



If you have a look at the "GPU Core Power" (up to 150W) you can see that the 120 ampere make sense by a voltage of 1.1-1.2V, because
Power = Voltage * Current

But these 120 ampere are "internal the graphic card", not from power supply to graphic card (which is powered by 12V). Hmmm ... your 12V drop down is a little bit high 12.20 -> 11.83 V ... thin cables ?

By the 12V you mean the 8-pin PCI Express cable from the PSU?  I'm using the cable that came with PSU when I bought it a little over 6 years ago.  OCZ600MXSP
https://www.newegg.com/Product/Product.aspx?Item=N82E16817341017

Thanks for your help, I thought the 120 amps were going through 12V, not 1.1-1.2V. That makes me feel a lot better.

 

[Martin]

119 Amps is well possible on those GPUs, so I believe it's an accurate value. >100 Amps on higher performance GPUs is no problem.

 

[TomWoB]

I think the 12V are measured on the motherboard, so it's more the 24-pin cable to the motherboard. But ... I think it's not a "cable issue", it's more that your power supply voltages goes "a little bit too much down under pressure". Btw ... your needed power "is not so high", 150W for a graphic card is "mid-range", my GTX 1080 Ti takes 280W and more, but my Seasonic power supply goes only down from 12.19 - 12.10V.

Anyway ... if your system is running stable ... everything seems to be fine