Many of our reviews in the past have included power consumption testing for motherboards, graphics cards, as well as CPUs and I’ll be very honest, coming from an electronics-background family, the accuracy of these results are not up to what I’d love them to. While I do have access to clamp meters and other more direct testing tools, the setup time just kept it from being an efficient workflow. I searched high and low around AliExpress to see if there’s an inline way to test PC components but that didn’t yield anything. That was around 3 or 4 years ago, a few months later I finally found my solution: Powenetics.
Aris Mpitziopoulos, together with Tom’s Hardware, rolled out Cybenetics’ Powenetics. Now prior to this, I’m aware who Cybenetics and their PSU certification body role as we’ve covered their debut and made a primer about their certification process. The Tom’s Hardware debut article also helped in walking me through the details and that helped me decide that Powenetics was the tool I needed so I contacted Cybenetics.
The Need for More Accurate Power Testing
Larger publication in the world have dedicated testing equipment when measuring power draw from different components and this involves scopes, multimeters, clamp meters, etc. This by itself is quite expensive and like I said I have access to these but its the time to set it all up and move it from one system test to another is what made me decide against it.
Before, I used a simple power outlet meter. Kill-a-watt is the popular brand in the west, in the Philippines its the Omni Power Meter. I’ve went through a lot of these things trying our functionalities but I settled from a Chinese model named the HOPI HP-9800. This meter comes in multiple configurations, I went with the model that has a USB interface to connect to the computer. This worked out for me since it has a CSV log of per-second reading from the outlet but much like all power outlet wattmeters, it still is measuring the entire system draw.
Many publishers and reviewers have worked out the math and have all agreed that we can at least infer the power draw of a component by stressing and subtracting from the idle system draw. Logically, that’s correct but there is a lot at play here including other system draws as well as the PSU efficiency. To oversimplify things: it’s just not that good a measurement.
The good thing about modern CPUs and power supplies is that they have API readings which shows power draws directly from the device. NVIDIA has theirs. AMD has theirs, Intel has theirs. And that’s where the problem lies. That means its questionable to use AMD API readings to directly compare that with NVIDIA API readings. While these readings may be close to real power draw, each company may implement it different and ultimately becomes useless when comparing between vendors. As the data is being provided by the component itself, the data is put to question as well. Motherboard vendors have recently been called out by HWINFO due to misreporting actual readings for AMD motherboards, so this only adds fuel to the fire of doubt towards motherboard and API readings.
Powenetics Concept
Powenetics’ logic is simple: read the power draw for each component only by going between the component and the power supply.
Think of it like the power meter we mentioned earlier. That power meter goes in between your system and the outlet measuring the entire system draw. Powenetics works the same,except instead of the AC outlet, its your PSU plug connecting to a measurement instrument and then passing that power to the component whether its the GPU, CPU or some SATA device.
Sounds easy, right? It is, but the hard part is getting those instrumentation in-between the component and power supply.
That’s done through this device. The photo above is the complete Powenetics hardware kit comprising of multiple measurement instruments from German DIY company Tinkerforge and as well as a PCIe riser. All of these came to around $1200 as shipment was extremely costly during the pandemic but on better times, it could drop to around $800 or less. Still, its an investment worth the money.
All of these come together to form this:
It’s not prettiest thing I know but then again its simple and it works. A little bit of background on how the system is installed.
Powenetics requires us to directly connect the Tinkerforge bricklets to the PSU which means cutting the wires and making basically a custom connector. We’ve went with the prescription Powenetics way of setting things up but I do have another kit I’d be building in the future so I might make an interposer/receiver card setup with the kit.
Take note of the PCIE riser as well, to fully measure GPU draw, the PCIe riser acts as a way to take readings from the PCIe slot without butchering the motherboard.
With all that set-up, all we need is the software.
Powenetics Software
With the bricklets in-place, all the readings comes to one place with the Powenetics software. This application shows us a lot of information separated into tabs for GPU readings, CPU readings and ATX (MOLEX/SATA) readings.
As we use multiple sensor bricklets on each power line, we have individualized reading as well as overall each component.
Data can be logged for visualization in the log tabs and can be exported to Excel.
The GPU log will look like this. The CPU and ATX log will of course have different columns. The important part here is how we use this information.
Here’s a quick draft of charts I whipped up from a quick run with an RTX 2080 Ti running AIDA64 GPUÂ load. This information is vital into presenting actual GPU, CPU and ATX power and will be key in our reviews going forward. It removes doubt from external factors from the measurement process and this makes it possible for media using the same reading to collaborate and exchange information.
Conclusion
I’m still actually building up my workflow using Powenetics but the important part is that the data measurement is there and there’s a lot for me to explore using the system. More importantly, I appreciate the peace of mind this has given me because a reviewers functions as a means for average consumers to have no more doubts in their purchase and with a measurement tool like this, doubt is obviously removed. This also allows us to police companies who may be misreporting power readings or to inform them of possible problems with the draw of their components.
We’d like to thank Aris Mpitziopoulos and the Powenetics development team at Cybenetics for sharing with us the system. We will put the system into good use going forward.
