GPU-Z is a lightweight utility that’s frequently used to identify GPUs and parse their various characteristics. It’s capable of logging a number of GPU variables in real-time, which is useful for measuring throttling, clock, and fan speeds, and it can tell you a great deal about the specifics of your GPU’s support for APIs like Vulkan, OpenGL, and DirectX. Now, the utility has added a new feature — the ability to detect counterfeit cards.
I was a bit surprised to read it because it’s been a long time since counterfeit parts were a major issue in the United States. 18-20 years ago, when I ran my own computer service and repair business, I had more than a few college students walk through the door with, shall we say, interesting hardware configurations. Back then, it wasn’t unheard of for unscrupulous individuals to save money by overclocking a CPU (typically by hardware jumper settings, though BIOS overclocking was in its early stages) and selling it as the more expensive part. Push a K6-2 266 up to 300MHz-333MHz, and you could pocket the additional premium between the two chips’ sale price. In one memorable case, I had a customer show up with a no-name knockoff motherboard and a Celeron that insisted it was actually a Pentium II.
Intel and AMD put a stop to this kind of hacking through a variety of methods, including locking the multipliers of various CPUs and, in some instances, building motherboards that didn’t respond well to bus overclocking to prevent counterfeit hardware from being hidden that way. We haven’t heard as much about this issue on the GPU side of the equation, though I believe it’s cropped up a time or two before, mostly when ATI or Nvidia would release a lower-end card that actually deployed the same version of a GPU as a higher-end solution. In these instances, a simple BIOS flash could sometimes result in a much faster, more powerful GPU, assuming that the additional resources had been locked off for market segmentation purposes rather than because they just didn’t work.
Evidently, GPU counterfeiting is now enough of an issue in Chinese markets that GPU-Z has added the ability to identify whether your card is what it claims to be. In the image below, a normal GPU is on the left, an identified fake card on the right.
To a knowledgeable eye, the GPU on the right obviously isn’t a GTX 1050 Ti — the core, shader, and ROP figures are all wrong. But since that’s not automatically clear to the end-user, the utility now displays a prominent FAKE warning next to the name of the card, with a warning logo where the company logo would normally sit. The GF116 was used for the GTX 550 Ti, GTX 560M, GT 545, and the GT 545 (OEM variant). And, as the app notes, it’s a seven-year-old GPU — not a two-year-old chip.
This utility could be particularly useful for laptops, where it’s much harder to check the make and model on the GPU, but it could also be useful in situations where the counterfeiter might have altered the labels on the desktop card. Printing replacement stickers and art are pretty trivial for most companies and one might not see the difference without close examination of the card itself. Hopefully, tools like this will help crack down on fraud against Chinese gamers and the unscrupulous companies rebranding hardware. Hat tip to Tom’s Hardware for spotting the update. GPU-Z may be downloaded here.
GPU-Z Adds Fake GPU Detection Capability to Combat Counterfeiting - ExtremeTech
