Hello,

I recently bought a used 3070 TI online, and thought I should come here to ask for some advice. The GPU I have currently I bought new from a retail store, so I didn’t have much need to verify it.

I am aware there are resellers out there with more shady interests, and sometimes “hack” a gpu chip to show up as, for example, a 3070 TI when it is really a [ insert any other inferior gpu here].

My question is, how do I test this 3070 TI to ensure it is, in fact, a 3070 TI and not something pretending to be one. My best guess is to run a benchmark and compare my scores to others with similar hardware, but there’s got to be a better way?

  • Fubar91@lemmy.world
    link
    fedilink
    English
    arrow-up
    6
    ·
    edit-2
    10 months ago

    CPU-z can assist in detection. Another option is disassemble of the card from the fans, heat sink, etc. And verify the physical components vs. A known good source card.

    You could run some GPU based benchmarks and compare results to other user submitted benchmarks to see if there’s a massive deviation from the norm. Ymmv, unless the results you’re comparing to have a similar entire system setup.

    There’s also nvidia profile inspector, can compare that with cpuz to point out any spoofed possible changes if they don’t align data wise.

  • smokedclover@feddit.de
    link
    fedilink
    English
    arrow-up
    4
    ·
    10 months ago

    GPU-Z can detect some reflashed GPUs, not sure to which extend. This in combination with your benchmark idea should do the trick. Or you take it apart and look up the markings on the die.

  • KoalaUnknown@lemmy.world
    link
    fedilink
    English
    arrow-up
    3
    ·
    10 months ago

    I would make sure GPU-Z claims it to be a 3070ti and then run the Time Spy benchmark from 3DMark to make sure it performs as expected.

  • MangoPenguin@lemmy.blahaj.zone
    link
    fedilink
    English
    arrow-up
    2
    ·
    10 months ago

    A benchmark is going to be the most reliable way I think, it seems like the fake GPUs tend to reflash a really crappy one, so benchmark results from a fake one would be extremely different from what they should be.