Image Image Image Image Image Image Image Image Image Image

This 8-Bit Life | September 26, 2016

Scroll to top



NVIDIA cripples firmware on their Video Cards - This 8-Bit Life

NVIDIA cripples firmware on their Video Cards
Leland Flynn

NVIDIA has some explaining to do here. They sell two separate video cards, the consumer level GTX690 and the professional level K5000. Both of these cards utilize the same GPUs but the GTX690 is simply a crippled K5000. It’s very frustrating to see NVIDIA treat customers this way. Moreover, they know the sort of people that they market to, they couldn’t have expected that their hardware would not be probed at some point.

They accomplish this by changing the analog values that determine what model the card represents itself as.

gtx 690 mod

A user named Gnif on the EEVBlog forums has found a way to unlock the hidden power of his GTX690 by desoldering one resistor and soldering on a 15k and 20k on specific pads of the board.

So why am I irritated by this? Because the difference in cost for the cards is about $800!

You can read all about Gnif’s hard work HERE.


  1. They’ve been doing this for close to a decade. I think the first time I remember making a resistor mod to make an nVidia desktop card think it was a workstation card was circa 2005. It has to do with “what the market will bear,” and you get other tools and features with the workstation card. It’s like GM selling a Chevy with leather seats as a Cadillac and marking it up 20K.

    • Leland Flynn

      Oh I understand. It’s still a poor business practice.

      • thisguy

        I guess you’ve never been to business school

        • Leland Flynn

          Can’t say that I have. Would it provide a moral argument to make this behavior acceptable?

          • Anonymous

            Business isn’t moral. Either every single card, regardless of market, costs a truckload more since the cost of manufacturing a die is astronomical, or they segment the market this way. Your choice.

  2. EbilPhish

    This isn’t any kind of scam, its just how hardware manufacturing works.

    They make the microchips, during the manufacturing their are a percentage that develop flaws but the chips are generally designed to work around the problem to some degree. The ones that fail the testing procedures have their performance scaled back.

    For example you can buy 3 core processors that are just failed 4 cores ones. Desktop processors are sold at different speeds depending on what speeds they are stable at. It’s why it is possible to overclock.

    You can also buy monitors that are the high priced 27-30″ Mac cinema displays for a few hundred instead of thousands, except you run the risk of getting some minor flaws (dead pixels, bad lighting, etc…). Look up Catskill monitors.

    If you are modifying a GTX690 into a K5000 you run the risk of having faulty hardware. Chances are results will vary depending on how imperfect the chip and the environment it is running in, some will work fine for some peoples usage patterns, others will become crashtastic or burnout.

  3. Alex

    You benefit from this practice, in terms of lower costs and access to ‘cutting-edge’ developments in the industry. Think it over and you should hopefully soon see why.

  4. NickG

    This has been standard practice in the semi-conductor industry for generations, you make a single high performing chip and bin, fuse, and disable it into lower markets. This is true of basically every type of processor you encounter nowadays from server workhorses to the desktop market to CUDA-core supercomputers, the lower end has some features disabled that they might be capable of.

    It\’s much cheaper to build semi-conductors this way for a couple of reasons, the primary two being yield recovery and the fact that designing chips is bloody expensive. To create a chip that _didn’t_ have Quadro features and sell it to you would be more expensive than selling a Quadro-disabled chip.

  5. exDM69

    The most expensive thing in workstation cards like Quadro and Tesla is warranty and support (this can easily account for a lot of the $800 price difference). This is also the reason that they are clocked down, they have to be more reliable than a gaming gpu.

    Gaming GPUs have some parts fused off. These parts are not essential for gaming performance, and powering them off with fuses gives better thermal properties, which allows more gaming performance. These things run HOT.

    If you want a gaming GPU, you pay for one. If you want something for serious work, you need something different at a different price. It’s not like they’re the exact same thing, despite the fact that they use chips that come from the same wafer. The cards, memories, thermal properties, etc are different.

    (also an important thing mentioned earlier in the comments: yield problems and selling partially defective chips for cheap with the defective parts powered down)

Submit a Comment