8

Someone once told me the following:

Processors, as good as they are, will always make mistakes.
When processors (i3, i5, i7, ...) are tested before distribution, they are categorized on the amount of mistakes they make. An i7 7700k for example which makes very few mistakes is labeled as a 'Type A', while another i7 7700k with the exact same name and specs makes a little bit more mistakes, and is labeled as 'Type B'.

All the 'Type A' processors are used and sold in business class laptops and workstations while the 'Type B' ones are sold to consumers.

After some research I couldn't find anything on it on the internet.

Anyone know if this is true or straight up bullshit?

Comments
  • 4
    Never heard that for processor but they do make ECC memory to correct for corrupt bits. They don't really use the same memory though and label it ECC, it's a totally seperate product that your cpu/mobo has to support, and it's typically slower. People usually use it for servers / scientific computing where errors / crashes are not an option.
  • 3
  • 6
    I'd call it bs.

    Remember AMD Phenom TLB Bug? It caused rare crashes on very specific scenarios. But it was fixed by disabling the feature and later fixing it on the silicon.

    If you make chips that frequently make mistakes, you will be out of business soon.
  • 4
    There are the celeron processors, that are low class Pentium/Core-i/Xeon processors with defective and / or disabled cache units (Celerons work otherwise normally) and lower frequencies.

    Warranty claims would cost Intel millions if processors are misbehaving / miscalculation (see the FDIV-bug).

    For business there are vPro-processors (with advanced security and maintanance capabilities) and Xeon processors (with ECC capabilities).

    All in all I would call this very, very unlikely.

    If you want to find misbehaving instructions, try https://github.com/xoreaxeaxeax/... out.
  • 1
    @hexc Thank you, I did not know this
  • 1
    @sbiewald Thank you very much! Very informative
  • 2
    @8BitOverdose to add, it's not uncommon for manufacturers to bin lower preforming chips and disable aspects of them and sell them as a different lower end products, though this typically doesn't have to do with miscalculation, rather they may not clock to the required frequency or they may have faulty components. That said they would only rebrand them IF they perform properly for the SKU they are going to rebrand them to. In short, they have to be working 100% for the spec they are being labeled.
Add Comment