They don’t exactly fail the quality checks, they get binned into a lower grade. It’s a common practice in many industries when reworking isn’t possible or financially viable.
It isn’t necessarily a bad thing either. Consumers can save some money when they don’t need top performance, the company gets some revenue, and the products don’t go into a landfill right away.
So it’s not good enough for the main product and gets put into a different pile? That sure sounds like a failed QC check to me. I agree with you though about the excess.
This is how chip fabrication works. It’s how chip fabrication has always worked. Chip yields are never perfect. The lithography processes aren’t 100% accurate, the substrates aren’t 100% pure, and some portion of chips come out of the oven with flaws. Always. That’s just how it is.
That doesn’t mean those chips don’t work, it just means they won’t work to their maximum design specifications. So to use a very simplified example, a 32 gig flash chip with a flaw found in it somewhere can be burned so the flawed portion will never be used, and then binned as a 16 gig chip. A processor core that’s not stable at its maximum rated speed can be burned so it’ll be a perfectly serviceable lower speed version. Etc. The process is already done, the material is already used, and that chip can still be serviceable. So why throw it away?
That’s the white hat version.
The black hat version is, some shady manufacturer could take the chips that are too flawed to be used reliably for any purpose, and use them anyway. Where that distinction is drawn is pretty important.
Depends on the nature of the failure. This is super common in CPUs for example. The absolutely perfect ones go in the top quality bin, those are your i9 and Ryzen 9 CPUs. Ones that are functionally fine, but a bit unstable at the higher clock speeds end up in the next lower quality bin, so something like an i7 or a Ryzen 7. Then you have ones that maybe have a failed core or two, those ones have the failed cores fused off and go in the next lower bin with lower core counts. Etc.
In the case of a flash memory chip as long as it isn’t corrupting data at a certain speed I’d expect it to be binned down to a lower quality bin. Assuming whoever buys that chip runs it at the clock speed the binning says to run it at it should be fine. Where you’d run into problems is if whoever buys it ignores what they were told and runs it at the full speed, in which case you’re going to end up with corrupted data.
Likewise other things that could lead to a usable but lower quality chip is if certain areas of memory were bad. As long as you disable those failed memory regions and sell it as a lower capacity chip that’s also fine. Once again though if the buyer re-enables the failed memory regions then that’s a problem.
They don’t exactly fail the quality checks, they get binned into a lower grade. It’s a common practice in many industries when reworking isn’t possible or financially viable.
It isn’t necessarily a bad thing either. Consumers can save some money when they don’t need top performance, the company gets some revenue, and the products don’t go into a landfill right away.
So it’s not good enough for the main product and gets put into a different pile? That sure sounds like a failed QC check to me. I agree with you though about the excess.
it fails the top tier qc check, but passes a low tier qc check. That’s how different price points/tiers for CPUs exist.
And then there’s those who fail the low tier check and “some friend” gets them for scrap en masse.
That sounds like two different ways of saying the same thing.
It failed to be the best, but it was good enough to be used for something else.
This is how chip fabrication works. It’s how chip fabrication has always worked. Chip yields are never perfect. The lithography processes aren’t 100% accurate, the substrates aren’t 100% pure, and some portion of chips come out of the oven with flaws. Always. That’s just how it is.
That doesn’t mean those chips don’t work, it just means they won’t work to their maximum design specifications. So to use a very simplified example, a 32 gig flash chip with a flaw found in it somewhere can be burned so the flawed portion will never be used, and then binned as a 16 gig chip. A processor core that’s not stable at its maximum rated speed can be burned so it’ll be a perfectly serviceable lower speed version. Etc. The process is already done, the material is already used, and that chip can still be serviceable. So why throw it away?
That’s the white hat version.
The black hat version is, some shady manufacturer could take the chips that are too flawed to be used reliably for any purpose, and use them anyway. Where that distinction is drawn is pretty important.
Depends on the nature of the failure. This is super common in CPUs for example. The absolutely perfect ones go in the top quality bin, those are your i9 and Ryzen 9 CPUs. Ones that are functionally fine, but a bit unstable at the higher clock speeds end up in the next lower quality bin, so something like an i7 or a Ryzen 7. Then you have ones that maybe have a failed core or two, those ones have the failed cores fused off and go in the next lower bin with lower core counts. Etc.
In the case of a flash memory chip as long as it isn’t corrupting data at a certain speed I’d expect it to be binned down to a lower quality bin. Assuming whoever buys that chip runs it at the clock speed the binning says to run it at it should be fine. Where you’d run into problems is if whoever buys it ignores what they were told and runs it at the full speed, in which case you’re going to end up with corrupted data.
Likewise other things that could lead to a usable but lower quality chip is if certain areas of memory were bad. As long as you disable those failed memory regions and sell it as a lower capacity chip that’s also fine. Once again though if the buyer re-enables the failed memory regions then that’s a problem.