4 ms·

Intel don’t have production runs of CPUs with different clock speeds. They make a batch, test them then work out the clock speed that batch is good for. Then se

Intel don’t have production runs of CPUs with different clock speeds. They make a batch, test them then work out the clock speed that batch is good for. Then sell them at a variety of prices. You can overclock and 90% of the time it’ll be fine. Maybe it dies after 5 years instead of 10. Or maybe after a week. In the case of Tesla maybe that move took brought forward the EOL of the battery by 2 years or maybe 2 weeks. Even our bodies don’t work that way - our brains prevent us from using our full strength except in extremis because if we did our muscles and tendons would be seriously damaged. Soft limits are everywhere.

colejohnson66 6y ago
The word for CPU manufacturers making multiple SKUs is “binning”. Every processor wafer made could attempts at an 8 core beast, but if 1 isn’t working, they disable another and sell it as a 6 core. If the chip isn’t stable enough for hyperthreading, that feature is disabled. Unstable at high frequencies? The turbo clock is lowered. Are the manufacturers benevolent in this? Not always. Sometimes an 8 core chip will be locked to 4 just to make more lower end SKU pieces available for purchase. Or Intel disabling overclocking so you have to buy a “K” processor that costs more. But when people say things like: > When you buy physical things, you should be paying the total cost for the hardware to be manufactured, distributed, and finally recycled or disposed of. They’re ignoring the fact that soft limits actually can make manufacture cheaper. The reason Intel/AMD/etc don’t have a separate silicon template for every SKU is it’s too wasteful (money wise). Each run has a setup cost, and if that can be lowered by reusing the same wafer template, they do that. It’s cheaper to have less manufacturing SKUs and just disable features later. In addition, when soft limits are used on chips, it actually prevents waste. Imagine if the yield on an 8 core wafer was 70%. That means only 70% of the produced chips works at spec. The other 30% would be tossed if binning didn’t happen. By binning, that 30% can be sold (with reduced feature sets).
- jay_kyburz 6y ago
  >It’s cheaper to have less manufacturing SKUs and just disable features later. I think you missed the point. If the chips are not stable or have a problem core, then fine, dial them down and sell them as capable of doing less. But if a chip _can_ do x+y and but you lock y away because you don't want your customers doing y, then there is a problem we should address.
  - tempay 6y ago
    They do make software limitations to make it fit inside their product line. If only one out of eight cores is dead it’ll likely get sold as a 6 core chip not a seven core. If the clock stability limit is slightly too low it will get binned 10% lower for no technical reason. If they didn’t do this everyone would be returning the 6 core versions of a chip until they get lucky and get the 7 core. People already do this a little with overclocking capable chips. Maybe you’d argue they should sell a 7 core chip as well for a slightly higher price but that also means having more product lines and increased manufacturing costs as a result.

colejohnson66 6y ago

The word for CPU manufacturers making multiple SKUs is “binning”. Every processor wafer made could attempts at an 8 core beast, but if 1 isn’t working, they disable another and sell it as a 6 core. If the chip isn’t stable enough for hyperthreading, that feature is disabled. Unstable at high frequencies? The turbo clock is lowered. Are the manufacturers benevolent in this? Not always. Sometimes an 8 core chip will be locked to 4 just to make more lower end SKU pieces available for purchase. Or Intel disabling overclocking so you have to buy a “K” processor that costs more. But when people say things like: > When you buy physical things, you should be paying the total cost for the hardware to be manufactured, distributed, and finally recycled or disposed of. They’re ignoring the fact that soft limits actually can make manufacture cheaper. The reason Intel/AMD/etc don’t have a separate silicon template for every SKU is it’s too wasteful (money wise). Each run has a setup cost, and if that can be lowered by reusing the same wafer template, they do that. It’s cheaper to have less manufacturing SKUs and just disable features later. In addition, when soft limits are used on chips, it actually prevents waste. Imagine if the yield on an 8 core wafer was 70%. That means only 70% of the produced chips works at spec. The other 30% would be tossed if binning didn’t happen. By binning, that 30% can be sold (with reduced feature sets).

jay_kyburz 6y ago
>It’s cheaper to have less manufacturing SKUs and just disable features later. I think you missed the point. If the chips are not stable or have a problem core, then fine, dial them down and sell them as capable of doing less. But if a chip _can_ do x+y and but you lock y away because you don't want your customers doing y, then there is a problem we should address.
- tempay 6y ago
  They do make software limitations to make it fit inside their product line. If only one out of eight cores is dead it’ll likely get sold as a 6 core chip not a seven core. If the clock stability limit is slightly too low it will get binned 10% lower for no technical reason. If they didn’t do this everyone would be returning the 6 core versions of a chip until they get lucky and get the 7 core. People already do this a little with overclocking capable chips. Maybe you’d argue they should sell a 7 core chip as well for a slightly higher price but that also means having more product lines and increased manufacturing costs as a result.