AMD is almost certainly disabling fully capable hardware to make 6-core parts because it's unlikely that they have enough defective hardware to meet the demand for 6-core parts. There aren't enough people willing to buy 8-core CPUs so some get the chop. Granted this also means that the bin for an 8-core CPU becomes better. They can find the clock speed that whatever percentage of chips they want for that segment to hit and set the bar there.
It's no different than the Phenom II days when people were buying x3 CPUs because most of them could be unlocked and work as a fully functional x4 part. A lot of people on these forums would buy an x3 to save $50 or so.
We probably have more variability in the top and bottom bins these days because there's fewer products on offer. Right now AMD has 10 Zen 4 CPUs. Half of them are 2 chiplet CPUs. Additionally there is at least one v-cache model in each of those halves. If you're getting a non-v-cache single chiplet Zen 4 CPU then you have four options, again segmented by core count.
If we use the Phenom II x4 as a comparison, AMD had 12 different product models for just their 4-core CPUs. You basically bought the CPU at 100 MHz clock increments for a large part of the product range. Intel wasn't all that different at the time either. All of the variability was wrung out by having so many different products.
AMDs economics are completely different due to chiplets. Before measuring the capabilities of one it could wind up in $10,000 server CPU or a $200 desktop CPU. The flexibility is something that didn't exist to such a broad extent previously and it makes AMD far more adaptable to changes in the market. Intel is going in this direction, but lead times on new designs are so long that anything that Pat has pushed forward won't emerge for a few years still. He might look ridiculous for cheerleading a few turkeys, but that's his job as CEO as much as getting better products to market.
It's no different than the Phenom II days when people were buying x3 CPUs because most of them could be unlocked and work as a fully functional x4 part. A lot of people on these forums would buy an x3 to save $50 or so.
We probably have more variability in the top and bottom bins these days because there's fewer products on offer. Right now AMD has 10 Zen 4 CPUs. Half of them are 2 chiplet CPUs. Additionally there is at least one v-cache model in each of those halves. If you're getting a non-v-cache single chiplet Zen 4 CPU then you have four options, again segmented by core count.
If we use the Phenom II x4 as a comparison, AMD had 12 different product models for just their 4-core CPUs. You basically bought the CPU at 100 MHz clock increments for a large part of the product range. Intel wasn't all that different at the time either. All of the variability was wrung out by having so many different products.
AMDs economics are completely different due to chiplets. Before measuring the capabilities of one it could wind up in $10,000 server CPU or a $200 desktop CPU. The flexibility is something that didn't exist to such a broad extent previously and it makes AMD far more adaptable to changes in the market. Intel is going in this direction, but lead times on new designs are so long that anything that Pat has pushed forward won't emerge for a few years still. He might look ridiculous for cheerleading a few turkeys, but that's his job as CEO as much as getting better products to market.