Originally Posted by sugarhell
It's irrelevant when you compare the volts between 2 different nodes. It's like comparing the volts of a SOI node to finfet.
There isn't any argument here. Doesnt change the fact that gpus are high performance parts and amd is using a mobile soc low power node.
I will correct you one last time.
It's not a low power node. AMD is releasing Zen on this node, a high performance cpu, they are releasing high performance gpus on this node. Why would they gimp themselves? it makes absolutely zero sense, much like your statements.
Performance is a measure of voltage. I will repeat that statement. TSMC 16nm- base voltage 0.70 volts, Samsung 14nm- base voltage 0.800 volts, Intel 14nm- base voltage 0.75 volts. The reason they call it "low power plus" or a low power node is because of Finfet. With finfet designs you don't need as much voltage to get high performance. This allows you to reduce the total voltage, improving efficiency and keeping high clocks.
Base voltage of the 28nm node is between 0.6 and 1.1 volts, 0.6 is the ultra low power stuff, and 1.1 is the super high performance stuff.
ok so if 28nm high performance took 1.1 volts, and the finfet 14nm takes 0.800 volts you can clearly see why they call it a low power node. Again it doesn't mean they cannot reach high clocks because finfets don't need no where near as much voltage to achieve higher clocks.