here are also FPGAs and GPUs made on TSMC 7nm I am not sure about FPGA, but NVIDIA GPU's are a 2020 product, as I assume are AMD's.
Wafer capacity from TSMC is way higher than from Intel. I thought we were talking leading edge capacity? Over 60% of TSMC output is 40nm or older. Intel is starting up their third 10nm factory, on the way to 6 total. Two of those will fairly quickly convert to 7nm. Do you have any data on TSMC 7nm versus 10nm INTC wafer capacity? TSMC overview is here: https://www.tsmc.com/english/dedicatedFoundry/manufacturing/fab_capacity.htm The X86 market is somewhere between 300MU and 400Mu/year with a fairly large die size. For comparison, we noticed the apple die size is much smaller. They are shipping around 175Mu/year of smartphones, with a total market of almost 1.5B units. I suspect most of those smartphone units are on older less expensive technologies.
How is an increased cache limiting clock speed? Both Intel 14nm and AMD products have 32KB L1 data caches. Sunnycove architecture increased that to 48K, or a 50% increase. They also increased the L1D cache latency to 5 cycles from 4, for a 25% increase. It is not clear if that was enough or not. What we do know is based on the benchmarks, 10nm is getting more work done per watt than 14nm.
News 
Market Data 
Discover 
