sc94597 said:
The point being made, repeatedly, is that the Switch 2, PS5, and Series S/X have specialized R&D for their chipsets. You can't buy another device with their specific combination of hardware (at least upon release, AMD has resold cut-down variants of the 9th Generation Console APUs to consumers.) They are what we typically call "semi-custom" which means there is extra R&D by the chipset designer put into them to tailor them to the specific SKU they were to be sold in.  This was not true of the original Switch. It literally had the same SOC as other devices. It's like buying a custom design for a home from an architect and making some modifications in the home plan, versus buying a traditional blueprint that is the same plan for many homes with no changes. For the specific discussion with bonzobanana, it is this distinction that mattered, because in order to say the "Switch is under-clocked compared to say an Nvidia Shield" all else really does need to be equal. Once you have different core counts, or a different CPU-variant the constraints that help determine things like optimal clock rate, and efficiency curves change, even with the same micro-architecture. The point wasn't to make a general comparison between the Switch and Switch 2, or whether or not the Switch 2 has a more suitable SOC. It was to illustrate that we don't have the same reference point when it comes to Switch 2 that we did with Switch. The closest thing to it is the T234, but the T234 is a very different chipset designed for a very different purpose, has a different CPU, has different memory and core configurations, has specialized edge-compute hardware, etc. | 
