Discussion

• @hattom Tom Hayward on x

  @IanCutress @Radeon_Rebels What size is TSMC's equivalent to Intel's 7nm? 5nm? I assume there is still a discrepancy between transistor density per “nm” between Intel's process and others, with Intel's “nm” label pessimistic compared to their density?

• @iancutress @iancutress on x

  Estimates put peak densities (million transistors per mm2) at: TSMC 16nm 28.2 TSMC 10nm 52.5 TSMC N7 is 91 TSMC N5 is ~171 TSMC N3 is ~290 Intel 14nm 37.5 Intel 10nm 101 Intel 7nm ~200-250 Note that these are peaks. Critical logic isn't made at peak densities.

• @handleym99 Maynard Handley on x

  @IanCutress Everything about M1/A14 looks like the priority (of course!) was getting x86/mac aspects working. Core and GPU look VERY similar. Probably no effort made to optimize for the new process. 2/

• @iancutress @iancutress on x

  @handleym99 Check the rest of the comments. Some of Apple/Huawei numbers were considered very suspect

• @thekanter David Kanter on x

  Most people don't grasp Intel's conservativism about new litho and correct EUV choice. EUV was absolutely not ready for HVM in 2017 for 10nm, nor in 2019 for 7nm. Unfortunately it turns out that 10nm wasn't ready for HVM for in 2017 or 2019 either. 1/N

• @migueldeicaza Miguel de Icaza on x

  My video review of the Apple M1, AMD Ryzen and Intel's latest Xeons: https://twitter.com/...

• @handleym99 Maynard Handley on x

  @IanCutress BUT Apple and Huawei chips up to 7nm were > 90% of TSMC max. Intel (even Atom) were1/3 to 1/2. TSMC is meaningful in products, INTC number is not. 5nm looks weird but we only have one data (Apple) and Apple was in a rush with M1/A14 1/