Új hozzászólás Aktív témák

• #6007 S_x96x_S addikt Petykemano #6005

  Új Válasz 2021-12-03 07:51:26 #6007

  Új hozzászólás

  Összes hozzászólása itt Válaszok az összes hozzászólására itt Válaszok erre a hozzászólásra

  Privát üzenet küldése

  

  S_x96x_S

  addikt

  válasz Petykemano #6005 üzenetére

  A fogyasztás és a TCO egyre fontosabb ..
  és itt az X86 -os irány - hátrányban van.

  https://semianalysis.substack.com/p/amazon-graviton-3-uses-chiplets-and

  """
  "Graviton3 is of course the star of the show. Amazon jumped forward as the first to many technologies on the server CPU side. They are using a chiplet design with 7 different dies. What stands out is that they are packaged using advanced packaging. The size of the ubumps connecting each die are <55um whereas every CPU from Intel and AMD is still at >100um. Intel and AMD only catch up with their next generation CPUs. This enables a design where IO is disaggregated from the CPU without ballooning in power budget. AMD IO dies on Rome and Milan server CPUs take as much as 100W. This 100W eats into the power budget for the cores, and cannot be geared towards computations. Graviton achieves 50% higher memory bandwidth than AMD Milan and PCIe 5.0 connectivity while keeping the entire CPU power consumption in that same ~100W range."
  ""

  """
  While x86 CPU vendors will maintain their peak performance per CPU lead, Intel and AMD are ignoring the more important battle. That battle is over total cost of ownership (TCO) per unit of compute on a server and rack level for generalized CPUs. Commoditization is here for the CPU market, and even if Intel's and AMD's individual core design is markedly better, it won't change the equation. Intel and AMD are hyper focusing on certain aspects, which make them miss crucial factors in system level design such as peak power being too high, density being too low, and clock speeds being pushed too far.

  Graviton3 should be making Intel and AMD executive’s quiver. In fact, all merchant silicon vendors should be terrified because Microsoft, Facebook, Google, and the major Chinese players want to replicate this vertical integration across networking, CPU, SSD, AI inference, and AI training. This basket of hyperscale firms are growing much faster than rest of the market, and they are swallowing up computing dollars spent like veracious beasts. Tech monopolies are going vertical, and there doesn't seem to be much being done to stop this long term tsunami.
  """

Új hozzászólás Aktív témák