← Back to context

Comment by einpoklum

4 days ago

If I try to cut through the hype, it seems the main features of this processor, or rather processor + memory controller + system architecture, is < 100 ns for accessing anything in system memory and 6 GB/sec for each of a large-ish number of cores, so a (much?) higher overall bandwidth than what we would see in a comparable Intel x86_64 machine.

Am I right or am I misunderstanding?

It's the same memory bandwidth as Intel and moderately higher than AMD.

  • Even if you get the 136 cores or whatever?

    • AMD old CPUs (to be replaced by the end of the year) have 192 cores per socket, where each core is significantly faster than Neoverse V3.

      The latest Intel server CPU, Clearwater Forest, uses Darkmont cores that have approximately the same performance, cost and power consumption as Neoverse V3, but Intel provides 288 cores per socket and 576 cores per board.

      Even supposing that Intel Xeons would be used in relatively big 2U servers, that still provides at least 50% more cores per rack than these new Arm AGI CPUs.

      The claim of Arm that they provide better performance per rack is false. They must have compared their new CPUs with some antique Intel Granite Rapids Xeon CPUs, instead of comparing with state-of-the-art Intel and AMD CPUs, which offer much more performance per rack than the new Arm AGI.