By: juanrga (nospam.delete@this.juanrga.com), November 10, 2014 8:00 am
Room: Moderated Discussions
David Kanter (dkanter.delete@this.realworldtech.com) on November 9, 2014 2:43 pm wrote:
> I really don't see HPC as being vulnerable at all. Single core performance matters,
> and there's a lot of ecosystem (e.g., infiniband) that just isn't there for ARM.
Supercomputing with commodity CPUs: are mobile SoCs ready for HPC?
gives beautiful graphics with the performance evolution of vector processors vs alpha/HP-PA/IBM... vs x86 and vs ARM-based processors. The replaced processors always began at higher performance levels but are replaced by processors that are characterized by faster evolution, closing the performance gap very fast. The faster evolution has its roots on a larger volume sales [#].
Then the authors write:
[#] AMD uses similar arguments for predicting that ARM will win over x86

> I really don't see HPC as being vulnerable at all. Single core performance matters,
> and there's a lot of ecosystem (e.g., infiniband) that just isn't there for ARM.
Supercomputing with commodity CPUs: are mobile SoCs ready for HPC?
gives beautiful graphics with the performance evolution of vector processors vs alpha/HP-PA/IBM... vs x86 and vs ARM-based processors. The replaced processors always began at higher performance levels but are replaced by processors that are characterized by faster evolution, closing the performance gap very fast. The faster evolution has its roots on a larger volume sales [#].
Then the authors write:
Given the trend discussed above, it is reasonable to consider whether the same market forces that replaced vectors with RISC microprocessors, and RISC processors with x86 processors, will replace x86 processors with mobile phone processors.
[#] AMD uses similar arguments for predicting that ARM will win over x86
