3 ms·

I'd love to see citations on both rphlx's and wmf's claims. Not trying to sound mean - I'm honestly interested in the future of low-power computing and I kind

• pieter 17y ago
  Here is one link: http://perspectives.mvdirona.com/2009/09/16/ARMCortexA9SMPDesignAnnounced.aspx http://perspectives.mvdirona.com/2009/09/16/ARMCortexA9SMPDe... Also Anandtech on the Cortex A9: http://www.anandtech.com/gadgets/showdoc.aspx?i=3714&p=3 http://www.anandtech.com/gadgets/showdoc.aspx?i=3714&p=3 and the pages around that.
• rphlx 17y ago
  Here is my reasoning -- An Atom Z530 1.6GHz core has almost exactly 2X the Dhrystone MIPS of a 1.0GHz ARMv5TE-based network SoC (I actually ran it; those are not bogus vendor marketing #s). A9 probably has much higher IPC than ARMv5 -- but I doubt it's high enough to beat an Atom running at 60% higher frequency; that is a hell of a gap to close architecturally. I also suspect Atom has higher DRAM bandwidth; the A4 pkg looks a bit small for x64 DRAM. Finally, A4 may or may not have more than 1 core; Atom definitely has 2. I am talking pure performance, not performance per watt, which is not interesting to me in this situation. I can almost always find AC if I'm using something iPad-sized.