From: user5857@newsgrouper.org.invalid   
      
   Terje Mathisen  posted:   
      
   > MitchAlsup wrote:   
   > >   
   > > anton@mips.complang.tuwien.ac.at (Anton Ertl) posted:   
   > >   
   > >> Bottom line: History-based branch prediction has won, any kind of   
   > >> delayed branches (including split-branch designs) turn out to be   
   > >> a bad idea.   
   > >   
   > > Or "Never bet against branch prediction".   
   >   
   > I have probably mentioned this before, once or twice, but I'm actually   
   > quite proud of the meeting I had with Intel Santa Clara in the spring of   
   > 1995:   
   >   
   > I had (accidentally) written the first public mention of the FDIV bug   
   > (on comp.sys.intel) in Oct 1994, then together with Cleve Moler of   
   > MathWorks/MatLab fame led the effort to develop a minimum cost sw   
   > workaround for the bug. (My code became part of all/most x86 compiler   
   > runtimes for the next few years.)   
   >   
   > Due to this Intel invited me to receive an early engineering prototype   
   > of the PentiumPro, together with an NDA-covered briefing about its   
   > architecture.   
   >   
   > Before the start of that briefing I suggested that I should start off on   
   > the blackboard by showing what I had been able to figure out on my own,   
   > then I proceeded to pretty much exactly cover every single feature on   
   > the cpu, with one glaring exception:   
   >   
   > Based on the useful but not great branch predictor on the Pentium I told   
   > them that I expected the P6 to employ eager execution, i.e execute both   
   > ways of one or two layers of branches, discarding the non-taken paths as   
   > the branch direction info became available.   
   >   
   > That's the point when they got to brag about how having a much, much   
   > better branch predictor was better both from a performance and a power   
   > viewpoint, since out of order execution could predict much deeper than   
   > any eager execution would have the resources for.   
      
   I remember you relating this story about 6-8 years ago.   
      
   > As you said: "Never bet against branch prediction".   
   >   
   > Terje   
   >   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)