From: johnrreagan@earthlink.net   
      
   On 2/24/2025 4:27 PM, John Dallman wrote:   
   > In article , seaohveh@hoffmanlabs.invalid   
   > (Stephen Hoffman) wrote:   
   >   
   >> The Itanium compiler optimizer just doesn't (and can't) know enough   
   >> about the system memory state, yes. Among other (no pun intended)   
   >> issues.   
   >>   
   >> The attempt to address that included providing run-time feedback   
   >> into the executables; providing post-link, post-execution tuning.   
   >> (Caliper / Atom / OM / etc.)   
   >   
   > "Attempt" is about right.   
   >   
   > I did several years porting work to Itanium. I tried run-time feedback   
   > zero times: doing the link of the instrumented build took over an hour,   
   > up from about a minute, because it was doing all the code generation at   
   > link time.   
   >   
   > The claim was "you only do this for the build you'll ship." My response   
   > was "The compiler is so immature that I'm reporting new bugs every week,   
   > and you want me to give the compiler new and difficult challenges?"   
   >   
   > I never heard of anyone who got anywhere with profile-guided optimisation   
   > on Itanium. Have you?   
   >   
   > John   
   Actually, the NonStop Itanium kernel is built using PGO.  Apparently,   
   for their test workload of transactions, the savings was a measurable   
   "few" percent.  [I think in the 3-5% range but I'm not sure anymore]   
      
   --- SoupGate-DOS v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)