Does the compiler you are using have an option for profile guided optimization? This can potentially realize very significant performance gains on SIMH.
-Henry On 15 September 2016 at 16:18, gérard Calliet <[email protected]> wrote: > > > Le 15/09/2016 à 20:04, Mark Pizzolato a écrit : > > Patrick's suggestion is a great one. Please explore that. > > I think so. It'll take some time, because I'm going now to USA for VMS > bootcamp. I'll have talks there with VMS enthousiasts, and all I ear in > this list will be tried. > Go on suggest. And if you think about tools to get figures, I'll take. > > Additionally, you're not necessarily making a fair comparison. To be > fair, you should compare the I2 system to an X86 system that was shipping > when that I2 system first shipped. I suspect you'll see closer results. > > I understand, but what is strange is when I have been able to use i4 > (network did work at that time) the performance results where as bad as > with i2. So I was thinking there is some structural issue with Itanium or > with OpenVMS. > > - Mark > > On Sep 15, 2016 10:22 AM, Patrick Finnegan <[email protected]> > <[email protected]> wrote: > > > > > > For #2, could you try comparing performance on VMS vs on Linux running > as the host OS on the same Itanium system? Narrowing it down to one > variable at a time makes it easier to test differences. > > > > You should be able to use a live CD on the Itanium system to test > without reinstalling the system. > > > > Pat > > > > On Thu, Sep 15, 2016 at 11:50 AM, gérard Calliet < > [email protected]> wrote: > > >> > > >> Hello, > >> > >> simh on OpenVMS again. > >> > >> 1) with the help of an Lan expert of VSI, I could use pcap on VMS on an > Itanium i4... some time and it is again impossible. I will meet the VSI > expert on OpenVMS bootcamp in Boston next week, anf I hope we will address > this issue again. Any idea welcomed, of course. (It seems PCAP on VMS > does'nt like larger frame with i4, because it is the only diference I can > see between i2 and i4 Itanium servers, and it works on i2). > >> > >> 2) there is a very bad performance issue: same batch on a windows10 / > i7 platform six times faster than on an i2 / OpenVMS platform. I don't > think it's a normal issue of windows or x86 being better than OpenVMS / > italium i2. Something goes wrong somewhere. I gave a huge working set to > the host process, and it is not doing paging, I tried Decram to avoid > physical I/O, asynchronous I/O is enabled for simh (and the simh > configuration is the same on windows and on OpenVMS). > >> > >> Any help will be welcomed: ideas, tools for tracking,... > >> > >> Best regards, > >> > >> Gérard Calliet > >> > >> > >> > >> --- > >> L'absence de virus dans ce courrier électronique a été vérifiée par le > logiciel antivirus Avast. > >> https://www.avast.com/antivirus <https://www.avast.com/antivirus> > >> > >> _______________________________________________ > >> Simh mailing list > >> [email protected] <[email protected]> > >> http://mailman.trailing-edge.com/mailman/listinfo/simh > <http://mailman.trailing-edge.com/mailman/listinfo/simh> > > > > > > > > > ------------------------------ > [image: Avast logo] > <https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient> > > L'absence de virus dans ce courrier électronique a été vérifiée par le > logiciel antivirus Avast. > www.avast.com > <https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient> > > > _______________________________________________ > Simh mailing list > [email protected] > http://mailman.trailing-edge.com/mailman/listinfo/simh >
_______________________________________________ Simh mailing list [email protected] http://mailman.trailing-edge.com/mailman/listinfo/simh
