What does the percentage mentioned in brackets refer to? Does ic miss (24.3%) means 24.3 % of total execution time was spent on inline cache misses? But surprisingly these percentages do not add to 100%.
Also, is it possible to see JITed code, where I can see how the code looks with inline caches? I am aware of --trace_ic which gives out the information about LOADIC and STOREIC. But I want to see how the original code looks after introducing code for inline caching. On Friday, 18 September 2015 18:44:43 UTC+5:30, Jakob Kummerow wrote: > > I don't think it's possible to measure property access times directly. One > reason is that a single property access is much, much faster than getting a > timestamp from the operating system. > > So your best bet is probably to measure it indirectly: take a long-running > loop, add a single property access in such a way that (1) it can't be > optimized away by the compiler and (2) it doesn't introduce additional > operations, which isn't easy to achieve at the same time, and measure how > much longer the overall loop takes than it did before. > > Of course that would be a very artificial micro-benchmark, and would only > measure one very specific situation. In the general case, property accesses > can take *very* different amounts of time, though: in the best case, > they're a single machine instruction; in the worst case, they're thousands > of machine instructions. It all depends on so many things... > > I can't think of a way to take an existing program and accurately measure > what fraction of its running time is spent on property accesses. > > On Fri, Sep 18, 2015 at 1:13 PM, dmonji <monika...@gmail.com <javascript:> > > wrote: > >> Is there any way to measure it? >> >> I tried to instrument the JavaScript code, and summed up the time between >> "before access" and "after access". But I am not very sure if this is >> the right way to measure the time required for property accesses. Any >> thoughts on this? >> >> On Friday, 18 September 2015 16:30:42 UTC+5:30, Jakob Kummerow wrote: >>> >>> Property access time is not measured separately. >>> >>> On Fri, Sep 18, 2015 at 12:56 PM, dmonji <monika...@gmail.com> wrote: >>> >>>> I want to compute the proportion of time spent on property accesses in >>>> JavaScript programs. I am aware of >>>> code.com/svn/branches/bleeding_edge/tools/profviz/profviz.html >>>> <http://profile.com/> which is a cpu profiler. >>>> >>>> But it is not clear what exactly does "execution" refer to since "ic >>>> cache" is shown seperately? >>>> >>>> Does "execution" only include "Access time of properties"? >>>> >>>> -- >>>> -- >>>> v8-users mailing list >>>> v8-u...@googlegroups.com >>>> http://groups.google.com/group/v8-users >>>> --- >>>> You received this message because you are subscribed to the Google >>>> Groups "v8-users" group. >>>> To unsubscribe from this group and stop receiving emails from it, send >>>> an email to v8-users+u...@googlegroups.com. >>>> For more options, visit https://groups.google.com/d/optout. >>>> >>> >>> -- >> -- >> v8-users mailing list >> v8-u...@googlegroups.com <javascript:> >> http://groups.google.com/group/v8-users >> --- >> You received this message because you are subscribed to the Google Groups >> "v8-users" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to v8-users+u...@googlegroups.com <javascript:>. >> For more options, visit https://groups.google.com/d/optout. >> > > -- -- v8-users mailing list v8-users@googlegroups.com http://groups.google.com/group/v8-users --- You received this message because you are subscribed to the Google Groups "v8-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to v8-users+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
