Geekbench 4

By: Yoav (no_drink.delete@this.gmail.com), July 29, 2016 11:22 pm
Room: Moderated Discussions
Yoav (no_drink.delete@this.gmail.com) on July 29, 2016 11:22 pm wrote:
>
> It jumped to my eyes also.
> thermal effects can make the testing a bit harder, and delays alone won;t solve this.
> The main issue is what do you want to test and how representative your workload is for that test?
>
> For example, if you want to test the sustain performance of a chip, by your workload
> is short, then you want to remove the benefit of starting from a cool system.
>
> I saw it very clearly when testing graphics - short benchmmarks (~30 seconds) would get very high
> scores, that would not be representative of actual frame rates for users who play an hour.
>
> Doug S (foo.delete@this.bar.bar) on July 29, 2016 7:59 pm wrote:
> > John Poole (john.delete@this.primatelabs.com) on July 29, 2016 11:57 am wrote:
> > > Doug S (foo.delete@this.bar.bar) on July 29, 2016 9:08 am wrote:
> > > > anon (anon.delete@this.anon.com) on July 28, 2016 10:58 pm wrote:
> > > > > Can you then provide a description of the benchmark here or somewhere that doesn't require sign up?
> > > >
> > > >
> > > > Agreed. If you have to sign up and participate in the beta just to learn what components are in the
> > > > benchmark, then the odds of it being discussed here are pretty low. RWT is probably one of the best
> > > > places to find people who could make good suggestions and provide constructive criticism. We certainly
> > > > have provided a lot of criticism (some of it even constructive :)) about Geekbench in the past, though
> > > > it isn't as though other benchmarks like SPEC and TPC have escaped criticism here either.
> > > >
> > > > Obviously there is still a need for testing as descriptions of the benchmark
> > > > components can only take you so far, comparing results you get on different
> > > > hardware is also necessary to see how it really works is also needed.
> > > >
> > > > But the starting point should be "what is in it, and how
> > > > much will each component figure into the overall score?"
> > >
> > > Sure thing. Here's some preliminary documentation on the CPU workloads included
> > > in Geekbench 4: http://www.primatelabs.com/beta/v4-cpu-workloads.pdf
> >
> >
> > On first glance, this looks like it will be a major improvement over previous versions of Geekbench!
> > I'll probably have more comments later, but the first thing that jumped out at me were the delays inserted
> > in the tests to give the CPU some time to cool down. While I think that's an excellent idea, I hope
> > there will also be a mode that lets you do the same test without the delay. It would provide an idea
> > of how quickly thermal effects cause the CPU to throttle or drop down from turbo mode.
> >
> > On PCs that's less of a concern, because you can always improve cooling if
> > you consider it an issue, but on phones there's really no way around it.
>
>

< Previous Post in ThreadNext Post in Thread >
TopicPosted ByDate
Geekbench 4John Poole2016/07/28 08:12 PM
  Geekbench 4anon2016/07/28 10:58 PM
    Geekbench 4Doug S2016/07/29 09:08 AM
      Geekbench 4John Poole2016/07/29 11:57 AM
        Geekbench 4someone2016/07/29 12:58 PM
        Geekbench 4Linus Torvalds2016/07/29 01:01 PM
          Geekbench 4John Poole2016/07/29 01:51 PM
            Geekbench 4Linus Torvalds2016/07/29 02:28 PM
        Geekbench 4Doug S2016/07/29 07:59 PM
          Geekbench 4Yoav2016/07/29 11:22 PM
            Geekbench 4Yoav2016/07/29 11:22 PM
        Geekbench 4anon2016/07/29 11:03 PM
          Geekbench 4Doug S2016/07/30 10:06 AM
            Geekbench 4Gabriele Svelto2016/07/30 12:49 PM
            Geekbench 4Maynard Handley2016/07/30 05:45 PM
        libjpeg, LLVM and sqliteGabriele Svelto2016/07/29 11:26 PM
        Geekbench 4none2016/07/31 10:48 AM
        Xcode7( llvm3.7?) vs Clang 3.8 xx2016/08/11 11:18 PM
Reply to this Topic
Name:
Email:
Topic:
Body: No Text
How do you spell green?