Performance analysis laughable

Article: Medfield, Intel's x86 Phone Chip
By: Linus Torvalds (torvalds.delete@this.linux-foundation.org), January 25, 2012 9:37 pm
Room: Moderated Discussions
tupper (no@thanks.com) on 1/25/12 wrote:
>
>Here is a research work done by UMich and ARM. How much sense does it make for the conclusion :

So I absolutely believe in the "tons of I$, ITLB and branch
prediction misses" thing.

Having done performance analysis of Spec and real desktop
applications (about ten years ago, so that's probably not
all that far from a smartphone), I can tell that Spec is
a set of incredibly well-behaved applications in general.

That includes gcc, which I obviously think is the only app
in the whole Spec suite that is worth even quoting numbers
for - the other ones are too effing broken, and much too
amenable to various trickery.

Graphical applications end up doing several layers of
indirection through libraries - something that you never
see in Spec. In fact, most Spec runs are outright lying,
in that they tend to flatten out even the one layer of
libraries there is - inlining a ton, and usually statically
linking the rest.

So spec is orders of magnitude better than the crazy toy
benchmarks the embedded world tends to historically use,
but spec isn't nearly gnarly enough to show what goes on
in a desktop app.

The paper you point to seems quite interesting from a quick
look. And it's really telling how the spec benchmarks show
basically no I$ component at all, but the interactive
ones do.

That said, there is one exception, and that one is not
shown in the paper. High-performance native games tend to
behave more like "traditional" loads than desktop loads.
Lots more tight loops etc. Of course, the actual graphics
stack is often quite big, and the 3D driver can get very
hard to gather data from.

Linus
< Previous Post in ThreadNext Post in Thread >
TopicPosted ByDate
Medfield article onlineDavid Kanter2012/01/23 02:51 PM
  server errorbakaneko2012/01/24 04:00 AM
    FixedDavid Kanter2012/01/24 05:02 AM
      FixedJoel2012/01/24 08:43 AM
      FixedRicardo B2012/01/24 12:25 PM
        FixedDavid Kanter2012/01/24 06:29 PM
      FixedGabriele Svelto2012/01/24 02:07 PM
        FixedDavid Kanter2012/01/24 06:30 PM
  Reference platform battery lifeDoug Siebert2012/01/24 03:03 PM
    standby timeFoo_2012/01/25 07:58 AM
      standby timeAnon2012/01/26 04:42 AM
        standby timeFoo_2012/01/26 05:02 AM
          standby timeDoug Siebert2012/01/26 01:39 PM
            standby timeAnon2012/01/26 02:22 PM
              standby timeanon2012/01/26 03:08 PM
                standby timeAnon2012/01/26 07:03 PM
                  standby timeanon2012/01/26 09:57 PM
                    standby timeanon2012/01/26 10:01 PM
                    standby timeAnon2012/01/27 10:32 PM
                standby timeDoug Siebert2012/01/27 03:15 PM
                  standby timeanon2012/01/27 03:41 PM
    Reference platform battery lifeDavid Kanter2012/01/27 11:09 AM
  Performance analysis laughableWilco2012/01/24 04:23 PM
    Performance analysis laughableDavid Kanter2012/01/24 06:19 PM
      Performance analysis laughableIntelUser20002012/01/24 08:30 PM
        Performance analysis laughableIntelUser20002012/01/24 08:32 PM
        Performance analysis laughableDavid Kanter2012/01/25 12:34 AM
          Performance analysis laughableIntelUser20002012/01/25 12:56 AM
            Performance analysis laughableDavid Kanter2012/01/25 03:07 AM
              Performance analysis laughableAlberto2012/01/25 01:54 PM
          Atom HT gainWilco2012/01/25 06:43 AM
            Atom HT gainIntelUser20002012/01/25 07:53 AM
              Atom HT gainnone2012/01/25 08:04 AM
                Atom HT gainIntelUser20002012/01/25 08:35 AM
            Atom HT gainFoo_2012/01/25 08:06 AM
      Performance analysis laughableWilco2012/01/24 09:21 PM
        Performance analysis laughableDavid Kanter2012/01/24 11:13 PM
          Performance analysis laughableWilco2012/01/25 05:30 AM
            Performance analysis laughablenone2012/01/25 07:14 AM
              Performance analysis laughableWilco2012/01/25 08:18 AM
                Performance analysis laughableobserver2012/01/26 05:17 AM
                  Performance analysis laughableWilco2012/01/26 07:25 AM
            Process numbersAlberto2012/01/26 10:29 AM
            Performance analysis laughableDavid Kanter2012/02/02 01:38 AM
          Performance analysis laughabletupper2012/01/25 05:27 PM
            Performance analysis laughableLinus Torvalds2012/01/25 09:37 PM
              Performance analysis laughableDoug Siebert2012/01/26 03:12 PM
  Medfield article onlineAndreas2012/01/25 04:10 AM
    Medfield article onlineAlberto2012/01/25 10:44 AM
    Medfield article onlineIntelUser20002012/01/25 11:24 AM
    Medfield article onlineDavid Kanter2012/01/25 10:58 PM
      Medfield article onlineDoug Siebert2012/01/26 02:20 PM
        Medfield article onlineEric2012/01/26 07:10 PM
          Medfield article onlineDoug Siebert2012/01/27 03:40 PM
  64-bitIngeneer2012/01/25 10:28 AM
    64-bitFoo_2012/01/25 11:23 AM
      64-bitIngeneer2012/01/25 03:34 PM
        64-bitUngo2012/01/25 05:08 PM
          64-bitEduardoS2012/01/26 01:55 PM
  Saltwell memcpySHK2012/01/26 03:41 AM
  Medfield WiFi & BluetoothRob Thorpe2012/01/26 04:09 AM
    Medfield WiFi & BluetoothDavid Kanter2012/01/27 06:54 PM
      Medfield WiFi & BluetoothRob Thorpe2012/01/28 03:22 PM
  Medfield article online (NT)Anil2012/01/26 06:57 PM
  Medfield article onlineAnil2012/01/26 07:11 PM
    Medfield article onlineMr. Camel2012/01/26 07:26 PM
    Medfield article onlinenone2012/01/27 02:41 AM
Reply to this Topic
Name:
Email:
Topic:
Body: No Text
How do you spell avocado?