Failing to cross the great divide

What this suggests is that the wikipedia authors he cites both share and shape his ignorance of, and correspondingly deep contempt for, non x86 processors.

Last week's series on the tendency of political and technological groups to barricade themselves among their friends drew this example - by "defconvegas" in reply to "Mactel's carbon footprint"

 

I wonder who came up with ACP and ACPI

ACP is a standard that was jointly developed by Intel and Microsoft

ACPI is a standard that was jointly developed by Intel, Microsoft, Toshiba, HP and Pheonix BIOS.

In case if you are wondering what ACP and ACPI are, they are the industry standard interface for power conservation

http://www.acpi.info/

I dont see any IBM, SUN, RedHat involved anywhere in any power conservation.

To top it off Windows offers the best power management features of any OS out there.

I dont see any Linux providing DECENT power conservation at all eg does not support hibernation.

Below is a link that shows a chart which displays the power consumption of CPU's and I think if you can read English you'll see x86 use less power than PowerPC. http://en.wikipedia.org/wiki/CPU_power_dissipation

Clearly shows that *nix people are just FUDsters who dont do anything useful but just sit around and complain about others and worse off lie about others.

The old saying that "birds of a feather flock together" applies here. Defcon's comments illustrate one of the subtler, but nastier, side-effects of that whole "great divide" thing - people with one sided views selecting compatible information sources.

He cites two sources for his opinions here: a wikipedia article and a Wintel power management consortium. To take the second one first, here's what that wikipedia entry says about the power consumption of what it (incorrectly) calls "IBM Processors" - starting with Freescale's MPC8641D:

 

IBM processors

PowerPC

  • Dual-core PowerPC MPC8641D 90 nm, 2 GHz, 1.2 V, 15-25 W
  • PowerPC 750FX 0.13 µm, 900 MHz, 1.2 V, 3.6 W
  • PowerPC 750CXe 0.18 µm, 600 MHz, 1.8 V, 6 W
  • PowerPC MGT560 0.20 µm, 56 MHz 2.7 V, .5 W (Performance=56MIPS)
  • PowerPC 440GX, 800 MHz, 4.5 W
  • PowerPC 970, 1.8 GHz, 1.3 V, 42 W
  • PowerPC 7400e, 1.0 GHz, 1.6 V, 30 W

Notice first that the one I mentioned in the blog, the MPC8641D is listed here at a rated 15-25 watts for the 2Ghz version. This is correct according to the original Freescale announcement but later data, including the Freescale production announcement quoted in the blog, put the actual at 15 watts for the dual core 1.5Ghz release.

Notice next that only two of the PPC chips listed here were ever used by Apple, the PPC7400 series in 1998 and the PPC970 1.8Ghz in 2002.

And, most importantly, if you check out the page you'll find that the authors limit their interest in PPC CPUs to the seven shown above, but go into loving detail on several hundred Intel chips - including 15 celerons, 6 Itanium 2s, 26 different Xeons, and this list of the chips Apple actually uses:

 

Intel Core Duo

 

  • T2600 2.16 GHz 31 W
  • T2500 2.0 GHz 31 W
  • T2400 1.83 GHz 31 W
  • T2300 1.66 GHz 31 W
  • T2050 1.60 GHz 31 W

Intel Core 2 Duo

 

  • E6700 2.66 GHz 65 W
  • E6600 2.40 GHz 65 W
  • E6400 2.13 GHz 65 W
  • E6300 1.86 GHz 65 W
  • E4300 1.80 GHz 65 W
  • X6800 2.93 GHz 75 W

Since, in reality, the PPC community is significantly larger and more diverse than the x86 one, this imbalance - along with the detailed AMD numbers and the apparently random and anachronistic choices the authors made on which "IBM" processors to list - suggests both a deep contempt for, and utter ignorance of, non x86 processors.

In other words, the source defcon chooses to cite provides an exact match to his prejudices by eliding exactly the information he would need to see if he were to have any chance at all of escaping his community limits.

His power use citation is similarly one sided. The organization he cites as the authority for his view that power management is done best on Windows is a Windows organization.

Look at the information provided by the ACPI, and power management originated with Intel and Microsoft - but, in reality, these concerns have been around for a long time. Thus NCR's tower machines, running AT&T Unix in the mid eighties, would shut down everything except RAM during temperature spikes or no load conditions, Solaris 2.5.1 came with default power management broadly similar to what Microsoft now offers, and Apple's original PPC603e design included power management capabilities broadly similar to what Intel offers today.

The point here isn't that defcon was wrong, but that he clearly tried to get it right on both counts, only to find himself betrayed by the closed and crippled nature of his community sources. Thus he did what he's supposed to do when arguing the facts: checked his beliefs against apparently authorative third parties and cited them for us to look at - but those sites misled him, and that's the tragedy writing QED to my fulminations from last week.