Business vs. Progress

It is an easy assumption to make that technological development drives a succesful business. We are being told that economic growth is essential for our lives, and that growth is made possible by faster and more efficient technology. However, the reality appears somewhat different. New and exciting technologies are being fought back by old and established businesses.


As a computer geek, my pet peeve is x86, and why it is the dominant player in workstations and servers despite its technical shortcomings. My simplistic theory is this:

  1. Most people run Windows, which is hopelessly stuck in x86. Perhaps not so much as the OS, but as the ecosystem of closed third-party software.
  2. Thus there is a huge market for x86, meaning
    1. The machines are widely available
    2. The machines are inexpensive
    3. A lot of development resources are poured into x86, thus making it "the dead horse beaten to the speed of light", as someone on Slashdot put it.
  3. As a result, even non-Windows users and developers find x86 practical.

So we end up with abominations like closed x86 software for Linux. The OS itself runs on pretty much any hardware platform, so we could choose the hardware on a technical basis. Instead, we are forced to choose x86 due to these business reasons. For example, scientific computing with Linux suffers from the Windows monoculture this way.

As if this were not bad enough for other architectures, businesses manage to screw them in various other ways.


G4/G5 in Apple computers

It is generally said that Apple moved from PPC to Intel due to the huge power consumption of the G5 processor. While the G4 was cool enough for laptops, its successor was never seen in those. So perhaps there was something inherently wrong about the architecture, and Intel was really the only option.

However, there was a similar change going on within Intel processors. Pentium 3 was pretty cool, but the P4 was a monster toaster. While the P4 occasionally made its way into laptops, the real future of laptops was with the Pentium M. It was a derivative of the P3, with the fast external bus and other tidbits borrowed from the P4. Later, two Pentium Ms were put onto a single die, and the result was called Core Duo. Further developments led to Core 2, which was also made in desktop versions, and the P4 officially became a dead end.

The natural parallel in the PPC world is that the G5 should have been phased out, and the G4 developed further. Like the P3, the G4 suffers from a slow external bus, so there is obvious room for improvement. Another well-known development is a process shrink, which further facilitates reduced power consumption, higher clock speeds, and multicore dies.

Why this did not happen is probably due to the smaller market of PowerPC. Unlike Intel, the AIM alliance could not maintain multiple parallel lines of CPU development. But perhaps another player could do that?


Apparently, P. A. Semi did just that with their PWRficient. A dual-core 64-bit PPC that consumes a few watts at 1 GHz. For one reason or another, this did not become a big enough business that you could actually buy one for yourself. After all, everyone wants x86. It would have been the perfect laptop CPU, and interestingly enough it was Apple that bought the company in 2008. But so far, little of interest has come out of it. The iPad uses an ARM CPU, for example.

Cell BE and PS3 Linux

If someone claims that PowerPC is dead, you can point them out that all of the current "7th" generation game consoles use it. All of Wii, PS3, and Xbox 360 — even Microsoft uses it. For a computer geek this would be exciting, but these are closed gaming machines. Naturally, there have been attempts to tap into this power and actually use the hardware you own, but so far these hacks have not been huge successes.

Except, to some extent, the PS3. It was originally marketed with the capability to install other operating systems, most notably Linux. Adding to the excitement, its CPU was probably the most interesting of the three. It was easy to imagine that you could learn Cell development on a PS3, and move on to big iron with more capable versions of the same CPU.

But it was pretty capable by itself, and for many research organizations, it was the most computing power you could buy per dollar. Until Sony decided to remove the functionality. Once again, a business decision working against technological progress.

The official explanation is that Linux was used for game piracy, even though that is not what actually happened. The OtherOS functionality did not expose all of the hardware capabilities, notably the GPU, and some people tried to work around those limitations. This is probably what triggered the response; with the GPU available to hobbyists, people could play open-source games instead of paying Sony.

Of course, even the usual OtherOS usage is bad for business. Research organizations do not play games on their PS3 clusters, so they are not offsetting the real hardware price that is less than what they paid for the PS3. So perhaps OtherOS was a bad business decision to begin with.

Nevertheless, it feels both sad and stupid that there is plenty of wonderful technology out there, that you cannot use. If you buy yourself a game console, you cannot actually use the hardware any way you want. It is truly a waste of resources — you need another machine to use as a typical computer, but not because of any technical reason. In fact, some amount of time and money was spent on putting those unnecessary limitations there, so you pay more for getting less.

MIPS and Sicortex

In the same vein of saving resources, Sicortex realized you can make a whisper-quiet supercomputing cluster, if you can live without x86. The problem was again, not enough people wanted to buy these, because everyone wants x86. Even the supercomputing people, because a lot of "scientific" software is only available as closed x86 binaries.


Having a regular Linux distro on a tablet or a phone is a brilliant and simple idea. No need to reinvent the wheel, just tweak the user interface a little to suit a smaller, mobile device.

Perhaps some people at Nokia thought about this in the early 2000s, and the first Maemo device was released in 2005. For some weird reason or another, this device from a famous mobile phone manufacturer did not feature phone capabilities. Maemo devices only reached this point in 2009.

This was probably too good, so Maemo was scrapped in favour of Meego. I have no idea what Meego means in practice, since nothing practical has come out of the Nokia-Intel alliance. Except perhaps the idea of an x86 phone, which gives every self-respecting geek the shudders.

Nokia's open source stance was finally nailed on 2011-02-11 with the announcement of a Microsoft deal. Meego tablets may still be on their way, but it will be Windows for the "smart"phones. Of course, Finns already have a name for the alliance that begins with an "M" and ends with "okia".


Risto A. Paju