Deep Inside Rochester

The title of this post came from a statement made by Frank Soltis in a recent article in TECHWORLD called "Q&A: Power chip inventor speaks out". What Frank actually said is this:

Today, most ISV-developed applications for the System i5 are multithreaded parallel applications.

Although I agree with Frank that "the System i5 and its predecessors have always been designed to run multiple applications in parallel," I don't agree that most ISV-developed applications are multithreaded, particularly because the use of threads in RPG is not prevalent.

However, will this statement become true eventually? Today's operating systems and ISVs will be pressured to fully exploit the capabilities of the compute power present in multi-core systems, driving the software industry towards greater use of multi-threading. In the article, Frank goes on to say that "writing parallel applications will in the future be the only way to get increased performance. There are some tools that can help, but much of the effort will come back to the way in which programmers design their applications." Yes, I think frank has identified a trend.

Btw, Frank works out of Rochester, Minnesota, not Rochester, Minneapolis as asserted in the article.

Real World Technologies (RWT) has an article on the IBM eCLipz project, "An eCLipz Looms on the Horizon" by David Kanter, which includes more discussion on POWER6.

In addition, the article is discussed here on Ars Technica.

The Kanter article briefly mentions iSeries and oS/400 as they relate to the economics of convergence and economies of scale. The author writes that "most platforms that did survive, such as OpenVMS, now share hardware with UNIX operating systems to better amortize the development costs. IBM took this same approach with the iSeries; OS/400 runs on a basically unmodified pSeries systems based on the POWER5."

I agree with the observation, but as many [AS/400, iSeries, i5] customers know, especially those that read The Four Hundred newsletter, the sharing of microprocessor technology between i and p goes back well before POWER5. In fact, it goes back to the "Apache" processor introduced in 1997. Sharing continued with the Star family of processors from 1998 to 2000 and the POWER4 processor in 2001.

In RPG, the "H" operation extender is used to indicate whether the contents of the result field are to be half adjusted (that is, rounded).

Applying terminology from the Clive Maxfield article to RPG, we can see that
rounding in RPG uses a symmetric implementation of round-half-up. Here is a description of rounding from Chapter 9, Calculation Specifications, of the RPG/400 Reference: "Half-adjusting is done by adding 5 (-5 if the field is negative) one position to the right of the last specified decimal position in the result field."

If the operation extender is not specified, the default is to round-towards-zero for decimal numbers. That is, the digits to the right of the last specified decimal position are truncated.

In the following code example, the value copied to r1 will be rounded, but the value copied to t2 will not:

     D t1              s              6p 2
     D t2              s              6p 1
     D r1              s              6p 1
      /free
        t1  = 0003.49;
        eval    t2 = t1;  // result is 3.4
        eval(h) r1 = t1;  // result is 3.5

Slashdot.org has a pointer to an interesting article on rounding algorithms, including those applied to decimal arithmetic. Here is the link to the article:

Clive Maxfield, "An introduction to different rounding algorithms," http://www.pldesignline.com/howto/showArticle.jhtml;?articleID=175801189

Given that Rochester hasn't seen the sun for at least two weeks, I think it's fair to say that the weather is simply cloudy here. But a recent regional forecast in the local paper (provided by AccuWeather) has a different way of describing it:

Tonight: "Clouds."
Thursday: "Cloudy most of the time."
Friday: "... considerable cloudiness."
Saturday: "Rather cloudy."
Sunday: "Mostly Cloudy"
Monday: "Overcast ..."