interesting quote

[Don Shepherd]

"If your program has no loops, it probably should be relegated to a less expensive machine than the (IBM) 650."

This quote is from the book, published in 1958, "Programming the IBM 650 Magnetic Drum Computer and Data-Processing Machine" by Richard V. Andree.

Is this as true today as it was in 1958?

[Maximilian Hohmann]

Hello!

(05-30-2021 09:13 AM)Don Shepherd Wrote:  Is this as true today as it was in 1958?

Personally, I would say: luckiliy not. At least as far as 99,9% of applications are concerned, storage and processing speed are virtually unlimited.

But if one designs ultra high performance embedded stuff (e.g. GPS-/Starlink-satellite communications, multi-megapixel-in-camera video processing, G5 mobile communication base transmitting stations, ...) then probably thinking twice about runtime efficiency will make a lot of difference.

Regards
Max

[Dave Britten]

It does make a good observation about choosing the right platform for the job so as not to grossly overspend on the work that needs to be done. You can frequently fit some simple jobs into an existing system that was meant for more complex work - and it often makes good sense to maximize an existing investment in this way - but if you're scoping a new system, or running out of capacity on the current one, don't overspend on what you need to accommodate the planned work.

This has become particularly important in cloud application environments that tend to be billed on resource consumption, and at vastly different rates for different types and tiers of services, not unlike the time sharing systems that this quote was referring to. For example I could set up an Azure SSIS runtime for $600/mo to run a job that simply extracts data from one system and delivers it to another, or I could do it for about $20/mo with a simple Azure Data Factory pipeline.

Stated another way, don't buy a 42S when all you need is a 20S.

[toml_12953]

(05-30-2021 01:28 PM)Dave Britten Wrote:  IYou can frequently fit some simple jobs into an existing system that was meant for more complex work - and it often makes good sense to maximize an existing investment in this way - but if you're scoping a new system, or running out of capacity on the current one, don't overspend on what you need to accommodate the planned work.

Stated another way, don't buy a 42S when all you need is a 20S.

Don't you know that applications are like a gas: They expand to fill whatever container you put them in!

[john gustaf stebbins]

(05-30-2021 09:13 AM)Don Shepherd Wrote:  "If your program has no loops, it probably should be relegated to a less expensive machine than the (IBM) 650."

This quote is from the book, published in 1958, "Programming the IBM 650 Magnetic Drum Computer and Data-Processing Machine" by Richard V. Andree.

Is this as true today as it was in 1958?

A bit of quick research indicates that an IBM 650 set-up rented from IBM for the equivalent of more than $13,000 per month in today's dollars. I'd say today you should have at least two loops to justify that kind of capital expense.

[Ren]

(05-30-2021 07:00 PM)john gustaf stebbins Wrote:  A bit of quick research indicates that an IBM 650 set-up rented from IBM for the equivalent of more than $13,000 per month in today's dollars. I'd say today you should have at least two loops to justify that kind of capital expense.

You are off by an order of magnitude, it would take 20 loops for 650 to justify $13,000.
B^)

[Ren]

(05-30-2021 01:28 PM)Dave Britten Wrote:  Stated another way, don't buy a 42S when all you need is a 20S.

That's like saying
"Why buy a Ferrari when a Yaris will get you there?"
B^)