… let’s hot wire the chip


My dad worked at IBM in Essex Junction, Vermont in the late sixties/early 70s. He was developing a dynamic RAM chip which was a new technology. He says that university researchers weren’t working on a dynamic RAM chip at the time so it was cutting edge hardware. The chip that he designed had a whopping 32kbits of memory–32,768 bits. His patent is US Patent 3,811,076.

The process they were using was originally called SAMOS for self-aligned metal oxide semiconductor, but I understand that later the acronym had a different meaning.

I just learned a new story about the project that I think is pretty astounding.

The first silicon that they were testing had a power trace that didn’t go to everywhere that it was needed. My dad had left out a power connection in the layout. Rather than go back to the drawing board and make a new wafer, they simply put a jumper connecting the two sides of the power network together with a couple of probes. It was good enough for them to complete the testing.

That is so clever to me. I’m sure there are very few stories of a chip design flaw that was worked around by hot wiring the chip.

It was possible because the dimensions of circuit components were much, much bigger than they are now. Also, the chip layout didn’t have as many layers as current circuits.

What an awesome engineering stunt!

Original image: 64-bit Chip. By Steve Jurvetson [Image license]

Transcending tricksy reading

I was just glancing over this paragraph and my vision played a trick on me. It comes from IUPUI’s website.

The Office of the Vice President for Information Technology has steering committees in place to help develop strategies and establish priorities for shared administrative and academic systems. The Oncourse Priorities Committee has specific responsibilities for ensuring thoughtful and timely decisions regarding priorities for Oncourse, IU’s online collaboration and learning system, and for recommending policies to guide its effective use. In light of next.iu.edu, the OPC will also provide guidance to the Learning Technologies Steering Committee regarding any future selection of a new learning management system for Indiana University. Below is a list of current committee members:

After glancing at the document, I was certain that I saw the word “transcend” somewhere in it. When I looked it over repeatedly, I found that “transcend” was not there at all.

Perhaps this is along the same line as the mistakes that can be present in eye-witness testimony. One can remember an event that might have taken only a fraction of a second. Over time the event will gradually shift from being a movie to a short story. This short story can be adjusted by people asking questions or reinforcing story facets with feedback and positive reinforcement. Talking with another witness can confound ones memories with the other person’s memories so that both are wrong.

One specific example of this is in police investigations. The police (for the most part) don’t try to manipulate the witness, but they have an effect by asking some questions and not others. Also, they can give reinforcement to certain details with their comments by thanking the witness for their information. In some situations, such as with children, these adjustments are very hard to avoid.

The movie and short story feel equally true, but after a while they won’t match the actual events that happened. As the short story then turns into a sound bite, the eye witness can be wrong, sometimes in catastrophic ways for the plaintiff defendant.