Chapter 30

STEVE - Wednesday, November 29, 1961

We woke up earlier than usual. I was eager to get back to the machine and try my new test cases. We had a quick breakfast, and Sandy said, "I need to stop by my office for a few minutes. While I'm there, I'll see if I can catch Paul and give him an update".

"I think I'd like to go directly to the test cell and start punching these programs. You won't be long, will you?"

"No, I'll be there by the time Miguel and Aaron get in".

I went to the test cell and, my notes in hand, started punching my program cards. By the time I was done, Sandy had arrived, along with Miguel and Aaron. They started the machine, and I soon had the new programs assembled and ready to run. I ran all of my test cases, noting which ones failed and which ran correctly.

I announced, "OK, here's what I know now. It looks like the only Y branch code point associated with the failures is the one I identified yesterday. I wrote a similar case that used the same branch condition, but as a Z branch, and it didn't fail. However, there are three different byte logic operation code points that fail; all of them use only one input to that unit. What's next?"

Sandy said, "Nice work! Let's start scoping that end-op line on all of these cases, using the no refetch as a trigger. Steve, can you run them all again?"

"Sure. Let me know when you're ready".

Aaron and Miguel had the scope set up and I ran through the entire suite again, announcing the cases that would fail as they came up. Sandy exclaimed, "It's definitely something about these code points affecting the end-op signal. It's clean on all of the good tests, and it's bad on all of the failing cases, although there are some definite differences in the pulse waveform".

I snorted, "So much for binary logic! I thought either it was or it wasn't!"

Just then Sarah walked in. Aaron smiled, and said, "Welcome to the underworld of logic. When it becomes a question of electricity, lots of things can go wrong. Right, Sarah?"

She said, "Absolutely! You wouldn't believe how messy it was in the dark days of vacuum tube logic!"

Sandy said, "Let's see if we can figure out what's causing the different pulse shapes for those cases".

I said, "While you're doing that, I'd like to look at something. How many pulse different shapes are you seeing?"

Sandy replied, "I know of three, but there may be a fourth. Why?"

"I want to see what cases are associated with each shape. That may help".

"OK, run them again. Let's call the waveform shapes A, B, and C. I'll tell you which shape I see for each case, and you can make a table of cases and shapes".

When we got done, it became evident that each shape corresponded to one of the three byte logic unit control codes that were associated with failures, and I exclaimed, "Each of these shapes is linked to a particular byte logic unit control code!"

Sandy got very excited and exclaimed, "You've done it again! That will help us enormously! Miguel, Aaron, can you start scoping the byte logic unit control lines? Steve and I were in early this morning, and now we are going to grab a quick sandwich. When we get back you guys can get your lunch".

SANDY - Wednesday, November 29, 1961

I said to Steve, "Let's go!"

We quickly made our way to the cafeteria and picked up sandwiches, along with sodas and a couple of apples. We took them back to the test cell, and found that Chet had joined the group. Miguel and Aaron went to lunch, and Steve and I told Chet what we had found so far. He was quite interested when he found out that the 'no refetch' flag was involved, along with one specific Y branch field code point. While I finished my sandwich, he studied the logic diagrams for a while, making notes on a scratch pad.

Finally, he asked Steve, "Can you set these test cases to loop?"

"Yes, which one do you want to run first?"

"Start at the beginning. Eventually, I'm going to want to scope them all".

Steve and Sarah sat down at the console, and ran the test cases as Chet directed. He and I started to scope the logic that generated the end-op signal, but we hadn't found anything by three o'clock.

While we were scoping, Steve was explaining the test cases to Sarah, along with what he had found out so far with his analysis. Finally, she exclaimed, "Good work, Steve! You certainly picked up the significant items quickly!"

Steve answered, "Maybe, but from here on out, you're going to have to help them. As I told you yesterday, I have to go back to Washington tomorrow unless there's a big snowstorm. I won't be back until the problems have been fixed, hopefully in January".

I asked Chet, "On the basis of what you've seen so far, what do you think?"

"Well, those ugly end-op signals are significant, and I would guess that they have a lot to do with the extra execution. But we shouldn't be seeing any signal that looks like that, at least if the machine is wired correctly".

Steve interrupted, "Yes! In one of my lab courses, I saw something like that. It turned out that it was a bad component. Do you see anything if you trace back from the end-op signal?"

I said, "We started to do that, trying the most obvious sources, but it's going to take another two days to check them all!"

Chet suggested, "Let's try to scope forward from that no-refetch line and the Y branch code point, and see if either gives us any clue."

We did that until six, with no luck. I said, "Tomorrow, let's take a different approach, and go back to the byte logic unit control lines, scoping forward from the decode".

We went back to my house, where Mom was fixing dinner for us. She told Steve, "You might want to watch the news tonight; they said something about a big snowstorm that could make your trip tomorrow at least unpleasant, and probably dangerous".

He gave me a tender look, and said, "Gee, I might have to stay over another day!"

I hugged him as we went to the living room and the TV. Our timing was perfect; the Weather Bureau had announced a severe winter storm warning for Maryland, Delaware, and New Jersey, with the possibility of light snow in New York City and maybe as far north as Poughkeepsie. It certainly sounded major, and I looked at Steve.

"Dearest, please don't go. I'd worry too much about you!"

"I'm certainly not going to go tomorrow. Murray won't give me any difficulty, I'm sure".

He hugged me and nuzzled my ear, and we went back into the kitchen. Steve told Mom, "It looks like you're stuck with me an extra night. The storm looks pretty bad, although I don't think you'll see much of anything here. I guess Sandy will be stuck with me in the test cell an extra day, also".

Mom smiled, saying, "I'm glad that you're going to be sensible. Are you making any progress understanding the problem?"

I said, "We know more, but there are still too many strange things that don't make sense!"

"I've seen cases like that in the hospital. Often, it turned out that the patient had multiple unrelated serious problems, and the interactions were confusing everyone until the doctors identified and treated the first problem. After that, the rest were easier to resolve, one at a time!"

"Mom, that's very helpful! I've been suspecting that we may have more than one bug and their interactions have been confusing the dickens out of us!"

Steve said, "Yes. There's the end-op signal, there's the Y branch, and there is the byte logic unit control field. Why don't you pick one and focus exclusively on it until you find something?"

"Well, I'd said we were going to work on the byte logic unit controls tomorrow. I guess we'll just be a little bit more stubborn before we let go of that method of attack!"

"While you're doing that, I'll take those test cases and make variants that will allow you to loop on a single instruction. I would think that would help you scope things".

"Yes! You'll see how much when I get you to look at the scope traces tomorrow!"

Mom interrupted, "Well, glad that I contributed something, even though I don't understand what you two are doing down there. At any rate, it's time for dinner!"

STEVE - Wednesday, November 29, 1961

We sat down and filled our plates. Sheila had warmed up the last of the Thanksgiving turkey, and I was happy to see that there was still plenty of the sauerkraut to go along with it.