Thought I'd hear better ones than most listed here. For example:
1. A failure that happened every day at a gradually changing time. Cause: heat from sunlight on a system. The failure happened after the sun came through a window at the correct angle. The time was changing as the sun moved with the season.
2. A falure that happened once a year at the same time: 10/10, at 10:10AM. The time was represented in BCD, and using IBM Bisync protocol, the bytes were escaped. That many 10s caused a buffer overflow.
48 bits? I recall 36 bits, split into 18 bit segment and 18 bit offset. Because of the memory mapping, standard files were limited to 2**18 words, approx 1 MB. Larger files were "MSF", or multi-segment files, which had a sort of builtin directory structure, which not all tools supported.
I will say, that the I/O speed impressed me!
(P.S. I used Multics at the U.S. Geological Survey arouund 1980. It was a fun machine.)
Bzzt. Don't forget optical techniques. Lenses have been made +- 1/4 wave (around 7 microinches) for many decades.
Also making something spherical is easier than arbitrary shapes. Standard ball bearing balls are +- 25 microinches, and have been for years. Think lapping.
Slashdot Mirror
I used Vol. 2 to improve the multiply algorithm in an open source program.
Thought I'd hear better ones than most listed here. For example:
1. A failure that happened every day at a gradually changing time. Cause: heat from sunlight on a system. The failure happened after the sun came through a window at the correct angle. The time was changing as the sun moved with the season.
2. A falure that happened once a year at the same time: 10/10, at 10:10AM. The time was represented in BCD, and using IBM Bisync protocol, the bytes were escaped. That many 10s caused a buffer overflow.
48 bits? I recall 36 bits, split into 18 bit segment and 18 bit offset. Because of the memory mapping, standard files were limited to 2**18 words, approx 1 MB. Larger files were "MSF", or multi-segment files, which had a sort of builtin directory structure, which not all tools supported.
I will say, that the I/O speed impressed me!
(P.S. I used Multics at the U.S. Geological Survey arouund 1980. It was a fun machine.)
Bzzt. Don't forget optical techniques. Lenses have been made +- 1/4 wave (around 7 microinches) for many decades.
Also making something spherical is easier than arbitrary shapes. Standard ball bearing balls are +- 25 microinches, and have been for years. Think lapping.