The sixties were a great time for computer architecture. Many recognizable ideas - caches, out-of-order, virtual memory, SIMD all emerged during the sixties. CPU was not yet a single integrated circuit. Vacuum tubes were still there in the early sixties. Large arrays of circuit boards with SLT (Solid Logic Technology) in the mid to late sixties.
From the many papers from this time, highlighting here two interesting write-ups that introduce "The Cache"
The author in the paper wrote:
"I wish in this note to draw attention to the use of a fast memory as a slave memory. By a slave memory I mean one which automatically accumulates to itself words that come from a slower main memory, and keeps them available for subsequent use without it being necessary for the penalty of main memory access to be incurred again. Since the slave memory can only be a fraction of the size of the main memory, words cannot be preserved in it indefinitely, and there must be wired into the system an algorithm by which they are progressively overwritten. In favorable circumstances, however, a good proportion of the words will survive long enough to be used on subsequent occasions and a distinct gain of speed results. The actual gain depends on the statistics of the particular situation."
This paper from the sixties has one of the first appearances of ideas that we now recognize as -
Cache Tags, Tag Arrays, Cache states - Valid/Invalid; Dirty/Not_Dirty
The Cache - one of the earliest appearances of the word cache, which has become common vocabulary to computer architects and programmers now.
Authors' original wording:
"Both the cache and main storage are logically divided into sectors, each consisting of 1K contiguous bytes starting on 1K-byte
boundaries"
My read about the use of the word sector: The word sector made its way into main memory and caches from the drums and disks of the day, which used circular sectors.
In today's terminology, we call these sets. What's described in this paper is a set-associative cache of size 16, 24, or 32K. Each set is 1K. And interestingly, the cache line size is 64 bytes. Yes, 64 bytes. To this day, 64-byte cache sizes stay prevalent in CPUs. (side note: GPUs have moved to larger cache line sizes).
The cache was write-through. The LRU granularity was a set, not a cache line.
retained the original wording on the historical title/quote; The relingo change is a modern sensitivity, which was unknown at that time.
Top comments (0)