HARVEST Archive Files
These are the scans of the archive files, and the documents they contain,
ordered by archive file number. The archive file folder cover contains archival
reference data. Concatenating all the files will reconstruct the original
The files were scanned as grayscale images at 600 dpi to TIFF, with deflate
(Zip) compression, then converted to PDF. Thus, the files can be rather large,
weighing in at about 4.5MB per page. This was necessary as the originals vary
greatly in quality, and often contain fine-line diagrams which would be
otherwise lost. Each page is legal-sized to match the photocopies. Most of the
documents are letter-sized, however. There is no OCR text processing.
Articles derived from these sources can be found at the HARVEST main page.
15881, HARVEST General Purpose Computer and 7950 Data Processin System [Loose Sheets] [Includes summary of HARVEST], 1957 - 1959
(Yes, that typo in the title is part of the archive file name.)
A series of graphs charting computer capability at the NSA, explaining the need for HARVEST, which was expected to be 100x faster and give 10x more capability per dollar. Estimated cost: $13,000,000
Fiscal, staff, and resource allocation report for MPRO's (Machine Processing, an NSA department?) planned 1960-61 fiscal years. Discusses specs and capabilities of the TRACTOR tape system. Estimated HARVEST delivery in September 1960.
Internal IBM document summarizing the high-level structure of HARVEST, with some comments about the high-speed memory being still under development. Explains HARVEST as a "pluggable" computer, but with the plugs under software control. Mentions use of ECC memory (72 bit, SECDED).
One-page memo assigning 10 maintenance personel to various parts of HARVEST.
Memo about the scope and process towards creating a diagnostic program suite.