ROM Dumping

The Wire ROM from machine CC2001 was read with the custom reader using its ‘stability’ mode where each address was read 16 times and the returned data analyzed for any variations from the same address. Every address read consistently with no variations.

The complete ROM was then dumped in raw hex mode multiple times and the set of dumps passed to Doug Miller, one of a (very) few Wang microcode gurus. He was able to compare with other images in his possession and to reverse engineer the contents of the four dead addresses.

The result was a fully restored wire ROM image, known as WireROM-CC2001.

Analysis and Comparison with other ROMs

Frequency analysis on the final CC2001 image shows a pretty good utilisation of all opcodes, this is to be expected considering the cost of microcode bits.

Val    AI    BI    ZO   AOP    AC    BC   MOP    KK    ST     J/C   JADh   JADm   JADl   JH  JL
  0  1273  1255   580  1447  1241  1739  1371   899  1533    1848    804    198    172  928 945
  1    21   676    59   274   807   308    72   113    26     200   1244    107    111  842 690
  2    21     4   123    64               129   102     7                   136    119   65  77
  3   211     7   455    70               127    89    62                   100    102   82  77
  4   106    47   205    31                79    51    68                   103    136    1 180
  5    63    23   138    54                82    52     5                   105    126  124   1
  6   345    35   485   106               146    77     5                   103    144    5  17
  7     8     2     1     1                 4    48    40                    95    122    1  61
  8                                         3    69    37                   142    117
  9                                         0    62     7                   177    123
  A                                         7    61    19                   190    128
  B                                         1    43    29                   153    156
  C                                         3    57    21                   121    139
  D                                         3   117   123                   105    138
  E                                         2   160     2                   106    103
  F                                        16    48    61                    99    106

Test Addresses in Wire ROM

During the ROM repair process Doug Miller made a very clever observation, noticing an emerging pattern in the last four addresses even though the ROM was being improperly read and the output was plagued by noise and errors.

He hypothesised that the last four addresses were test locations:

• Address 7FC: Alternating one and zero
• Address 7FD: Alternating zero and one (ie complement of 7FD)
• Address 7FE: All ones
• Address 7FF: Mostly zeros, with the value “130” in the 3-digit JAD field

As the ROM and Reader were brought to final form, these patterns were confirmed and were very useful reassurances that the ROM was being read reliably. The complementary patterns of alternating 1s and 0s in locations 7FC and 7FD are obviously useful for bit and cross-talk tests. The all-1s address allows testing at one extreme. Address 7FF is a bit of a puzzle because it is not all zeros and the value 130 reads very consistently, it is higly likely to be a programmed value and not an error. What the 130 represents is a mystery, perhaps it is a ROM revision level, or perhaps even a serial number?

Nevertheless, it was very satisfying to see that these four locations did read as above, confirming Doug’s insight and the operation of the reader.

The WireROM-CC2001 image will be loaded into the ROM emulator and used to drive the restored machine. The Wire ROM will be left in situ for appearance but cannot be used for machine operation due to the four missing addresses