On 10/30/2017 06:18 AM, Eric Smith via cctalk wrote:
2. What IBM defined as emulation was use of extremely
specialized
hardware and/or microcode (specifically, not the machine's general-purpose
microcode used for natively programming the host machine).
As far as I know,
IBM's 360s did NOT have any special
purpose hardware associated with their emulation packages.
The only thing special was that additional microcode cards
were installed. In the 360/30, these were Mylar cards with
word lines on them, they were punched on a card punch to
punch out the capacitor plates to make zeroes.
On the 360/40, they were Mylar "tapes" that were punched to
cut traces to go through or bypass the sense transformers.
On 360/50 and 65, they were etched word line boards that had
traces that weaved under the bit line capacitor plates.
So, these were all custom to fit the specific instruction
set to be emulated. But, as far as I know, they added no
additional logic to the machine to support the emulation.
In some cases, this made things fairly inefficient. At
least on the 360/30, when running 14xx emulation, there were
many holes in memory, because they did not convert between
decimal and binary addresses. So, memory locations 0-9 were
used, A-F were inaccessible, 10-19 were used, 1A-1F were
skipped, and so on. Some of this made I/O buffers kind of
strange, as the I/O buffers had to be repacked between the
real I/O devices and the emulator's buffers.
Jon