Hardware Reference
In-Depth Information
File 0
File 1
File 2
File 3
File 4
File 5
File 6
File 7
File 8
File 9
File 10
File name:
Rubber-ducky
Length:
Type:
Creation date:
Last access:
Last change:
Total accesses:
Block 0:
Block 1:
Block 2:
Block 3:
1840
image/jpeg
March 16, 1066
September 1, 1492
July 4, 1776
144
Tr a ck 4
Track 19
Track 11
Track 77
Sector 6
Sector 9
Sector 2
Sector 0
Figure 6-23.
A user file directory and the contents of a typical entry in a file di-
rectory.
parallel, certain virtual instructions are needed. These instructions will be dis-
cussed in the following sections.
The laws of physics provide yet another reason for the current interest in paral-
lel processing. According to Einstein's special theory of relativity, it is impossible
to transmit electrical signals faster than the speed of light, which is nearly 1 ft/nsec
in vacuum, less in copper wire or optical fiber. This limit has important implica-
tions for computer organization. For example, if a CPU needs data from the main
memory 1 ft away, it will take at least 1 nsec for the request to arrive at the memo-
ry and another nanosecond for the reply to get back to the CPU. Consequently,
subnanosecond computers will need to be extremely tiny. An alternative approach
to speeding up computers is to build machines with many CPUs. A computer with
a thousand 1-nsec CPUs may (in theory) have the same computing power as one
CPU with a cycle time of 0.001 nsec, but the former may be much easier and
cheaper to construct. Parallel computing is discussed in detail in Chap. 8.
On a computer with more than one CPU, each of several cooperating processes
can be assigned to its own CPU, to allow the processes to progress simultaneously.
If only one processor is available, the effect of parallel processing can be simulated
by having the processor run each process in turn for a short time. In other words,
the processor can be shared among several processes.
Figure 6-24 shows the difference between true parallel processing, with more
than one physical processor, and simulated parallel processing, with only one phys-
ical processor. Even when parallel processing is simulated, it is useful to regard
each process as having its own dedicated virtual processor. The same communica-
tion problems that arise when there is true parallel processing arise also in the sim-
ulated case. In both cases, debugging the problems is very difficult.
