Generic Computer
 |
Group Name: Performance Modeling
Directory: perfmod
|
|
|
Description
The GenericComputer device models the basic functioning of a
computer and collects statistics. The GenericComputer is designed
to behave similarly to any real computer but it simulates a
compute operation with a representative delay instead of an actual
computation. GenericComputers are driven by program files (usually
from compiled DFGs) or instruction lines.
Instructions may be associated with a particular process in the
GenericComputer. If there exist multiple processes in a
GenericComputer, the GenericComputer will interleave execution of
instructions from different processes. This concept is known
multitasking. The way in which the GenericComputer chooses the
next process to execute instructions from is known as
scheduling. The default scheduler in the GenericComputer will
schedule based on the priority of the instructions. The default
scheduler will preempt (interrupt) the current instruction if a
higher priority instruction is send to the processor.
Ports
Input
- instruction lines
Accepts instructions for the computer to
execute. An instruction line may have any name except
"io_in".
- io_in
Input port for arriving messages.
Output
- io_out
Output port for outgoing messages.
Attributes
- speed_up = 1.0
Scales the compute time of instructions.
- boot_time = 0.0ms
The time this computer takes to boot.
- context_switch_time = 0.0ms
The time to switch between processes.
- tcp_send_overhead = 1000.0ms
.
- tcp_per_byte_cost = 1000.0ms
.
Output
- processing timeline
The time to switch between processes.
- utilization
The time to switch between processes.
Generic CPU
 |
Group Name: Performance Modeling
Directory: perfmod
|
|
|
Description
The GenericCPU device models the basic functioning of a
computer and collects statistics. The GenericCPU is designed to
behave similarly to any real computer but it simulates a compute
operation with a representative delay instead of an actual
computation. GenericCPUs are driven by program files (usually from
compiled DFGs) or instruction lines.
Instructions may be associated with a particular process in the
GenericCPU. If there exist multiple processes in a GenericCPU, the
GenericCPU will interleave execution of instructions from
different processes. This concept is known multitasking. The way
in which the GenericCPU chooses the next process to execute
instructions from is known as scheduling. The default scheduler in
the GenericCPU will schedule based on the priority of the
instructions. The default scheduler will preempt (interrupt) the
current instruction if a higher priority instruction is send to
the processor.
Ports
Input
- instruction lines
Accepts instructions for the CPU to execute. An
instruction line may have any name except "io_in".
- io_in
Input port for arriving messages.
Output
- io_out
Output port for outgoing messages.
Attributes
- speed_up = 1.0
Scales the compute time of instructions.
- boot_time = 0.0
The time this computer takes to boot.
- context_switch_time = 0.0
The time to switch between processes.
Output
- processing timeline
The time to switch between processes.
- utilization
The time to switch between processes.
Protocol Stack
 |
Group Name: Performance Modeling
Directory: perfmod
|
|
|
Description
Ports
Input
- out_in
Input port for outgoing messages.
- io_in
Input port for arriving messages.
Output
- io_out
Output port for outgoing messages.
- in_out
Output port for arriving messages.
- ins
Output port for cpu instructions.
Attributes
- tcp_send_overhead = 1000.0ms
.
- tcp_per_byte_cost = 500.0ms
.
- udp_send_overhead = 500.0ms
.
- udp_per_byte_cost = 100.0ms
.
Message Distributer
 |
Group Name: Performance Modeling
Directory: perfmod
|
|
|
Description
The MsgDistributer device routes incomming performance model
messages to the output port with a name that matches a particular
field in the message.
Ports
Input
- in
Input port for arriving messages.
Output
- output lines
Output port for outgoing messages.
Attributes
- field
Sets which field to compare the output port names
to.
Message Funnel
 |
Group Name: Performance Modeling
Directory: perfmod
|
|
|
Description
The MsgFunnel device routes all incomming performance model
messages to a single output port named "out".
Ports
Input
- input lines
Input port for arriving messages.
Output
- out
Output port for outgoing messages.
Hardware Communications Device
 |
Group Name: Performance Modeling
Directory: perfmod
|
|
|
Description
The MsgFunnel device routes all incomming performance model
messages to a single output port named "out".
Ports
Input
- io_out
Input port for outgoing messages.
- io_port
Input port for arriving messages.
Output
- io_in
Output port for arriving messages.
- io_port
Output port for outgoing messages.
Generic Store & Forward Switch
 |
Group Name: Performance Modeling
Directory: perfmod
|
|
|
Description
The MsgFunnel device routes all incomming performance model
messages to a single output port named "out".
Ports
Input
- io_out
Input port for outgoing messages.
- io_port
Input port for arriving messages.
Output
- io_in
Output port for arriving messages.
- io_port
Output port for outgoing messages.
Aron Rubin
Last modified: Mon Apr 9 16:50:21 EDT 2001