![Lattice Semiconductor LatticeMico32 Скачать руководство пользователя страница 29](http://html1.mh-extra.com/html/lattice-semiconductor/latticemico32/latticemico32_hardware-developer-user-manual_3843852029.webp)
U
SING
THE
L
ATTICE
M
ICO
S
YSTEM
S
OFTWARE
:
Creating the Microprocessor Platform in MSB
LatticeMico32 Hardware Developer User Guide
23
have default priorities assigned in their components' .xml files when you add
the components to the platform. Since each multi-master slave has its own
arbiter in this scheme, arbitration priorities are assigned per slave. However,
you can change these priorities by selecting Platform Tools > Edit Arbitration
Priorities and changing the priorities in the Edit Arbitration Priorities dialog
box. When you perform a DRC check, MSB checks the validity of the priorities
that you have changed.
Slave-Side Round-Robin Arbitration
The slave-side round robin
arbitration scheme is similar to the slave-side fixed arbitration scheme in that
each multi-master slave has one arbiter, but all masters have the same
priority. The arbiter grants access to all the masters that request a slave in a
round-robin, or circular, fashion. Once the requesting master is finished with
its transfer, the next master obtains access to the slave.
In the slave-side round-robin scheme, the Platform Tools > Edit Arbitration
Priorities command is not available.
Comparing the Arbitration Schemes
The difference between the slave-side fixed arbitration scheme and the slave-
side round-robin arbitration scheme is how the arbiter grants requesting
masters access to the bus. The slave-side fixed scheme always gives the
highest-priority master access to the bus if that master requests it. The slave-
side round-robin scheme grants masters access to the bus in a round-robin
fashion.
Both the slave-side fixed and the slave-side round-robin arbitration schemes
use separate arbiters for each multi-master slave, so the area of the platforms
generated with these schemes is slightly larger than that resulting from the
shared-bus arbitration scheme. For example, for a typical system consisting
of four multi-master slaves, the slave-side fixed and the slave-side round-
robin schemes require four arbiters, but the shared-bus scheme requires only
one arbiter. The area required by the system with the slave-side arbitration
schemes is approximately three times more than the area required by the
system with the shared-bus arbitration scheme.
The slave-side arbitration schemes offer better performance than the shared-
bus arbitration scheme. For example, the SoC used in this topic (a CPU with a
DMA controller) yields better performance with a slave-side arbitration
scheme than with a shared-bus arbitration scheme. When a slave-side
arbitration scheme is used in this SoC, the DMA controller's read and write
ports can work in parallel and transfer the data from the external SRAM
memory to on-chip memory. When a shared-bus arbitration scheme is used in
the SoC, data cannot be transferred in parallel because there is a single
arbiter for both memories.
Whether you select a slave-side fixed or slave-side round robin arbitration
scheme depends on the application. If the application requires each master to
have equal access to a slave, the slave-side round-robin scheme is a better
option. If the application requires a certain master to have access to a slave
as soon as the current master is finished with the data transfer, the slave-side
fixed scheme is the best option.