поискавой системы для электроныых деталей
  Russian  ▼
ALLDATASHEETRU.COM

X  

RM0319 датащи(PDF) 79 Page - STMicroelectronics

номер детали RM0319
подробное описание детали  SPEAr320S architecture and functionality
PDF  368 Pages
Scroll/Zoom Zoom In 100%  Zoom Out
производитель  STMICROELECTRONICS [STMicroelectronics]
домашняя страница  http://www.st.com
Logo STMICROELECTRONICS - STMicroelectronics

RM0319 датащи(HTML) 79 Page - STMicroelectronics

Back Button RM0319 Datasheet HTML 75Page - STMicroelectronics RM0319 Datasheet HTML 76Page - STMicroelectronics RM0319 Datasheet HTML 77Page - STMicroelectronics RM0319 Datasheet HTML 78Page - STMicroelectronics RM0319 Datasheet HTML 79Page - STMicroelectronics RM0319 Datasheet HTML 80Page - STMicroelectronics RM0319 Datasheet HTML 81Page - STMicroelectronics RM0319 Datasheet HTML 82Page - STMicroelectronics RM0319 Datasheet HTML 83Page - STMicroelectronics Next Button
Zoom Inzoom in Zoom Outzoom out
 79 / 368 page
background image
RM0319
Multiport DDR controller (MPMC)
Doc ID 022640 Rev 3
79/368
maximize efficiency of the core through command grouping and bank splitting. Once placed
into the command queue, the relative order of commands is constant.
Many of the rules used in placement may be individually enabled/disabled. In addition, the
queue may be programmed by the placement_en parameter to disable the placement logic
entirely, resulting in an in-line queue that attends requests in the order they are received. If
the placement_en parameter is cleared to 1'b0, the placement algorithm will be ignored.
Rules of the placement algorithm
The factors affecting command placement together work to identify where a new command
fits into the execution order. They are listed in order of importance.
Address collision/Data coherency violation
The order in which READ and WRITE commands are processed inside MPMC is critical to
proper system behavior. While READs and WRITEs to different addresses are independent
and may be re-ordered without affecting system performance, READs and WRITEs that
access the same address are significantly related. If the port requests a READ after a
WRITE to the same address, then repositioning the READ before the WRITE would return
the original data, not the changed data. Similarly, if the READ was requested ahead of the
WRITE but accidentally positioned after the WRITE, the READ would return the new data,
not the original data prior to being overwritten. These are significant data coherency
mistakes.
To avoid address collisions, READs or WRITEs that access the same chip select, bank and
row as a command already in the command queue will be inserted into the command queue
after the original command, even if the new command is of a higher priority.
This factor may be enabled/disabled through the addr_cmp_en parameter and should only
be disabled if the system can guarantee coherency of READs and WRITEs.
Source ID collision
Each port is assigned a specific source ID that identifies the source uniquely. This allows the
memory controller to map data from/to the correct source/destination.
Note:
A source ID does contain port identification information which means that the rules for
placement are dependent on the requesting port. There will not be source ID collisions
between ports.
In general, commands of the same type from the same source ID will be placed in the
command queue in order. Therefore, a READ/WRITE command with the same source ID of
a READ/WRITE command being already in the command queue will be processed
afterward.
The behavior of commands of different types from the same source ID is dependent on the
user configuration. For MPMC, the placement of new READ/WRITE commands that collide
in terms of source ID with existing entries in the command queue will only depend on other
commands of the same type, not on different types. This means that, if there are no address
conflicts, a READ command could be executed ahead of a WRITE command with the same
source ID, as a WRITE command could be executed ahead of a READ command with the
same source ID as well.
This feature is always enabled.
Write buffer collision
Incoming WRITE requests in the command queue are allocated to one of the 8 WRITE
buffers of the core automatically based on availability. New WRITE commands will be



Html Pages

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100  ...More


датащи скачать

Go To PDF Page


ссылки URL



Вашему бизинису помогли Аллдатащит?  [ DONATE ] 

Что такое Аллдатащит   |   реклама   |   контакт   |   Конфиденциальность   |   Ссылка на техническое описание    |   обмен ссыками   |   поиск по производителю
All Rights Reserved©Alldatasheet.com


Mirror Sites
English : Alldatasheet.com  |   English : Alldatasheet.net  |   Chinese : Alldatasheetcn.com  |   German : Alldatasheetde.com  |   Japanese : Alldatasheet.jp
Russian : Alldatasheetru.com  |   Korean : Alldatasheet.co.kr  |   Spanish : Alldatasheet.es  |   French : Alldatasheet.fr  |   Italian : Alldatasheetit.com
Portuguese : Alldatasheetpt.com  |   Polish : Alldatasheet.pl  |   Vietnamese : Alldatasheet.vn
Indian : Alldatasheet.in  |   Mexican : Alldatasheet.com.mx  |   British : Alldatasheet.co.uk  |   New Zealand : Alldatasheet.co.nz
Family Site : ic2ic.com  |   icmetro.com