欢迎访问ic37.com |
会员登录 免费注册
发布采购

PM7350-PI 参数 Datasheet PDF下载

PM7350-PI图片预览
型号: PM7350-PI
PDF下载: 下载PDF文件 查看货源
内容描述: 双串行链路物理层复用器 [DUAL SERIAL LINK PHY MULTIPLEXER]
分类和应用: 复用器ATM集成电路SONET集成电路SDH集成电路电信集成电路电信电路异步传输模式
文件页数/大小: 241 页 / 1939 K
品牌: PMC [ PMC-SIERRA, INC ]
 浏览型号PM7350-PI的Datasheet PDF文件第82页浏览型号PM7350-PI的Datasheet PDF文件第83页浏览型号PM7350-PI的Datasheet PDF文件第84页浏览型号PM7350-PI的Datasheet PDF文件第85页浏览型号PM7350-PI的Datasheet PDF文件第87页浏览型号PM7350-PI的Datasheet PDF文件第88页浏览型号PM7350-PI的Datasheet PDF文件第89页浏览型号PM7350-PI的Datasheet PDF文件第90页  
RELEASED  
PM7350 S/UNI-DUPLEX  
DATA SHEET  
PMC-1980581  
ISSUE 5  
DUAL SERIAL LINK PHY MULTIPLEXER  
register (0x5E) to be four cells deep as per the I.432 specification. If flow control  
asserted by the far-end device prevents cells from a particular serial data  
interface from being read, the corresponding FIFO will begin to fill. In general,  
proper system engineering should ensure that the far-end is able to keep up, but  
should flow control continue to be asserted by the far-end an overflow condition  
could occur. Overflow results in an optional interrupt being raised, an error  
status register being set, and subsequently received cells being discarded until  
FIFO space is available.  
Regardless of whether configured a bus master, bus slave, or clocked serial data  
device, the S/UNI-DUPLEX monitors the per-channel FIFOs and performs round  
robin scheduling of cells onto the transmit LVDS links. The control channel FIFO  
is given equal priority to the PHYs.  
The device attached to the far-end of the LVDS link is typically a S/UNI-VORTEX  
(see Fig. 1) or a S/UNI-DUPLEX (see Fig. 4 and Fig. 5). Although the S/UNI-  
DUPLEX only implements one form of transmit flow control, to help understand  
the end-to-end flow control we will discuss the two situations separately.  
Case #1: the far-end device is a S/UNI-VORTEX  
The S/UNI-VORTEX is a parallel bus slave with a single receive buffer for each  
of its eight LVDS links. It uses a weighted round robin polling of these buffers  
and schedules the received cells onto the parallel bus under the control of the  
bus master. The S/UNI-VORTEX’s eight LVDS links running at 200 Mb/s can  
provide aggregate traffic bursts that exceed the capacity of a 800 Mb/s parallel  
bus. Also, link polling weights may be set such that one or more links receive  
relatively less of the aggregate bus bandwidth. Therefore, it is quite likely that  
congestion will sometimes occur at the S/UNI-VORTEX’s LVDS receive buffers.  
Since all traffic received on the eight LVDS links is heading to the S/UNI-  
VORTEX’s parallel bus there was no need to partition this FIFO into individual  
channels. Therefore the S/UNI-VORTEX implements flow control on each of its  
LVDS links in a binary fashion – all 32 logical channels are enabled or all  
channels are disabled. The S/UNI-VORTEX may or may not treat the  
microprocessor control channel independently from the user channels depending  
on how it is configured. If it is programmed to route control channel cells out  
over the parallel bus then the control channel flow control is identical to the 32  
user channels. If the control channel is being routed to the microprocessor port  
then the control channel flow control is independent from the user channels.  
When the S/UNI-VORTEX’s LVDS receiver cell buffer becomes congested the  
S/UNI-VORTEX will immediately clear the CA bits and possibly the UPCA fields  
in Table 16 in the cells sent on the downstream LVDS link. When the S/UNI-  
PROPRIETARY AND CONFIDENTIAL TO PMC-SIERRA, INC., AND FOR ITS CUSTOMERS’ INTERNAL USE  
74  
 复制成功!