欢迎访问ic37.com |
会员登录 免费注册
发布采购

PM7351-BGI 参数 Datasheet PDF下载

PM7351-BGI图片预览
型号: PM7351-BGI
PDF下载: 下载PDF文件 查看货源
内容描述: [Support Circuit, 1-Func, CMOS, PBGA304, 31 X 31 MM, 1.51 MM HEIGHT, 1.27 MM PITCH, SBGA-304]
分类和应用: ATM异步传输模式电信电信集成电路
文件页数/大小: 174 页 / 1840 K
品牌: PMC [ PMC-SIERRA, INC ]
 浏览型号PM7351-BGI的Datasheet PDF文件第133页浏览型号PM7351-BGI的Datasheet PDF文件第134页浏览型号PM7351-BGI的Datasheet PDF文件第135页浏览型号PM7351-BGI的Datasheet PDF文件第136页浏览型号PM7351-BGI的Datasheet PDF文件第138页浏览型号PM7351-BGI的Datasheet PDF文件第139页浏览型号PM7351-BGI的Datasheet PDF文件第140页浏览型号PM7351-BGI的Datasheet PDF文件第141页  
RELEASED  
PM7351 S/UNI-VORTEX  
DATA SHEET  
PMC-1980582  
ISSUE 5  
OCTAL SERIAL LINK MULTIPLEXER  
12  
OPERATION  
This section complements Section 9 Functional Description, by describing in  
greater detail some of the programming and cross-functional issues that a  
designer using the S/UNI-VORTEX may want to consider.  
12.1 Determining the Value for FREADY[5:0]  
As discussed in Section 9.1.1, a downstream cell transfer should be initiated only  
after polling the appropriate downstream channel and ensuring that the buffer is  
empty, as indicated by that channel’s TPA (Transmit Packet Available) output  
being asserted. The cell buffer status for each downstream channel (i.e. the  
status that will be driven onto the TPA output when the channel is polled) is  
deasserted when the first byte of a cell is written into the buffer. It is re-asserted  
only after the number of bytes programmed into the FREADY[5:0] field of the  
associated Downstream Logical Channel FIFO Ready Level register have been  
serialized onto a high-speed link. Determining whether or not it is necessary for  
you to adjust the default value of FREADY[5:0] is discussed in this section.  
Each S/UNI-VORTEX has a total of 264 downstream cell buffers. There are 33  
buffers per LVDS link (32 PHYs plus the microprocessor port), and 8 links in  
total. Each link is independent of, but identical to the others, so we need only  
describe how a single link works for you to understand how all downstream links  
function.  
At the end of each cell transfer on the LVDS link the downstream scheduler polls  
(round-robin) its 33 buffers looking for the next buffer with a whole cell to send  
(the scheduler ignores buffers that are in the process of having a cell written to  
them). If there is no cell to send the scheduler immediately injects a stuff cell on  
the link. Once the link starts sending out the cell from a buffer, the entire cell will  
be sent before the scheduler begins its next scheduling cycle. Since a buffer  
cannot have more than one cell stored in it at a time, no single downstream  
channel can ever operate at a bandwidth greater than ½ the bandwidth of the  
LVDS link. Even if there are no other channels active, the scheduler will inject a  
stuff cell before the next cell from the same channel is available.  
So what does this mean to the system designer? The maximum per channel  
bandwidth (i.e. ½ the LVDS line rate) will only be sustained if the bus master  
always writes the next cell into the downstream buffer within the time it takes to  
transmit a single cell over the LVDS link. Since the system bus is normally  
running much faster than the LVDS link this is not a problem in most designs.  
However, there are two scenarios in which the designer may be concerned about  
the time delay between TPA asserted and the last byte of the cell being written  
into the buffer:  
PROPRIETARY AND CONFIDENTIAL TO PMC-SIERRA, INC., AND FOR ITS CUSTOMERS’ INTERNAL USE  
127