- 您现在的位置：买卖IC网 > PDF目录67681 > IBM21P100BGB PCI BUS CONTROLLER, PBGA304 PDF资料下载
参数资料
| 型号： | IBM21P100BGB |
| 元件分类： | 总线控制器 |
| 英文描述： | PCI BUS CONTROLLER, PBGA304 |
| 封装： | 31 X 31 MM, HEAT SINK, PLASTIC, BGA-304 |
| 文件页数： | 52/140页 |
| 文件大小： | 2032K |
| 代理商： | IBM21P100BGB |
第1页第2页第3页第4页第5页第6页第7页第8页第9页第10页第11页第12页第13页第14页第15页第16页第17页第18页第19页第20页第21页第22页第23页第24页第25页第26页第27页第28页第29页第30页第31页第32页第33页第34页第35页第36页第37页第38页第39页第40页第41页第42页第43页第44页第45页第46页第47页第48页第49页第50页第51页当前第52页第53页第54页第55页第56页第57页第58页第59页第60页第61页第62页第63页第64页第65页第66页第67页第68页第69页第70页第71页第72页第73页第74页第75页第76页第77页第78页第79页第80页第81页第82页第83页第84页第85页第86页第87页第88页第89页第90页第91页第92页第93页第94页第95页第96页第97页第98页第99页第100页第101页第102页第103页第104页第105页第106页第107页第108页第109页第110页第111页第112页第113页第114页第115页第116页第117页第118页第119页第120页第121页第122页第123页第124页第125页第126页第127页第128页第129页第130页第131页第132页第133页第134页第135页第136页第137页第138页第139页第140页
IBM21P100BGB
IBM 133 PCI-X Bridge R1.1
ppb11_intro.fm.03
July 9, 2001
General Information
Page 11 of 131
2.5.2.2 Posted Write Buffers
The posted write buffers each have a capacity of 1 KB to hold data from posted memory write transactions.
Each is logically divided into eight independent 128-byte segments to allow transactions to be issued on the
target bus before they have been completed on the initiating bus. Unlike the burst read buffers, the amount of
space assigned to each transaction is dynamic; that is, a single transaction can utilize from one to all eight
128-byte subsections as needed. Hence, each posted write queue is an 8-entry FIFO, and up to eight active
write transactions in each direction are possible. Again, activity generally occurs in the bridge when a 128-
byte segment is filled or emptied, and the design keeps data flowing through the bridge by using or re-using
128-byte subsections as they become available.
2.5.2.3 Single Data Phase Buffers
There is one single data phase buffer for each direction to hold read or write data from 4-byte split or delayed
transactions. These transactions include all I/O or configuration operations as well as doubleword memory
read operations.
2.5.3 Address Decoding
The IBM 133 PCI-X Bridge R1.1 is a transparent bridge and utilizes a flat addressing model, meaning that
both PCI and PCI-X address spaces are split between the primary bus and the secondary bus. Address
ranges residing on the secondary bus are defined by the I/O, Memory, and Prefetchable Memory Base and
Limit registers in the bridge configuration space. All other addresses are assumed to reside on the primary
bus; in other words, inverse address decoding is used to determine when to forward transactions upstream.
The only exception to this is when the optional opaque address range is defined by its Base and Limit regis-
ters and is enabled. By definition, the IBM 133 PCI-X Bridge R1.1 does not recognize transactions on either
bus to addresses within the opaque range. This region may be used, for example, for peer-to-peer communi-
cation between devices on the secondary bus.
The IBM 133 PCI-X Bridge R1.1 supports full 64-bit addressing and handles dual address cycles on both
interfaces. The device provides no capability for translating addresses.
The IBM 133 PCI-X Bridge R1.1 configuration registers are accessible only from the primary interface. There
are no other I/O- or memory-mapped facilities in the device, so the optional Base Address registers 0 and 1 of
the configuration header are not implemented. On the secondary interface, the bridge ignores all configura-
tion read transactions as well as Type 0 configuration write transactions, and it only claims Type 1 configura-
tion write transactions that specify conversion to a special cycle on an upstream bus segment.
2.5.4 Bus Arbitration
The IBM 133 PCI-X Bridge R1.1 contains an arbiter for the secondary interface that is enabled or disabled via
an input signal pin. It provides bus arbitration for up to six additional masters, each of which may be assigned
high or low priority or may be masked off. When the internal arbiter is being used and the IBM 133 PCI-X
Bridge R1.1 request is not masked off, the bus will be parked at the bridge whenever there are no pending
requests.
The arbiter implements a two-level fairness algorithm that allows each device within a level to receive grant
requests cyclically. The arbiter uses the Arbitration Priority register to determine which agents are high
priority (HP) devices and which are low priority (LP) devices. At different points in time, snapshots are taken
of all pending requests at each priority level. All captured HP requests are serviced first, then one of the
captured LP requests is serviced. At this point, a new HP snapshot is taken, picking up any new HP requests,
相关PDF资料 |
PDF描述 |
|---|---|
| IBM21P100BGC | PCI BUS CONTROLLER, PBGA304 |
| IBM25403GCX-3JC76C2 | RISC PROCESSOR, PQFP16 |
| IBM25405GP-3BA200C2 | RISC PROCESSOR, PBGA456 |
| IBM25EMPPC603EFG-100 | 32-BIT, 100 MHz, RISC PROCESSOR, PQFP240 |
| IBM25EMPPC603EBG-100 | 32-BIT, 100 MHz, RISC PROCESSOR, CBGA255 |
相关代理商/技术参数 |
参数描述 |
|---|---|
| IBM24L5086 | 制造商:AVED MEMORY PRODUCTS 功能描述: 制造商:AVED Memory Products 功能描述: |
| IBM25403GCX-3BC80C2 | 制造商:IBM 功能描述:RISC PROCESSOR, 160 Pin Plastic BGA |
| IBM25403GCX-3JC50C2 | 制造商:IBM 功能描述:403GCX-3JC50C2 |
| IBM25403GCX-3JC66C2 | 制造商:IBM 功能描述: |
| IBM25403GCX3JC76C2 | 制造商:IBM 功能描述: |
发布紧急采购，3分钟左右您将得到回复。