欢迎访问ic37.com |
会员登录 免费注册
发布采购

IBM25PPC970FX6TB429RT 参数 Datasheet PDF下载

IBM25PPC970FX6TB429RT图片预览
型号: IBM25PPC970FX6TB429RT
PDF下载: 下载PDF文件 查看货源
内容描述: [RISC Microprocessor, 64-Bit, 2200MHz, CMOS, CBGA576, 25 X 25 MM, 1 MM PITCH, CERAMIC, BGA-576]
分类和应用: 时钟外围集成电路
文件页数/大小: 68 页 / 1960 K
品牌: IBM [ IBM ]
 浏览型号IBM25PPC970FX6TB429RT的Datasheet PDF文件第7页浏览型号IBM25PPC970FX6TB429RT的Datasheet PDF文件第8页浏览型号IBM25PPC970FX6TB429RT的Datasheet PDF文件第9页浏览型号IBM25PPC970FX6TB429RT的Datasheet PDF文件第10页浏览型号IBM25PPC970FX6TB429RT的Datasheet PDF文件第12页浏览型号IBM25PPC970FX6TB429RT的Datasheet PDF文件第13页浏览型号IBM25PPC970FX6TB429RT的Datasheet PDF文件第14页浏览型号IBM25PPC970FX6TB429RT的Datasheet PDF文件第15页  
.
Data Sheet  
Preliminary  
PowerPC 970FX  
1. General Information  
1.1 Description  
The IBM PowerPC 970FX RISC Microprocessor,  
is a 64-bit implementation of the IBM PowerPC®  
family of reduced instruction set computer (RISC)  
microprocessors that are based on the PowerPC  
Architecture. This microprocessor, also called the  
PowerPC 970FX, includes a Vector/SIMD facility  
which supports high-bandwidth data processing and  
algorithmic-intensive computations. This  
microprocessor is also designed to support multiple  
system organizations, including desktop and low-  
end server applications, and uniprocessor up  
through four-way SMP configurations.  
Figure 1-1 on page 13 is a block diagram of the  
PowerPC 970FX.  
The PowerPC 970FX is comprised of three main  
components:  
• PowerPC 970FX Core which includes VMX  
execution units  
• PowerPC 970FX Storage subsystem which  
includes core interface logic, non-cacheable  
unit, L2 cache and controls, and the Bus  
Interface Unit  
• PowerPC 970FX Pervasive Functions  
Note: The terms microprocessor and processor are  
used interchangeably in this document.  
This document also provides pertinent physical  
characteristics of the PowerPC 970FX single chip  
modules (SCM).  
1.2 Features  
• 64-bit implementation of the PowerPC  
Architecture (Version 2.01)  
• Dynamic instruction cracking for some  
instructions allows for simpler inner core  
dataflow  
• Binary compatibility for all PowerPC  
application level code (problem state)  
• Binary compatibility for all PowerPC  
application level code (problem state)  
• Support for 32-bit O/S bridge facility  
• Vector/SIMD Multimedia eXtension  
- Dedicated dataflow for cracking one  
instruction into two internal operations  
- Microcoded templates for longer  
emulation sequences  
• Speculative superscalar inner core organization  
• Aggressive branch prediction  
- Prediction for up to two branches per  
cycle  
• Layered implementation strategy for very high  
frequency operation  
• Deeply pipelined design  
- 16 stages for most fixed-point register-  
register operations  
- Support for up to 16 predicted branches  
in flight  
- 18 stages for most load and store  
operations (assuming L1 Dcache hit)  
- 21 stages for most floating point  
operations  
- Prediction support for branch direction  
and branch addresses  
• Out of order issue of up to ten operations  
into 10 execution pipelines  
- 19, 22, and 25 stages for fixed-point,  
complex-fixed, and floating point  
operations, respectively in the VALU.  
- 19 stages for VMX permute operations  
- Two load or store operations  
- Two fixed-point register-register  
operations  
- Two floating-point operations  
- One branch operation  
- One condition register operation  
- One VMX permute operation  
- One VMX ALU operation  
General Information  
Page 11 of 68  
July 15, 2005