欢迎访问ic37.com |
会员登录 免费注册
发布采购

IBM25PPC970FX6TR348ET 参数 Datasheet PDF下载

IBM25PPC970FX6TR348ET图片预览
型号: IBM25PPC970FX6TR348ET
PDF下载: 下载PDF文件 查看货源
内容描述: [RISC Microprocessor, 64-Bit, 2000MHz, CMOS, CBGA576, 25 X 25 MM, 1 MM PITCH, ROHS COMPLIANT, CERAMIC, BGA-576]
分类和应用: 时钟外围集成电路
文件页数/大小: 74 页 / 3280 K
品牌: IBM [ IBM ]
 浏览型号IBM25PPC970FX6TR348ET的Datasheet PDF文件第8页浏览型号IBM25PPC970FX6TR348ET的Datasheet PDF文件第9页浏览型号IBM25PPC970FX6TR348ET的Datasheet PDF文件第10页浏览型号IBM25PPC970FX6TR348ET的Datasheet PDF文件第11页浏览型号IBM25PPC970FX6TR348ET的Datasheet PDF文件第13页浏览型号IBM25PPC970FX6TR348ET的Datasheet PDF文件第14页浏览型号IBM25PPC970FX6TR348ET的Datasheet PDF文件第15页浏览型号IBM25PPC970FX6TR348ET的Datasheet PDF文件第16页  
Data Sheet  
PowerPC 970FX  
Preliminary  
• Dynamic instruction cracking for some instructions allows for simpler inner core dataflow  
- 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  
- Support for up to 16 predicted branches in flight  
- Prediction support for branch direction and branch addresses  
• Out of order issue of up to ten operations into 10 execution pipelines  
- 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  
• In order dispatch of up to five operations into distributed issue queue structure  
• Register renaming on GPRs, FPRs, VRFs, CR Fields, XER (parts), FPSCR, VSCR, VRSAVE, Link  
and Count  
• Large number of instructions in flight (theoretical maximum of 215 instructions)  
- Up to 16 instructions in instruction fetch unit (fetch buffer and overflow buffer)  
- Up to 32 instructions in instruction fetch buffer in instruction decode unit  
- Up to 35 instructions in 3 decode pipe stages and 4 dispatch buffers  
- Up to 100 instructions in the inner-core (after dispatch)  
- Up to 32 stores queued in the STQ (available for forwarding)  
• Fast, selective flush of incorrect speculative instructions and results  
• Specific focus on storage latency management  
• Out-of-order and speculative issue of load operations  
• Support for up to 8 outstanding L1 cache line misses  
• Hardware initiated instruction prefetching from L2 cache  
• Software initiated data stream prefetching  
- Support for up to 8 active streams  
• Critical word forwarding / critical sector first  
• New branch processing / prediction hints added to branch instructions  
• Power management  
• Static power management  
- Software initiated doze and nap  
• Dynamic power management  
- Parts of the design stop their (hardware initiated) clocks when not in use  
• PowerTune  
- Software initiated slow down of the processor; selectable to half of the nominal operating  
frequency  
General Information  
October 14, 2005  
Page 12 of 74