CPU time Seconds Instructions Cycles Seconds x Program
- Slides: 50
Παράγοντες που επηρεάζουν την επίδοση της CPU time = Seconds = Instructions Cycles Seconds ---------------- x -------Program Instruction Cycle Instr. count CPI Clock rate Program Χ Compiler Χ Χ Instruction Set Architecture (ISA) Οργάνωση X Χ Χ X Χ Τεχνολογία Χ
5 -Stage Pipelined Datapath Instruction Fetch Execute Addr. Calc Instr. Decode Reg. Fetch Next SEQ PC Adder Zero? RS 1 Imm IR ← Mem[PC] RD RD RD MUX Sign Extend MEM/WB Data Memory EX/MEM MUX ALU MUX ID/EX Reg File IF/ID Memory Address RS 2 WB Data 4 Write Back MUX Next PC Memory Access PC ← PC + 4 A ← Reg[IRrs] rslt ← B ← Reg[IRrt] A op. IRop B Mem[rslt] Reg[IRrd]←WB
Διάγραμμα χρονισμού Time (clock cycles) Reg DMem Ifetch Reg ALU O r d e r Ifetch ALU I n s t r. Cycle 6 Cycle 7 ALU Cycle 1 Cycle 2 Cycle 3 Cycle 4 Cycle 5 Reg Reg DMem Reg
Παράδειγμα για την εντολή lw: Instruction Fetch (IF) Instruction fetch 0 M u x 1 IF/ID ID/EX EX/MEM MEM/WB Add Add result 4 PC Address Instruction memory Instruction Shift left 2 Read register 1 Read data 1 Read register 2 Registers Read Write data 2 register Write data 0 M u x 1 Zero ALU result Address Data memory Write data 16 Sign extend 32 Read data 1 M u x 0
Παράδειγμα για την εντολή lw: Instruction Decode (ID) Instruction decode 0 M u x 1 IF/ID ID/EX EX/MEM MEM/WB Add Add result 4 PC Address Instruction memory Instruction Shift left 2 Read register 1 Read data 1 Read register 2 Registers Read Write data 2 register Write data 0 M u x 1 Zero ALU result Address Data memory Write data 16 Sign extend 32 Read data 1 M u x 0
Παράδειγμα για την εντολή lw: Execution (EX) Execution 0 M u x 1 IF/ID ID/EX EX/MEM MEM/WB Add Add result 4 PC Address Instruction memory Instruction Shift left 2 Read register 1 Read data 1 Read register 2 Registers Read Write data 2 register Write data 0 M u x 1 Zero ALU result Address Data memory Write data 16 Sign extend 32 Read data 1 M u x 0
Παράδειγμα για την εντολή lw: Memory (MEM) Memory 0 M u x 1 IF/ID ID/EX EX/MEM MEM/WB Add Add result 4 PC Address Instruction memory Instruction Shift left 2 Read register 1 Read data 1 Read register 2 Registers Read Write data 2 register Write data 0 M u x 1 Zero ALU result Address Data memory Write data 16 Sign extend 32 Read data 1 M u x 0
Παράδειγμα για την εντολή lw: Writeback (WB) Writeback 0 M u x 1 IF/ID ID/EX EX/MEM MEM/WB Add Add result 4 PC Address Instruction memory Instruction Shift left 2 Read register 1 Read data 1 Read register 2 Registers Read Write data 2 register Write data 0 M u x 1 Zero ALU result Address Data memory Write data 16 Sign extend 32 Read data 1 M u x 0
Παράδειγμα structural hazard: ένα διαθέσιμο memory port Time (clock cycles) Instr 2 Instr 3 Instr 4 DMem Ifetch Reg ALU Instr 1 Reg ALU Ifetch ALU O r d e r Load ALU I n s t r. Cycle 6 Cycle 7 ALU Cycle 1 Cycle 2 Cycle 3 Cycle 4 Cycle 5 Reg Reg DMem Reg
Παράδειγμα structural hazard: επίλυση με stall Time (clock cycles) Instr 1 Instr 2 Stall Instr 3 Reg DMem Ifetch Reg ALU Ifetch Bubble Cycle 6 Cycle 7 Reg Bubble Ifetch Reg DMem Bubble Reg Bubble ALU O r d e r Load ALU I n s t r. ALU Cycle 1 Cycle 2 Cycle 3 Cycle 4 Cycle 5 Bubble DMem Reg
Παράδειγμα data hazard στον r 1 Time (clock cycles) or r 8, r 1, r 9 xor r 10, r 11 DMem Ifetch Reg ALU and r 6, r 1, r 7 Reg ALU O r d e r sub r 4, r 1, r 3 Ifetch WB ALU add r 1, r 2, r 3 MEM ALU I n s t r. ID/RF EX ALU IF Reg Reg DMem Reg
Προώθηση DMem Ifetch Reg ALU and r 6, r 1, r 7 Reg ALU sub r 4, r 1, r 3 Ifetch ALU O r d e r add r 1, r 2, r 3 ALU I n s t r. ALU Time (clock cycles) or r 8, r 1, r 9 xor r 10, r 11 • μειώνονται τα RAW hazards Reg Reg DMem Reg
Αλλαγές στο hardware για την υποστήριξη προώθησης Next. PC mux MEM/WR EX/MEM ALU mux ID/EX Registers Data Memory mux Immediate Προώθηση ΕΧ->ΕΧ, ΜΕΜ->ΕΧ
Προώθηση ΜΕΜ->ΜΕΜ or r 8, r 6, r 9 xor r 10, r 9, r 11 DMem Ifetch Reg ALU sw r 4, 12(r 1) Reg ALU lw r 4, 0(r 1) Ifetch ALU O r d e r add r 1, r 2, r 3 ALU I n s t r. ALU Time (clock cycles) Reg Reg DMem Reg
Τα data hazards δεν εξαφανίζονται πλήρως με την προώθηση and r 6, r 1, r 7 or r 8, r 1, r 9 Reg DMem Ifetch Reg Reg DMem ALU O r d e r sub r 4, r 1, r 6 Ifetch ALU lw r 1, 0(r 2) ALU I n s t r. ALU Time (clock cycles) Reg DMem Reg
Τα data hazards δεν εξαφανίζονται πλήρως με την προώθηση and r 6, r 1, r 7 or r 8, r 1, r 9 Reg DMem Ifetch Reg Bubble DMem Ifetch Bubble Reg Bubble Ifetch Reg Reg DMem ALU O r d e r sub r 4, r 1, r 6 Ifetch ALU lw r 1, 0(r 2) ALU I n s t r. ALU Time (clock cycles) Reg DMem
Αναδιάταξη εντολών για την αποφυγή RAW hazards Πώς μπορούμε να παράξουμε γρηγορότερο κώδικα assembly για τις ακόλουθες πράξεις? a = b + c; d = e – f; Slow code: LW LW ADD SW LW LW SUB SW Rb, b Rc, c Ra, Rb, Rc a, Ra Re, e Rf, f Rd, Re, Rf d, Rd Fast code: LW LW LW ADD LW SW SUB SW Rb, b Rc, c Re, e Ra, Rb, Rc Rf, f a, Ra Rd, Re, Rf d, Rd
36: xor r 10, r 11 Reg DMem Ifetch Reg ALU 22: add r 8, r 1, r 9 Ifetch ALU 18: or r 6, r 1, r 7 Reg ALU 14: and r 2, r 3, r 5 Ifetch ALU 10: beq r 1, r 3, 36 ALU Κίνδυνοι ελέγχου στις εντολές διακλάδωσης: stalls 3 σταδίων DMem Reg Reg DMem Reg
Τροποποιήσεις στο pipeline Instruction Fetch Next SEQ PC Memory Access Write Back Adder MUX Next PC Zero? RS 1 MUX Sign Extend RD MEM/WB Data Memory EX/MEM ALU MUX ID/EX Imm Reg File IF/ID Memory Address RS 2 WB Data 4 Execute Addr. Calc Instr. Decode Reg. Fetch RD RD
4 εναλλακτικές προσεγγίσεις για την αντιμετώπιση των control hazards #4: Delayed Branches branch instruction sequential successor 1 sequential successor 2. . . . sequential successorn branch target if taken Branch delay μήκους n: οι εντολές εκτελούνται είτε το branch είναι Taken είτε όχι – delay ενός slot: επιτρέπει απόφαση και υπολογισμό διεύθυνσης-στόχου στο 5 -stage pipeline χωρίς stalls
Εναλλακτικά. . . Pipeline CPI = υπερβαθμωτή εκτέλεση Ideal pipeline CPI + register renaming δυναμική εκτέλεση loop unrolling static scheduling, software pipelining προώθηση Structural Stalls + Data Hazard Stalls + Control Stalls πρόβλεψη διακλαδώσεων υποθετική εκτέλεση delayed branches, branch scheduling
Παράδειγμα Clock Cycles MUL. D ADD. D L. D S. D 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 IF ID M 1 M 2 M 3 M 4 M 5 M 6 M 7 M WB IF ID A 1 A 2 A 3 A 4 M WB IF ID EX M WB
RAW hazards και αύξηση των stalls Clock Cycles L. D F 4, 0(R 2) MUL. D F 0, F 4, F 6 ADD. D F 2, F 0, F 8 S. D F 2, 0(R 2) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 IF ID EX M WB IF ID S M 1 M 2 M 3 M 4 M 5 M 6 M 7 M WB IF S ID S S S A 1 A 2 A 3 A 4 M WB IF S S S ID EX S S S M WB • full bypassing/forwarding • η S. D πρέπει να καθυστερήσει έναν κύκλο παραπάνω για να αποφύγουμε το conflict στο ΜΕΜ της ADD. D
Απόδοση του multi-cycle FP pipeline • stalls ανά εντολή + breakdown • από 0. 65 μέχρι 1. 21 stalls ανά εντολή • κυριαρχούν τα RAW hazard stalls ( «FP result stalls» )
Case-study: MIPS R 4000 Pipeline IS Instruction Memory RF Reg EX ALU IF DF DS Data Memory TC WB Reg Branch target and condition eval. • Deeper Pipeline (superpipelining): επιτρέπει υψηλότερα clock rates • Fully pipelined memory accesses (2 cycle delays για loads) • Predicted-Not-Taken πολιτική – Not-taken (fall-through) branch : 1 delay slot – Taken branch: 1 delay slot + 2 idle cycles
Case-study: MIPS R 4000 Pipeline IF IS Instruction Memory complete ICache access decode, register read, hazard checking, ICache hit detection RF 1 st half of DCache access EX Reg ALU PC selection, initiation of ICache access effective address calculation, ALU operation, branchtarget computation, condition evaluation DF Dcache hit detection DS Data Memory complete DCache access TC WB Reg register write-back
Branch delay (3 cycles)
- Cpu instructions
- Cpu instructions
- Keyboard mouse scanner and microphone are blank devices
- Teknik yang digunakan cpu dalam menangani program interupsi
- In cpu scheduling the clock is of type
- For minutes. start.
- Program control instruction
- 60 seconds times 60 minutes
- Mandarin cycles
- Alliteration words for cycle
- Biogeochemical cycles of water
- Pdsa cycles
- Precession milankovitch cycles
- Rvguev suicide
- Relationship between transaction cycles
- Cpi cycles per instruction
- Biogeochemical cycles performance task
- Apes cycles
- What is the major abiotic reservoir for phosphorus
- What happens to phosphorus that erodes from rock and soil?
- Enam securities portfolio
- Joke cycles
- Joke cycles
- Biogeochemical cycles class 9 ppt
- Sleep wake cycle
- Biogeochemical cycles foldable
- Design science research cycles
- Compare and contrast carbon and nitrogen cycles
- Biogeochemical cycles poster
- Apa itu pdsa
- Alphonse mucha advertising posters
- Management performance cycle
- Water cycles of matter
- Isagenix presidents pack
- Multiplication of cycles
- Cpi in cpu
- Tides
- Kiwi bird life cycle
- 4 major biogeochemical cycles
- Pearson
- Section 26.3 life cycles of stars
- Mandarin cycles
- Tsw cycles
- Business cycles economics
- Stockspotter.com
- How to calculate real gdp per person
- Paragraph on global warming
- Multiple procurement cycles
- Biogeochemical cycles
- Generalization in database
- Milankovitch cycles