1 / 30

CSL718 : Superscalar Processors

CSL718 : Superscalar Processors. Speculative Execution 2nd Feb, 2006. Handling Control Dependence. Simple pipeline Branch prediction reduces stalls due to control dependence Wide issue processor Mere branch prediction is not sufficient

lajos
Download Presentation

CSL718 : Superscalar Processors

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. CSL718 : Superscalar Processors Speculative Execution 2nd Feb, 2006 Anshul Kumar, CSE IITD

  2. Handling Control Dependence • Simple pipeline • Branch prediction reduces stalls due to control dependence • Wide issue processor • Mere branch prediction is not sufficient • Instructions in the predicted path need to be fetched and EXECUTED (speculated execution) Anshul Kumar, CSE IITD

  3. What is required for speculation? • Branch prediction to choose which instructions to execute • Execution of instructions before control dependences are resolved • Ability to undo the effects of incorrectly speculated sequence • Preserving of correct behaviour under exceptions Anshul Kumar, CSE IITD

  4. Types of speculation • Hardware based speculation • done with dynamic branch prediction and dynamic scheduling • used in Superscalar processors • Compiler based speculation • done with static branch prediction and static scheduling • used in VLIW processors Anshul Kumar, CSE IITD

  5. i i x x x f x f Extending Tomasulo’s scheme for speculative execution • Introduce re-order buffer (ROB) • Add another stage – “commit” Normal execution • Issue • Execute • Write result Speculative execution • Issue • Execute • Write result • Commit Anshul Kumar, CSE IITD

  6. Extending Tomasulo’s scheme for speculative execution – contd. • Write results into ROB in the “write result” stage • Write results into register file or memory in the “commit” stage • Dependent instructions can read operands from ROB • A speculative instruction commits only if the prediction is determined to be correct • Instructions may complete execution out-of-order, but they commit in-order Anshul Kumar, CSE IITD

  7. Recall Tomasulo’s scheme ...... Anshul Kumar, CSE IITD

  8. Issue • Get next instruction from instruction queue • Check if there is a matching RS which is empty • no: structural hazard, instruction stalls • yes: issue the instruction to that RS • For each operand, check if it is available in RF • yes: put the operand in the RS • no: keep track of FU that will produce it Anshul Kumar, CSE IITD

  9. Execute • If one or more operands not available, wait and monitor CDB • When an operand becomes available, it is placed in RS • When all operands are available, start execution • Choice may need to be made if multiple instructions become ready at the same time Anshul Kumar, CSE IITD

  10. Write result • When result is available • write it on CDB and • from there into RF and relevant RSs • Mark RS as available Anshul Kumar, CSE IITD

  11. More formal description ...... Anshul Kumar, CSE IITD

  12. RS and RF fields Anshul Kumar, CSE IITD

  13. Issue • Get instruction <op, rd, rs, rt> from instruction queue • Wait until  r  RS[r].busy = no • if (RF[rs].Qi  0) {RS[r].Qj  RF[rs].Qi} else {RS[r].Vj  RF[rs].val; RS[r].Qj  0} • similarly for rt • RS[r].op  op; RS[r].busy  yes; RF[rd].Qi  r Anshul Kumar, CSE IITD

  14. Execute • Wait until RS[r].Qj = 0 and RS[r].Qk = 0 • Compute result: operation is RS[r].op, operands are RS[r].Vj and RS[r].Vk Anshul Kumar, CSE IITD

  15. Write result • Wait until execution complete at r and CDB available •  x if (RF[x].Qi = r) {RF[x].val  result; RF[x].Qi  0} •  x if (RS[x].Qj = r) {RS[x].Vj  result; RS[x].Qj  0} • similarly for Qk / Vk • RS[r].busy  no Anshul Kumar, CSE IITD

  16. Tomasulo’s scheme plus ROB...... Anshul Kumar, CSE IITD

  17. Issue • Get next instruction from instruction queue • Check if there is a matching RS which is empty and an empty slot in ROB • no: structural hazard, instruction stalls • yes: issue the instruction to that RS and mark the ROB slot, also put ROB slot number in RS • For each operand, check if it is available in RF or ROB • yes: put the operand in the RS • no: keep track of FU that will produce it Anshul Kumar, CSE IITD

  18. Execute (no change) • If one or more operands not available, wait and monitor CDB • When an operand becomes available, it is placed in RS • When all operands are available, start execution • Choice may need to be made if multiple instructions become ready at the same time Anshul Kumar, CSE IITD

  19. Write result • When result is available • write it on CDB with ROB tag and • from there into ROB RF and relevant RSs • Mark RS as available Anshul Kumar, CSE IITD

  20. Commit (non-branch instruction) • Wait until instruction reaches head of ROB • Update RF • Remove instruction from ROB Anshul Kumar, CSE IITD

  21. Commit (branch instruction) • Wait until instruction reaches head of ROB • If branch is mispredicted, • flush ROB • Restart execution at correct successor of the branch instruction • else • Remove instruction from ROB Anshul Kumar, CSE IITD

  22. More formal description ...... Anshul Kumar, CSE IITD

  23. RS fields Anshul Kumar, CSE IITD

  24. RF fields Anshul Kumar, CSE IITD

  25. ROB fields Anshul Kumar, CSE IITD

  26. Issue • Get instruction <op, rd, rs, rt> from instruction queue • Wait until r  RS[r].busy=no and ROB[b].busy=no, where b = ROB tail • if (RF[rs].busy) {h  RF[rs].Qi; if (ROB[h].rdy) {RS[r].Vj  ROB[h].val; RS[r].Qj  0} else {RS[r].Qj  h} } else {RS[r].Vj  RF[rs].val; RS[r].Qj  0} • similarly for rt • RS[r].op  op; RS[r].busy  yes; RS[r].Qi b • RF[rd].Qi b; RF[rd].busy  yes; ROB[b].busy  yes • ROB[b].inst  op; ROB[b].dst  rd; ROB[b].rdy  no Anshul Kumar, CSE IITD

  27. Execute (no change) • Wait until RS[r].Qj = 0 and RS[r].Qk = 0 • Compute result: operation is RS[r].op, operands are RS[r].Vj and RS[r].Vk Anshul Kumar, CSE IITD

  28. Write result • Wait until execution complete at r and CDB available • b  RS[r].Qi; RS[r].busy  no •  x if (RF[x].Qi = r) {RF[x]  result; RF[x].Qi  0} •  x if (RS[x].Qj = b) {RS[x].Vj  result; RS[x].Qj  0} • similarly for Qk / Vk • ROB[b].rdy  yes; ROB[b].val  result Anshul Kumar, CSE IITD

  29. Commit (non-branch instruction) • Wait until instruction reaches head of ROB (entry h) and ROB[h].rdy = yes • d  ROB[h].dst • RF[d].val  ROB[h].val • ROB[h].busy  no • if (RF[d].Qi = h) {RF[d].busy  no} Anshul Kumar, CSE IITD

  30. Commit (branch instruction) • Wait until instruction reaches head of ROB (entry h) and ROB[h].rdy = yes • If branch is mispredicted, • clear ROB, RF[ ].Qi • fetch branch dest • else • ROB[h].busy  no • if (RF[d].Qi = h) {RF[d].busy  no} Anshul Kumar, CSE IITD

More Related