Sorry for the confusion. I mean Reorder Buffer Entries. :)

Mansour Shafaei Moghaddam <mansoor.shafaei <at> gmail.com> writes:

> 
> 
> Hi
> 
> Could you please let me know in which of the following state(s), a row 
buffer entry can be distinguished as a mis-speculated one? What is the last 
state a speculative row buffer entry can reside(switch to) before getting 
the acknowledge that the speculation was correct and it can go further (or 
be committed)? 
> 
> Thanks 
> 
> 
>  StateList rob_free_list;                             // Free ROB entyry
>  StateList rob_frontend_list;                         // Frontend in 
progress (artificial delay)
>  StateList rob_ready_to_dispatch_list;                // Ready to dispatch
>  StateList rob_dispatched_list[MAX_CLUSTERS];         // Dispatched but 
waiting for operands
>  StateList rob_ready_to_issue_list[MAX_CLUSTERS];     // Ready to issue 
(all operands ready)
>  StateList rob_ready_to_store_list[MAX_CLUSTERS];     // Ready to store 
(all operands except possibly rc are ready)
>  StateList rob_ready_to_load_list[MAX_CLUSTERS];      // Ready to load 
(all operands ready)
>  StateList rob_issued_list[MAX_CLUSTERS];             // Issued and in 
progress (or for loads, returned here after address is generated)
>  StateList rob_completed_list[MAX_CLUSTERS];          // Completed and 
result in transit for local and global forwarding
>  StateList rob_ready_to_writeback_list[MAX_CLUSTERS]; // Completed; result 
ready to writeback in parallel across all cluster register files
>  StateList rob_cache_miss_list;                       // Loads only: wait 
for cache miss to be serviced
>  StateList rob_tlb_miss_list;                         // TLB miss waiting 
to be serviced on one or more levels
>  StateList rob_memory_fence_list;                     // mf uops only: 
wait for memory fence to reach head of LSQ before completing
>  StateList rob_ready_to_commit_queue;                 // Ready to commit
> 
> 
> <div><div dir="ltr">
> <div>Hi</div>
> <div><br></div>
> <div>Could you please let me know in which of the following state(s), a 
row buffer entry can be distinguished as a mis-speculated one? 
What is the last state a speculative row buffer entry can reside(switch to) 
before getting the acknowledge that the speculation was correct 
and it can go further (or be committed)? </div>
> <div><br></div>
> <div>Thanks </div>
> <div><br></div>
> <div><br></div>
> <div> StateList rob_free_list;           
                  // Free ROB 
entyry</div>
> <div> StateList rob_frontend_list;           
              // Frontend in progress 
(artificial delay)</div>
> <div> StateList rob_ready_to_dispatch_list;       
         // Ready to dispatch</div>
> <div> StateList rob_dispatched_list[MAX_CLUSTERS];     
    // Dispatched but waiting for operands</div>
> <div> StateList rob_ready_to_issue_list[MAX_CLUSTERS];     
// Ready to issue (all operands ready)</div>
> <div> StateList rob_ready_to_store_list[MAX_CLUSTERS];     
// Ready to store (all operands except possibly rc are ready)</div>
> <div> StateList rob_ready_to_load_list[MAX_CLUSTERS];     
 // Ready to load (all operands ready)</div>
> <div> StateList rob_issued_list[MAX_CLUSTERS];       
      // Issued and in progress (or for loads, returned here 
after address is generated)</div>
> <div> StateList rob_completed_list[MAX_CLUSTERS];     
     // Completed and result in transit for local and global 
forwarding</div>
> <div> StateList rob_ready_to_writeback_list[MAX_CLUSTERS]; // 
Completed; result ready to writeback in parallel across all cluster register 
files</div>
> <div> StateList rob_cache_miss_list;         
              // Loads only: wait for 
cache miss to be serviced</div>
> <div> StateList rob_tlb_miss_list;           
              // TLB miss waiting to be 
serviced on one or more levels</div>
> <div> StateList rob_memory_fence_list;         
            // mf uops only: wait for memory 
fence to reach head of LSQ before completing</div>
> <div> StateList rob_ready_to_commit_queue;       
          // Ready to commit</div>
> </div></div>
> 





_______________________________________________
http://www.marss86.org
Marss86-Devel mailing list
[email protected]
https://www.cs.binghamton.edu/mailman/listinfo/marss86-devel

Reply via email to