Difference between revisions of "InOrder ToDo List"

Revision as of 17:42, 12 January 2010

Python Configurability

InOrderCPU configuration (uniprocessor options)

   * MTApproach - Multi-Threading (MT) Approach
   * Resource - Resource Configuration (will eventually be used to autogenerate C++)

         * ResourceType - Type of resource (Enum type)
               * ResourceParams - Parameters for this type of resource
               * Request - List of requests for this type of resource (Enum type)
                     * Latency - operation latency and issue latency (intra/inter thread)
               * Count - Number of such resource type

   * PipelineDesc - Pipeline Description

         * InstSchedule - Instruction schedule specified as a vector of InstClassSchedule
               * InstClassSchedule - Vector of schedules per instruction class - load/store, Int execute, FP execute, specialized inst, etc. (do we still want a distinction between front end and back end schedules?)
                     * ResourceRequestList - Vector of ResourceRequest (per stage, right?
                           * ResourceRequest - Vector of requests for resources

   * Other params

MTApproach options

   * None (single threaded)
   * Fine-grained (switch context every cycle or every few cycles, like Ultrasparc T2)
   * Coarse-grained (switch context on thread stalls, like 'SwitchOnCacheMiss' currently)
   * SMT (all contexts active, like 'SMT' currently)

ResourceType/Request options

   * FetchUnit
         o AssignNextPC
         o UpdateTargetPC
         o SelectThread (for fine-grained and coarse grained MT approaches; may also be done in a separate thread selection unit or pick unit as in Ultrasparc T2)
         o Any other fetch address generation?
   * MemPort (DataReadPort, DataWritePort, FetchPort)
         o Access
         o InitiateAccess
         o CompleteAccess
   * DecodeUnit
         o DecodeInst
   * BPredUnit
         o Predict
         o Update
   * UseDefUnit
         o ReadReg
         o WriteReg
   * AgenUnit
         o GenerateAddr
   * ExecUnit
         o Exec
         o InitiateExec
         o CompleteExec
         o GenerateAddr
   * GradUnit
         o Graduate
   * Interstage buffers (only for SMT?!)

Simulation Speed

   * Instruction Schedule Work
         o Use Vector of Vectors instead of Priority Queue
         o Identify Instruction Schedule Types (via Tuple)
         o Cache Instruction Schedule, Generate On-Demand
         o Instructions walk through schedule by incrementing pointer instead of popping from queue
         o If dynamic schedule is needed, then copy the remaining part of schedule and let the instruction add/remove as it pleases
               + Better solution here?
   * Event-Sleeping Work
         o Sleep instructions waiting for an long-delay event
         o Sleep CPU w/no activity (partially implemented)