[ Pobierz całość w formacie PDF ]
.1 Cache sizesL1 Instruction L1 Data L2 cachePowerPC 970 64 kB 32 kB 512 kBPentium 4 It s a secret 8 kB 512 kBFor maximum microprocessor speed we need a high clock speedcombined with the maximum use being made of every part of themicroprocessor.The early 8-bit microprocessors would accept the firstinstruction and it would pass through the microprocessor beingdecoded, then acted upon, then having the results stored before itconsidered the next instruction.This meant that each bit of the microwas doing nothing for much of the time.Modern micros load many instructions at the same time and split upthe tasks so that as many as possible can be carried out at the sametime to have the minimum time wastage.As with the Pentium 4, the PPC970 makes use of level 1 caches that,as is now common, are split into an Instruction cache and a Datacache.There is also a level 2 cache and an external level 3 cache.Loading the instructionsThe instructions pour down from the Instruction cache at a maximumrate of eight per cycle, though five is a more likely overall figure.Butthis is still fast.The PP970 uses a very long pipeline and can be handling up to 200instructions simultaneously.The price of such a long pipeline is thatwe must be careful to ensure that it is filled with the most usefulinstructions and hence we need to back it up with very effectivebranch prediction techniques.Branch predictionTo obtain the maximum possible speed, the PP970 has devoted a greatdeal of resources into its branch prediction.As the instructions areloaded, the branch prediction circuitry scans the incoming instructionlooking for branch instructions.Every time we meet a branchinstruction that offers a choice of outcome the branch will have to beaccepted or rejected.The 970 has two branch prediction methods.The first is very similar tothat used in the Pentium 4 and, to over simplify the situation, it followsthe same sort of reasoning as we often adopt in everyday life.If it190 The PowerPCusually happens, it is most likely to happen again.The 970 keeps arecord of the previous 16384 branches in its BHT (Branch HistoryTable) to see how often each choice was made and then thisinformation is further sorted by a prediction program before it comesto a final decision.The second method involves a similar sized table called a GlobalPredictor.This method also comes up with a final go/no go for thebranch but it decides by generating an 11-bit vector that stores theactual execution path taken by the previous eleven fetch groupsleading up to the branch.So there are two independent mechanisms that make a decision as towhether the branch should be taken.If they disagree, we need areferee.This job is performed by a  Selector Table that stores thesuccess rate for each of the two previous methods for each particularbranch.It then makes the final decision  and it is said (by IBM) to bevery successful, which it probably is.Handling the instructionsHaving combined the incoming instruction stream from the Instructioncache with the information from the Branch predict, the instruction arequeued and passed to the Decode, Crack and Group Formation Unit.At this stage, in order to keep the instruction handling speed at amaximum, this unit takes the instruction codes from the Instructioncache, decodes them and cracks them into their component partscalled Internal Operations (IOPs).These very small but simple tasksare passed out to specialized units like the five blocks shown along thebottom of Figure 13.5.Figure 13.5The PowerPC 970191 Introduction to Microprocessors and MicrocontrollersThe IOPs are executed in whatever order that will result in the fastestthroughput and to reduce the complexity of keeping track of theexecution of each and every one, they are organized in groups of fiveand then the groups are tracked.Of the final row shown, there are the arithmetically based block thathandle the vectors, floating point and integer calculations, the load-store that handles the transfer of data to the memory via the secondlevel cache and finally the feedback path for the branch predictioninformation.The PC market placeThe PowerPC may not be in our PC but it may well be in our car.TheFord Motor Company has elected to use the PowerPC as first choicefor their engine management computer into the next century.Quiz time 13In each case, choose the best option.1 The maximum number of instructions that thePowerPC 970 can be dealing simultaneously is:(a) 200.(b) 3.(c) 16384.(d) 128 [ Pobierz całość w formacie PDF ]

  • zanotowane.pl
  • doc.pisz.pl
  • pdf.pisz.pl
  • funlifepok.htw.pl

    •