[理工] 計組 99 100台大電機

看板Grad-ProbAsk作者ken52011219 (呱)時間9年前 (2016/11/24 09:43)推噓4(4推 0噓 11→)

留言15則, 3人參與討論串1/1

大家好 ,手邊有兩題有點疑問台大電機 100 (10)Forwarding is a common technique th eliminate data hazard occuring among pipelining instruction. 　　However,not all data hazard can be eliminate by forwarding. Support on a particular machine,if an instruction following a load instr -uction depends on the result of the load instruction,then a data hazard has occured , and pipeline is stalled for one cycle Which of following statements are not correct? (a) In addition to data hazard,there can be control hazard in this situation because we cannot determine the proper data to load in the MEM pipeline stage. 應改為 ID pipeline stage (b) Assume the percentage of load instruction is 20% in a program , and half the time the instruction following a load needs the result of the load instruction. Then , the performance degradtion due to the data hazard is 1.1 1 + 0.2* (1/2) Performance = ────── = 1.1 1 performance degradtion due to the data hazard = 1.1 - 1 = 0.1 (e) All of the above are correct. Ans: a,b,e 台大電機 99 Which of the following statement are true ? (a) If the data requested by a load instruction has not yet become availble when it is requested , the load-use data hazard occur. when its next and dependency instruction is requested (b) Data hazard may be resolved with bypassing technique which retrieve the hitting data element from internal buffers rather than waitting for it to arrive from external memory hitting 應改為 missing (c) If the proper instruction cannot execute in proper clock cycle because the instruction that was fetched is not that is needed , the data hazard occurs . control hazard -- 　　有一個香錦囊，是從一個神話般的守軍的血屍頂上剝下的。那一次我們部隊遭受從未有過的頑強抵抗，我們犧牲了三個艦隊，一個裝甲師和無以數計小組推進的敢死排，才摧毀了那處隘口的碉堡。但是竟然發現，使我們遭受如此慘烈傷亡的守軍，總數只有一人。　　士兵們起鬨地在他胸前發現這枚香袋，大家都相信這是一枚具有神奇力量的護身符。我們把他的頭顱砍斷，取下香袋，剝開，　　裡面一張被血浸紅的宣紙竟用漢字娟娟秀秀四個整齊的楷書寫著－「盼君早歸。」 -- ※ 發信站: 批踢踢實業坊(ptt.cc), 來自: 36.224.3.107 ※ 文章網址: https://www.ptt.cc/bbs/Grad-ProbAsk/M.1479951786.A.5F9.html 100 (a): ID ※ 編輯: ken52011219 (36.224.3.107), 11/24/2016 10:22:39

→

gary19941208

11/24 10:45, , 1^F

11/24 10:45, 1^F

→

gary19941208

11/24 10:46, , 2^F

11/24 10:46, 2^F

了解感謝

→

gary19941208

11/24 10:50, , 3^F