[ { "key": "S98E6CQQ", "version": 7, "library": { "type": "group", "id": 167322, "name": "InvestigacionAdvanced Branch Prediction", "links": { "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction", "type": "text/html" } } }, "links": { "self": { "href": "https://api.zotero.org/groups/167322/items/S98E6CQQ", "type": "application/json" }, "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction/items/S98E6CQQ", "type": "text/html" } }, "meta": { "createdByUser": { "id": 1378368, "username": "sergiogq", "name": "Sergio", "links": { "alternate": { "href": "https://www.zotero.org/sergiogq", "type": "text/html" } } }, "creatorSummary": "Cliff et al.", "numChildren": 0 }, "data": { "key": "S98E6CQQ", "version": 7, "itemType": "document", "title": "A Comparative Analysis of Schemes for Correlated Branch Prediction", "creators": [ { "creatorType": "author", "firstName": "Young", "lastName": "Cliff" }, { "creatorType": "author", "firstName": "Gloy", "lastName": "Nicolas" }, { "creatorType": "author", "firstName": "Smith", "lastName": "Michael D." } ], "abstractNote": "Modern high-performance architectures require extremely accurate\nbranch prediction to overcome the performance limitations of con-\nditional branches. We present a framework that categorizes branch\nprediction schemes by the way in which they partition dynamic\nbranches and by the kind of predictor that they use. The framework\nallows us to compare and contrast branch prediction schemes, and\nto analyze why they work. We use the framework to show how a\nstatic correlated branch prediction scheme increases branch bias\nand thus improves overall branch prediction accuracy. We also use\nthe framework to identify the fundamental differences between\nstatic and dynamic correlated branch prediction schemes. This\nstudy shows that there is room to improve the prediction accuracy\nof existing branch prediction schemes.", "type": "", "date": "", "publisher": "", "place": "", "DOI": "", "citationKey": "", "url": "http://www.cs.cmu.edu/afs/cs/academic/class/15740-f02/public/doc/discussions/uniprocessors/branch_pred/msmith_isca95.pdf", "accessDate": "", "archive": "", "archiveLocation": "", "shortTitle": "", "language": "", "libraryCatalog": "", "callNumber": "", "rights": "", "extra": "", "tags": [], "collections": [ "C2662ZQB" ], "relations": {}, "dateAdded": "2013-05-17T08:04:06Z", "dateModified": "2013-05-17T08:04:06Z" } }, { "key": "WFP4I2IW", "version": 7, "library": { "type": "group", "id": 167322, "name": "InvestigacionAdvanced Branch Prediction", "links": { "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction", "type": "text/html" } } }, "links": { "self": { "href": "https://api.zotero.org/groups/167322/items/WFP4I2IW", "type": "application/json" }, "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction/items/WFP4I2IW", "type": "text/html" } }, "meta": { "createdByUser": { "id": 1378368, "username": "sergiogq", "name": "Sergio", "links": { "alternate": { "href": "https://www.zotero.org/sergiogq", "type": "text/html" } } }, "creatorSummary": "Daniel A.", "numChildren": 0 }, "data": { "key": "WFP4I2IW", "version": 7, "itemType": "document", "title": "Fast Path-Based Neural Branch Prediction", "creators": [ { "creatorType": "author", "firstName": "Jim ́enez", "lastName": "Daniel A." } ], "abstractNote": "Microarchitectural prediction based on neural learning\nhas received increasing attention in recent years. However,\nneural prediction remains impractical because its superior\naccuracy over conventional predictors is not enough to off-\nset the cost imposed by its high latency. We present a new\nneural branch predictor that solves the problem from both\ndirections: it is both more accurate and much faster than\nprevious neural predictors. Our predictor improves accu-\nracy by combining path and pattern history to overcome\nlimitations inherent to previous predictors. It also has much\nlower latency than previous neural predictors. The result is\na predictor with accuracy far superior to conventional pre-\ndictors but with latency comparable to predictors from in-\ndustrial designs. Our simulations show that a path-based\nneural predictor improves the instructions-per-cycle (IPC)\nrate of an aggressively clocked microarchitecture by 16%\nover the original perceptron predictor.", "type": "", "date": "", "publisher": "", "place": "", "DOI": "", "citationKey": "", "url": "http://www.microarch.org/micro36/html/pdf/jimenez-FastPath.pdf", "accessDate": "", "archive": "", "archiveLocation": "", "shortTitle": "", "language": "", "libraryCatalog": "", "callNumber": "", "rights": "", "extra": "", "tags": [], "collections": [ "C2662ZQB" ], "relations": {}, "dateAdded": "2013-05-17T08:04:06Z", "dateModified": "2013-05-17T08:04:06Z" } }, { "key": "9QSZIMPF", "version": 7, "library": { "type": "group", "id": 167322, "name": "InvestigacionAdvanced Branch Prediction", "links": { "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction", "type": "text/html" } } }, "links": { "self": { "href": "https://api.zotero.org/groups/167322/items/9QSZIMPF", "type": "application/json" }, "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction/items/9QSZIMPF", "type": "text/html" } }, "meta": { "createdByUser": { "id": 1378368, "username": "sergiogq", "name": "Sergio", "links": { "alternate": { "href": "https://www.zotero.org/sergiogq", "type": "text/html" } } }, "creatorSummary": "Tse Yu and Yale N.", "numChildren": 0 }, "data": { "key": "9QSZIMPF", "version": 7, "itemType": "document", "title": "Two-Level Adaptive Training Branch Prediction", "creators": [ { "creatorType": "author", "firstName": "Yeh", "lastName": "Tse Yu" }, { "creatorType": "author", "firstName": "Patt", "lastName": "Yale N." } ], "abstractNote": "High-performance microarchitectures use, among other structures, deep pipelines to help speed up exe-\ncution. The importance of a good branch predictor to the effectiveness of a deep pipeline in the presence of condi-\ntional branches is well-known. In fact, the literature contains proposals for a number of branch prediction schemes.\nSome are static in that they use opcode information and profiling statistics to make predictions. Others are dynamic\nin that they use run-time execution history to make predictions.\nThis paper proposes a new dynamic branch predictor, the Two-Level Adaptive Paining scheme, which alters the\nbranch prediction algorithm on the basis of information collected at run-time.\nSeveral configurations of the Two-Level Adaptive Training Branch Predictor are introduced, simulated, and\ncompared to simulations of other known static and dynamic branch prediction schemes. Two-Level Adaptive\nTraining Branch Prediction achieves 97 percent accuracy on nine of the ten SPEC benchmarks, compared to less\nthan 93 percent for other schemes. Since a prediction miss requires flushing of the speculative execution already in\nprogress, the relevant metric is the miss rate. The miss rate is 3 percent for the Two-Level Adaptive Training scheme\nvs. 7 percent (best case) for the other schemes. This represents more than a 100 percent improvement in reducing\nthe number of pipeline hushes required.", "type": "", "date": "", "publisher": "", "place": "", "DOI": "", "citationKey": "", "url": "http://www.inf.pucrs.br/~calazans/graduate/SDAC/saltos.pdf", "accessDate": "", "archive": "", "archiveLocation": "", "shortTitle": "", "language": "", "libraryCatalog": "", "callNumber": "", "rights": "", "extra": "", "tags": [], "collections": [ "C2662ZQB" ], "relations": {}, "dateAdded": "2013-05-17T08:04:06Z", "dateModified": "2013-05-17T08:04:06Z" } }, { "key": "8Q5KRKUD", "version": 7, "library": { "type": "group", "id": 167322, "name": "InvestigacionAdvanced Branch Prediction", "links": { "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction", "type": "text/html" } } }, "links": { "self": { "href": "https://api.zotero.org/groups/167322/items/8Q5KRKUD", "type": "application/json" }, "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction/items/8Q5KRKUD", "type": "text/html" } }, "meta": { "createdByUser": { "id": 1378368, "username": "sergiogq", "name": "Sergio", "links": { "alternate": { "href": "https://www.zotero.org/sergiogq", "type": "text/html" } } }, "creatorSummary": "Tse Yu and Yale N.", "numChildren": 0 }, "data": { "key": "8Q5KRKUD", "version": 7, "itemType": "document", "title": "A Comparison of Dynamic Branch Predictors that use Two Levels of Branch History", "creators": [ { "creatorType": "author", "firstName": "Yeh", "lastName": "Tse Yu" }, { "creatorType": "author", "firstName": "Patt", "lastName": "Yale N." } ], "abstractNote": "Recent attention to speculative execution as a mechanism for increasing performance of single instruction streams has demanded substantially better branch prediction than what has been previously available. We [1, 2] and Pan, So, and Rahmeh [4] have both proposed variations of the same aggressive dynamic branch predictor for handling those needs. We call the basic model Two-Level Adaptive Branch Prediction; Pan, So, and Rahmeh call it Correlation Branch Prediction. In this paper, we adopt the terminology of [2] and show that there are really nine variations of the same basic model. We compare the nine variations with respect to the amount of history information kept. We study the effects of different branch history lengths and pattern history table configurations. Finally, we evaluate the cost effectiveness of the nine variations.", "type": "", "date": "", "publisher": "", "place": "", "DOI": "", "citationKey": "", "url": "", "accessDate": "", "archive": "", "archiveLocation": "", "shortTitle": "", "language": "", "libraryCatalog": "", "callNumber": "", "rights": "", "extra": "", "tags": [], "collections": [ "C2662ZQB" ], "relations": {}, "dateAdded": "2013-05-17T08:04:06Z", "dateModified": "2013-05-17T08:04:06Z" } }, { "key": "KBBUQT55", "version": 7, "library": { "type": "group", "id": 167322, "name": "InvestigacionAdvanced Branch Prediction", "links": { "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction", "type": "text/html" } } }, "links": { "self": { "href": "https://api.zotero.org/groups/167322/items/KBBUQT55", "type": "application/json" }, "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction/items/KBBUQT55", "type": "text/html" } }, "meta": { "createdByUser": { "id": 1378368, "username": "sergiogq", "name": "Sergio", "links": { "alternate": { "href": "https://www.zotero.org/sergiogq", "type": "text/html" } } }, "creatorSummary": "Chih-Chieh et al.", "numChildren": 0 }, "data": { "key": "KBBUQT55", "version": 7, "itemType": "document", "title": "The Bi-Mode Branch Predictor", "creators": [ { "creatorType": "author", "firstName": "Lee,", "lastName": "Chih-Chieh" }, { "creatorType": "author", "firstName": "Chen,", "lastName": "I-Cheng K." }, { "creatorType": "author", "firstName": "Mudge", "lastName": "Trevor N." } ], "abstractNote": "Dynamic branch predictors are popular because they\ncan deliver accurate branch prediction without changes to\nthe instruction set architecture or pre-existing binaries.\nHowever, to achieve the desiredprediction accuracy, existing\ndynamic branch predictors require considerable\namounts of hardware to minimize the integerence effects\ndue to aliasing in the prediction tables. We propose a new\ndynamic predictor, the bi-mode predictor, which divides the\nprediction tables into two halves and, by dynamically determining\nthe current “mode” of the program, selects the appropriate\nhalf of the table for prediction. This approach is\nshown to preserve the merits of global history basedprediction\nwhile reducing destructive aliasing and, as a result, improving\nprediction accuracy. Moreover, it is simple enough\nthat it does not impact a processor’s cycle time. We conclude\nby conducting a comprehensive study into the mechanism\nunderlying two-level dynamic predictors and\ninvestigate the criteria for their optimal designs. The analysis\npresented provides a general framework for studying\nbranch predictors.", "type": "", "date": "", "publisher": "", "place": "", "DOI": "", "citationKey": "", "url": "", "accessDate": "", "archive": "", "archiveLocation": "", "shortTitle": "", "language": "", "libraryCatalog": "", "callNumber": "", "rights": "", "extra": "", "tags": [], "collections": [ "C2662ZQB" ], "relations": {}, "dateAdded": "2013-05-17T08:04:06Z", "dateModified": "2013-05-17T08:04:06Z" } }, { "key": "GW25V8EQ", "version": 7, "library": { "type": "group", "id": 167322, "name": "InvestigacionAdvanced Branch Prediction", "links": { "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction", "type": "text/html" } } }, "links": { "self": { "href": "https://api.zotero.org/groups/167322/items/GW25V8EQ", "type": "application/json" }, "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction/items/GW25V8EQ", "type": "text/html" } }, "meta": { "createdByUser": { "id": 1378368, "username": "sergiogq", "name": "Sergio", "links": { "alternate": { "href": "https://www.zotero.org/sergiogq", "type": "text/html" } } }, "creatorSummary": "Scott", "numChildren": 0 }, "data": { "key": "GW25V8EQ", "version": 7, "itemType": "document", "title": "Combining Branch Predictors", "creators": [ { "creatorType": "author", "firstName": "McFarling", "lastName": "Scott" } ], "abstractNote": "One of the key factors determining computer performance is the degree to\nwhich the implementation can take advantage of instruction-level paral-\nlelism. Perhaps the most critical limit to this parallelism is the presence of\nconditional branches that determine which instructions need to be executed\nnext. To increase parallelism, several authors have suggested ways of\npredicting the direction of conditional branches with hardware that uses the\nhistory of previous branches. The different proposed predictors take advan-\ntage of different observed patterns in branch behavior. This paper presents\na method of combining the advantages of these different types of predictors.\nThe new method uses a history mechanism to keep track of which predictor\nis most accurate for each branch so that the most accurate predictor can be\nused. In addition, this paper describes a method of increasing the usefulness\nof branch history by hashing it together with the branch address. Together,\nthese new techniques are shown to outperform previously known approaches\nboth in terms of maximum prediction accuracy and the prediction accuracy\nfor a given predictor size. Specifically, prediction accuracy reaches 98.1%\ncorrect versus 97.1% correct for the most accurate previously known ap-\nproach. Also, this new approach is typically at least a factor of two smaller\nthan other schemes for a given prediction accuracy. Finally, this new ap-\nproach allows predictors with a single level of history array access to outper-\nform schemes with multiple levels of history for all but the largest predictor\nsizes", "type": "", "date": "", "publisher": "", "place": "", "DOI": "", "citationKey": "", "url": "http://www.hpl.hp.com/techreports/Compaq-DEC/WRL-TN-36.pdf", "accessDate": "", "archive": "", "archiveLocation": "", "shortTitle": "", "language": "", "libraryCatalog": "", "callNumber": "", "rights": "", "extra": "", "tags": [], "collections": [ "C2662ZQB", "ZCDARHQ9" ], "relations": {}, "dateAdded": "2013-05-14T05:51:11Z", "dateModified": "2013-05-14T05:51:11Z" } }, { "key": "QDDG7A6C", "version": 7, "library": { "type": "group", "id": 167322, "name": "InvestigacionAdvanced Branch Prediction", "links": { "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction", "type": "text/html" } } }, "links": { "self": { "href": "https://api.zotero.org/groups/167322/items/QDDG7A6C", "type": "application/json" }, "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction/items/QDDG7A6C", "type": "text/html" } }, "meta": { "createdByUser": { "id": 1378368, "username": "sergiogq", "name": "Sergio", "links": { "alternate": { "href": "https://www.zotero.org/sergiogq", "type": "text/html" } } }, "creatorSummary": "Daniel A.", "numChildren": 0 }, "data": { "key": "QDDG7A6C", "version": 7, "itemType": "document", "title": "Neural Methods for Dynamic Branch Prediction", "creators": [ { "creatorType": "author", "firstName": "Jimenez", "lastName": "Daniel A." } ], "abstractNote": "This article presents a new and highly accurate method for branch prediction. The key idea is to\nuse one of the simplest possible neural methods, the perceptron, as an alternative to the commonly\nused two-bit counters. The source of our predictor’s accuracy is its ability to use long history lengths,\nbecause the hardware resources for our method scale linearly, rather than exponentially, with the\nhistory length. We describe two versions of perceptron predictors, and we evaluate these predictors\nwith respect to five well-known predictors. We show that fo\nra4KB\nhardware budget, a simple\nversion of our method that uses a global history achieves a misprediction rate of 4.6% on the SPEC\n2000 integer benchmarks, an improvement of 26% over\ngshare\n. We also introduce a global/local\nversion of our predictor that is 14% more accurate than the McFarling-style hybrid predictor of the\nAlpha 21264. We show that for hardware budgets of up to 256 KB, this global/local perceptron pre-\ndictor is more accurate than Evers’ multicomponent predictor, so we conclude that ours is the most\naccurate dynamic predictor currently available. To explore the feasibility of our ideas, we provide\na circuit-level design of the perceptron predictor and describe techniques that allow our complex\npredictor to operate quickly. Finally, we show how the relatively complex perceptron predictor can\nbe used in modern CPUs by having it override a simpler, quicker Smith predictor, providing IPC\nimprovements of 15.8% over\ngshare\nand 5.7% over the McFarling hybrid predictor.", "type": "", "date": "", "publisher": "", "place": "", "DOI": "", "citationKey": "", "url": "http://delivery.acm.org/10.1145/580000/571639/p369-a_jimenez.pdf?ip=201.206.114.1&acc=ACTIVE%20SERVICE&key=C2716FEBFA981EF11F6DB77B659BED58F7160FA7C9FD07CA&CFID=215505304&CFTOKEN=11720391&__acm__=1368039975_85955e598728d72ac127e1c442853a12", "accessDate": "", "archive": "", "archiveLocation": "", "shortTitle": "", "language": "", "libraryCatalog": "", "callNumber": "", "rights": "", "extra": "", "tags": [], "collections": [ "C2662ZQB", "ZCDARHQ9" ], "relations": {}, "dateAdded": "2013-05-14T05:51:11Z", "dateModified": "2013-05-14T05:51:11Z" } }, { "key": "T5WK5GJD", "version": 7, "library": { "type": "group", "id": 167322, "name": "InvestigacionAdvanced Branch Prediction", "links": { "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction", "type": "text/html" } } }, "links": { "self": { "href": "https://api.zotero.org/groups/167322/items/T5WK5GJD", "type": "application/json" }, "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction/items/T5WK5GJD", "type": "text/html" } }, "meta": { "createdByUser": { "id": 1378368, "username": "sergiogq", "name": "Sergio", "links": { "alternate": { "href": "https://www.zotero.org/sergiogq", "type": "text/html" } } }, "creatorSummary": "Tee-Yu and Yale N.", "numChildren": 0 }, "data": { "key": "T5WK5GJD", "version": 7, "itemType": "document", "title": "Alternative Implementation of Two-Level Adaptive Branch Prediction", "creators": [ { "creatorType": "author", "firstName": "Yeh", "lastName": "Tee-Yu" }, { "creatorType": "author", "firstName": "Patt", "lastName": "Yale N." } ], "abstractNote": "", "type": "", "date": "", "publisher": "", "place": "", "DOI": "", "citationKey": "", "url": "http://www.eecg.utoronto.ca/~moshovos/ACA06/readings/two-level-bpred.pdf", "accessDate": "", "archive": "", "archiveLocation": "", "shortTitle": "", "language": "", "libraryCatalog": "", "callNumber": "", "rights": "", "extra": "", "tags": [], "collections": [ "C2662ZQB", "ZCDARHQ9" ], "relations": {}, "dateAdded": "2013-05-14T05:51:11Z", "dateModified": "2013-05-14T05:51:11Z" } }, { "key": "BECTRBNW", "version": 6, "library": { "type": "group", "id": 167322, "name": "InvestigacionAdvanced Branch Prediction", "links": { "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction", "type": "text/html" } } }, "links": { "self": { "href": "https://api.zotero.org/groups/167322/items/BECTRBNW", "type": "application/json" }, "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction/items/BECTRBNW", "type": "text/html" } }, "meta": { "createdByUser": { "id": 1378368, "username": "sergiogq", "name": "Sergio", "links": { "alternate": { "href": "https://www.zotero.org/sergiogq", "type": "text/html" } } }, "creatorSummary": "Shien–Tai et al.", "numChildren": 0 }, "data": { "key": "BECTRBNW", "version": 6, "itemType": "document", "title": "Improving the Accuracy of Dynamic Branch Prediction sing Branch Correlation", "creators": [ { "creatorType": "author", "firstName": "Pan", "lastName": "Shien–Tai" }, { "creatorType": "author", "firstName": "So", "lastName": "Kimrning" }, { "creatorType": "author", "firstName": "Rahmeh", "lastName": "Joseph T." } ], "abstractNote": "Improving the Accuracy of Dynamic Branch\nPrediction Using Branch Correlation Shien–Tai Pan Advanced Workstation Division IBM Corporation, 9440 11400 Burnet Road Austin, Texas 78758-3493\nKimrning So Joseph T. Rahmeh Advanced\nWorkstation Division Electrical and\nComputer IBM Corporation,\n2808 Engineering Department 11400 Burnet Road University of Texas\nat Austin Austin,\nTexas\n78758-3493\nAustin,\nTexas\n78712\nsok@\nwatson.ibm.com\nrahmeh@cerc.austin.edu\nAbstract\nLong\nbranch\ndelay\nis\na\nwell–known\nproblem\nin\ntoday’s\nhigh\nper-\nformance\nsuperscalar\nand\nsupetpipeline\nprocessor\ndesigns.\nA\ncom-\nmon\ntechnique\nused\nto\nalleviate\nthis\nproblem\nis\nto\npredict\nthe\ndirec-\ntion\nof\nbranches\nduring\nthe\ninstruction\nfetch.\nCounter-based\nbranch\nprediction,\nin\nparticular,\nhas\nbeen\nreported\nas\nan\neffective\nscheme\nfor\npredicting\nthe\ndirection\nof\nbranches.\nHowever,\nits\naccuracy\nis\ngener-\nally\nlimited\nby\nbranches\nwhose\nfuture\nbehavior\nis\nalso\ndependent\nupon\nthe\nhistory\nof\nother\nbranches.\nTo\nenhance\nbranch\nprediction\nac-\ncuracy\nwith\na\nminimum\nincrease\nin\nhardware\nCOSL\nwe\npropose\na\ncor-\nrelation-based\nscheme\nand\nshow\nhow\nthe\nprediction\naccuracy can be improved\nby incorporating information, not only from the\nhistory of a specific brsncb\nbut also from the history of other branches. Specif- ically, we use the\ninformation provided by a proper subhistory\nof a branch to predict the outcome of that branch.\nThe proper subhistory\nis selected based on the outcomes of the most recently executedM branches.\nThe new scheme is evaluated\nusing traces collected from running the\nSPEC benchmark suite on an IBM RISC\nSystem/6000 workstation.\nThe results show that,\nas compared with the 2-bit counter-based prediction scheme, the correlation-based\nbranch prediction achieves up to 11~0 additional accuracy at the extra hardware cost of one shift register.\nThe results also show that the accuracy of the new scheme surpasses that of the counter–based branch predction at saturation.", "type": "", "date": "", "publisher": "", "place": "", "DOI": "", "citationKey": "", "url": "http://delivery.acm.org/10.1145/150000/143490/p76-pan.pdf?ip=201.206.114.1&acc=ACTIVE%20SERVICE&key=C2716FEBFA981EF11F6DB77B659BED58F7160FA7C9FD07CA&CFID=327511259&CFTOKEN=66466897&__acm__=1367946000_96670048534592da598b05bf4a87238f", "accessDate": "", "archive": "", "archiveLocation": "", "shortTitle": "", "language": "", "libraryCatalog": "", "callNumber": "", "rights": "", "extra": "", "tags": [], "collections": [ "ZCDARHQ9" ], "relations": {}, "dateAdded": "2013-05-14T05:51:11Z", "dateModified": "2013-05-14T05:51:11Z" } }, { "key": "XI49ZRA5", "version": 6, "library": { "type": "group", "id": 167322, "name": "InvestigacionAdvanced Branch Prediction", "links": { "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction", "type": "text/html" } } }, "links": { "self": { "href": "https://api.zotero.org/groups/167322/items/XI49ZRA5", "type": "application/json" }, "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction/items/XI49ZRA5", "type": "text/html" } }, "meta": { "createdByUser": { "id": 1378368, "username": "sergiogq", "name": "Sergio", "links": { "alternate": { "href": "https://www.zotero.org/sergiogq", "type": "text/html" } } }, "creatorSummary": "Ciprian et al.", "numChildren": 0 }, "data": { "key": "XI49ZRA5", "version": 6, "itemType": "presentation", "title": "Advanced Branch Prediction Simulator (ABPS)", "creators": [ { "creatorType": "presenter", "firstName": "V. RADU", "lastName": "Ciprian" }, { "creatorType": "presenter", "firstName": "CALBOREAN", "lastName": "Horia" }, { "creatorType": "presenter", "firstName": "CRAPCIU", "lastName": "Adrian" }, { "creatorType": "presenter", "firstName": "VINŢAN", "lastName": "Lucian" }, { "creatorType": "presenter", "firstName": "FLOREA", "lastName": "Adrian" } ], "abstractNote": "", "presentationType": "", "date": "", "meetingName": "", "place": "", "series": "", "sessionTitle": "", "DOI": "", "citationKey": "", "url": "http://www.slideshare.net/adrianflorea/abps-advanced-branch-prediction-simulator", "accessDate": "", "shortTitle": "", "language": "", "rights": "", "extra": "", "tags": [], "collections": [ "ZCDARHQ9" ], "relations": {}, "dateAdded": "2013-05-14T05:51:11Z", "dateModified": "2013-05-14T05:51:11Z" } }, { "key": "9M78URTT", "version": 6, "library": { "type": "group", "id": 167322, "name": "InvestigacionAdvanced Branch Prediction", "links": { "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction", "type": "text/html" } } }, "links": { "self": { "href": "https://api.zotero.org/groups/167322/items/9M78URTT", "type": "application/json" }, "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction/items/9M78URTT", "type": "text/html" } }, "meta": { "createdByUser": { "id": 1378368, "username": "sergiogq", "name": "Sergio", "links": { "alternate": { "href": "https://www.zotero.org/sergiogq", "type": "text/html" } } }, "creatorSummary": "Rajib", "numChildren": 0 }, "data": { "key": "9M78URTT", "version": 6, "itemType": "videoRecording", "title": "Advanced Branch Prediction Techniques YouTube", "creators": [ { "creatorType": "director", "firstName": "Mall", "lastName": "Rajib" } ], "abstractNote": "", "videoRecordingFormat": "", "seriesTitle": "", "volume": "", "numberOfVolumes": "", "studio": "", "place": "", "date": "", "runningTime": "", "ISBN": "", "DOI": "", "citationKey": "", "url": "http://www.youtube.com/watch?v=gwNeh4Ej6GU", "accessDate": "", "archive": "", "archiveLocation": "", "shortTitle": "", "language": "", "libraryCatalog": "", "callNumber": "", "rights": "", "extra": "", "tags": [], "collections": [ "ZCDARHQ9" ], "relations": {}, "dateAdded": "2013-05-14T05:51:11Z", "dateModified": "2013-05-14T05:51:11Z" } }, { "key": "XAS3K5NW", "version": 6, "library": { "type": "group", "id": 167322, "name": "InvestigacionAdvanced Branch Prediction", "links": { "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction", "type": "text/html" } } }, "links": { "self": { "href": "https://api.zotero.org/groups/167322/items/XAS3K5NW", "type": "application/json" }, "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction/items/XAS3K5NW", "type": "text/html" } }, "meta": { "createdByUser": { "id": 1378368, "username": "sergiogq", "name": "Sergio", "links": { "alternate": { "href": "https://www.zotero.org/sergiogq", "type": "text/html" } } }, "creatorSummary": "Erik", "numChildren": 0 }, "data": { "key": "XAS3K5NW", "version": 6, "itemType": "document", "title": "A Mechanism for Reducing Negative Branch History Interference", "creators": [ { "creatorType": "author", "firstName": "Sprangle", "lastName": "Erik" } ], "abstractNote": "", "type": "", "date": "", "publisher": "", "place": "", "DOI": "", "citationKey": "", "url": "http://courses.engr.illinois.edu/ece512/Papers/Sprangle.1997.ISCA.pdf", "accessDate": "", "archive": "", "archiveLocation": "", "shortTitle": "", "language": "", "libraryCatalog": "", "callNumber": "", "rights": "", "extra": "", "tags": [], "collections": [ "ZCDARHQ9" ], "relations": {}, "dateAdded": "2013-05-14T05:51:11Z", "dateModified": "2013-05-14T05:51:11Z" } }, { "key": "UFAXTXVI", "version": 6, "library": { "type": "group", "id": 167322, "name": "InvestigacionAdvanced Branch Prediction", "links": { "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction", "type": "text/html" } } }, "links": { "self": { "href": "https://api.zotero.org/groups/167322/items/UFAXTXVI", "type": "application/json" }, "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction/items/UFAXTXVI", "type": "text/html" } }, "meta": { "createdByUser": { "id": 1378368, "username": "sergiogq", "name": "Sergio", "links": { "alternate": { "href": "https://www.zotero.org/sergiogq", "type": "text/html" } } }, "numChildren": 0 }, "data": { "key": "UFAXTXVI", "version": 6, "itemType": "presentation", "title": "Advanced Dynamic Branch Prediction", "creators": [], "abstractNote": "", "presentationType": "", "date": "", "meetingName": "", "place": "", "series": "", "sessionTitle": "", "DOI": "", "citationKey": "", "url": "http://homepages.stca.herts.ac.uk/~7ele0042/lectures/AdvancedDynamicBranchPrediction.pdf", "accessDate": "", "shortTitle": "", "language": "", "rights": "", "extra": "", "tags": [], "collections": [ "ZCDARHQ9" ], "relations": {}, "dateAdded": "2013-05-14T05:51:11Z", "dateModified": "2013-05-14T05:51:11Z" } }, { "key": "662BK2FR", "version": 6, "library": { "type": "group", "id": 167322, "name": "InvestigacionAdvanced Branch Prediction", "links": { "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction", "type": "text/html" } } }, "links": { "self": { "href": "https://api.zotero.org/groups/167322/items/662BK2FR", "type": "application/json" }, "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction/items/662BK2FR", "type": "text/html" } }, "meta": { "createdByUser": { "id": 1378368, "username": "sergiogq", "name": "Sergio", "links": { "alternate": { "href": "https://www.zotero.org/sergiogq", "type": "text/html" } } }, "creatorSummary": "Linley", "numChildren": 0 }, "data": { "key": "662BK2FR", "version": 6, "itemType": "document", "title": "Digital 21264 Sets New Standard", "creators": [ { "creatorType": "author", "firstName": "Gwennap", "lastName": "Linley" } ], "abstractNote": "", "type": "", "date": "", "publisher": "", "place": "", "DOI": "", "citationKey": "", "url": "http://atc2.aut.uah.es/~nacho/Micros/Caracteristicas/21264_articulo.pdf", "accessDate": "", "archive": "", "archiveLocation": "", "shortTitle": "", "language": "", "libraryCatalog": "", "callNumber": "", "rights": "", "extra": "", "tags": [], "collections": [ "ZCDARHQ9" ], "relations": {}, "dateAdded": "2013-05-14T05:51:11Z", "dateModified": "2013-05-14T05:51:11Z" } }, { "key": "DIPHEU5V", "version": 6, "library": { "type": "group", "id": 167322, "name": "InvestigacionAdvanced Branch Prediction", "links": { "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction", "type": "text/html" } } }, "links": { "self": { "href": "https://api.zotero.org/groups/167322/items/DIPHEU5V", "type": "application/json" }, "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction/items/DIPHEU5V", "type": "text/html" } }, "meta": { "createdByUser": { "id": 1378368, "username": "sergiogq", "name": "Sergio", "links": { "alternate": { "href": "https://www.zotero.org/sergiogq", "type": "text/html" } } }, "creatorSummary": "Paj", "numChildren": 0 }, "data": { "key": "DIPHEU5V", "version": 6, "itemType": "document", "title": "Branch Prediction Schemes", "creators": [ { "creatorType": "author", "firstName": "Parihar", "lastName": "Paj" } ], "abstractNote": "", "type": "", "date": "", "publisher": "", "place": "", "DOI": "", "citationKey": "", "url": "http://www.ece.rochester.edu/~parihar/pres/Pres_BranchPrediction-Survey.pdf", "accessDate": "", "archive": "", "archiveLocation": "", "shortTitle": "", "language": "", "libraryCatalog": "", "callNumber": "", "rights": "", "extra": "", "tags": [], "collections": [ "ZCDARHQ9" ], "relations": {}, "dateAdded": "2013-05-14T05:51:11Z", "dateModified": "2013-05-14T05:51:11Z" } }, { "key": "Q7TVSR5A", "version": 6, "library": { "type": "group", "id": 167322, "name": "InvestigacionAdvanced Branch Prediction", "links": { "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction", "type": "text/html" } } }, "links": { "self": { "href": "https://api.zotero.org/groups/167322/items/Q7TVSR5A", "type": "application/json" }, "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction/items/Q7TVSR5A", "type": "text/html" } }, "meta": { "createdByUser": { "id": 1378368, "username": "sergiogq", "name": "Sergio", "links": { "alternate": { "href": "https://www.zotero.org/sergiogq", "type": "text/html" } } }, "creatorSummary": "Mikko H.", "numChildren": 0 }, "data": { "key": "Q7TVSR5A", "version": 6, "itemType": "presentation", "title": "Advanced Branch Prediction", "creators": [ { "creatorType": "presenter", "firstName": "Lipasti", "lastName": "Mikko H." } ], "abstractNote": "", "presentationType": "", "date": "", "meetingName": "", "place": "", "series": "", "sessionTitle": "", "DOI": "", "citationKey": "", "url": "http://ece752.ece.wisc.edu/lect09-adv-branch-prediction.pdf", "accessDate": "", "shortTitle": "", "language": "", "rights": "", "extra": "", "tags": [], "collections": [ "ZCDARHQ9" ], "relations": {}, "dateAdded": "2013-05-14T05:51:11Z", "dateModified": "2013-05-14T05:51:11Z" } }, { "key": "M5D6BW7T", "version": 6, "library": { "type": "group", "id": 167322, "name": "InvestigacionAdvanced Branch Prediction", "links": { "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction", "type": "text/html" } } }, "links": { "self": { "href": "https://api.zotero.org/groups/167322/items/M5D6BW7T", "type": "application/json" }, "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction/items/M5D6BW7T", "type": "text/html" } }, "meta": { "createdByUser": { "id": 1378368, "username": "sergiogq", "name": "Sergio", "links": { "alternate": { "href": "https://www.zotero.org/sergiogq", "type": "text/html" } } }, "creatorSummary": "Sunghoon", "numChildren": 0 }, "data": { "key": "M5D6BW7T", "version": 6, "itemType": "document", "title": "Branch Prediction using Advanced Neural Methods", "creators": [ { "creatorType": "author", "firstName": "Kim", "lastName": "Sunghoon" } ], "abstractNote": "Among the hardware techniques,\ntwo-level adaptive branch predictors with two-bit saturating counters are acknowledged as best branch predict\nors. They accomplish very competitive performance at low hardware cost. However, with the rapid of evolution of superscalar processors, the more\naccurate predictors are desired for more\ncorrect branch prediction as one of\nspeculation method. They will lead to higher performance of processors with no doubt.\nThis article presents alternative new and potential neural net methods for branch prediction. The advanced applications of the neural networks more than perceptron or backpropagation are examined as alternative methods. They are radial basis networks,\nElman networks and Learning vector\nquantization (LVQ) networks. I demonstrate that these neural methods can achieve misprediction rate comparable to the conventional two-level adaptive predictors,\nrepresentatively a Gshare method, without consideration of hardware and prediction latency. I also present the effects of the history length of the global history shift register (HR) and the size of the pattern history table (PHT) on the misprediction rate for each neural method.", "type": "", "date": "", "publisher": "", "place": "", "DOI": "", "citationKey": "", "url": "http://www.cs.berkeley.edu/~kubitron/courses/cs252-F03/projects/reports/project10_report_ver3.pdf", "accessDate": "", "archive": "", "archiveLocation": "", "shortTitle": "", "language": "", "libraryCatalog": "", "callNumber": "", "rights": "", "extra": "", "tags": [], "collections": [ "ZCDARHQ9" ], "relations": {}, "dateAdded": "2013-05-14T05:51:11Z", "dateModified": "2013-05-14T05:51:11Z" } }, { "key": "9CDWFMET", "version": 6, "library": { "type": "group", "id": 167322, "name": "InvestigacionAdvanced Branch Prediction", "links": { "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction", "type": "text/html" } } }, "links": { "self": { "href": "https://api.zotero.org/groups/167322/items/9CDWFMET", "type": "application/json" }, "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction/items/9CDWFMET", "type": "text/html" } }, "meta": { "createdByUser": { "id": 1378368, "username": "sergiogq", "name": "Sergio", "links": { "alternate": { "href": "https://www.zotero.org/sergiogq", "type": "text/html" } } }, "creatorSummary": "Kunle", "numChildren": 0 }, "data": { "key": "9CDWFMET", "version": 6, "itemType": "presentation", "title": "Lecture 8: Compiling for ILP and Branch Prediction", "creators": [ { "creatorType": "presenter", "firstName": "Olukotun", "lastName": "Kunle" } ], "abstractNote": "", "presentationType": "", "date": "", "meetingName": "", "place": "", "series": "", "sessionTitle": "", "DOI": "", "citationKey": "", "url": "http://www.stanford.edu/class/ee282h/handouts/Handout21.pdf", "accessDate": "", "shortTitle": "", "language": "", "rights": "", "extra": "", "tags": [], "collections": [ "ZCDARHQ9" ], "relations": {}, "dateAdded": "2013-05-14T05:51:11Z", "dateModified": "2013-05-14T05:51:11Z" } }, { "key": "7IXGC4WQ", "version": 6, "library": { "type": "group", "id": 167322, "name": "InvestigacionAdvanced Branch Prediction", "links": { "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction", "type": "text/html" } } }, "links": { "self": { "href": "https://api.zotero.org/groups/167322/items/7IXGC4WQ", "type": "application/json" }, "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction/items/7IXGC4WQ", "type": "text/html" } }, "meta": { "createdByUser": { "id": 1378368, "username": "sergiogq", "name": "Sergio", "links": { "alternate": { "href": "https://www.zotero.org/sergiogq", "type": "text/html" } } }, "creatorSummary": "Toni", "numChildren": 0 }, "data": { "key": "7IXGC4WQ", "version": 6, "itemType": "document", "title": "A Third Level of Adaptivity for Branch Prediction", "creators": [ { "creatorType": "author", "firstName": "Juan", "lastName": "Toni" } ], "abstractNote": "Accurate branch prediction is essential for obtaining\nhigh performance in pipelined superscalar processors that\nexecute instructions speculatively. Some of the best current\npredictors combine a part of the branch address with a fixed\namount of global history of branch outcomes in order to\nmake a prediction. These predictors cannot perform uni-\nformly well across all workloads because the best amount\nof history to be used depends on the code, the input data and\nthe frequency of context switches. Consequently, all predic-\ntors that use a fixed history length are therefore unable to\nperform up to their maximum potential.\nWe introduce a method —called DHLF— that dynami-\ncally determines the optimum history length during execu-\ntion, adapting to the specific requirements of any code, in-\nput data and system workload. Our proposal adds an extra\nlevel of adaptivity to two-level adaptive branch predictors.\nThe DHLF method can be applied to any one of the predic-\ntors that combine global branch history with the branch ad-\ndress. We apply the DHLF method to gshare (\ndhlf-gshare\n)\nand obtain near-optimal results for all\nSPECint95\nbench-\nmarks, with and without context switches. Some results are\nalso presented for gskewed (\ndhlf-gskewed\n), confirming that\nother predictors can benefit from our proposal", "type": "", "date": "", "publisher": "", "place": "", "DOI": "", "citationKey": "", "url": "http://www.eecg.toronto.edu/~moshovos/ACA06/readings/History-Length-Fitting.pdf", "accessDate": "", "archive": "", "archiveLocation": "", "shortTitle": "", "language": "", "libraryCatalog": "", "callNumber": "", "rights": "", "extra": "", "tags": [], "collections": [ "ZCDARHQ9" ], "relations": {}, "dateAdded": "2013-05-14T05:51:11Z", "dateModified": "2013-05-14T05:51:11Z" } }, { "key": "34DFU3X3", "version": 6, "library": { "type": "group", "id": 167322, "name": "InvestigacionAdvanced Branch Prediction", "links": { "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction", "type": "text/html" } } }, "links": { "self": { "href": "https://api.zotero.org/groups/167322/items/34DFU3X3", "type": "application/json" }, "alternate": { "href": "https://www.zotero.org/groups/investigacionadvanced_branch_prediction/items/34DFU3X3", "type": "text/html" } }, "meta": { "createdByUser": { "id": 1378368, "username": "sergiogq", "name": "Sergio", "links": { "alternate": { "href": "https://www.zotero.org/sergiogq", "type": "text/html" } } }, "creatorSummary": "Yoav Etsion", "numChildren": 0 }, "data": { "key": "34DFU3X3", "version": 6, "itemType": "document", "title": "Advanced Branch Prediction", "creators": [ { "creatorType": "author", "firstName": "Dan Tsafrir", "lastName": "Yoav Etsion" } ], "abstractNote": "", "type": "", "date": "", "publisher": "", "place": "", "DOI": "", "citationKey": "", "url": "http://webcourse.cs.technion.ac.il/234267/Spring2013/ho/WCFiles/3-branch.pdf", "accessDate": "", "archive": "", "archiveLocation": "", "shortTitle": "", "language": "", "libraryCatalog": "", "callNumber": "", "rights": "", "extra": "", "tags": [], "collections": [ "ZCDARHQ9" ], "relations": {}, "dateAdded": "2013-05-14T05:51:11Z", "dateModified": "2013-05-14T05:51:11Z" } } ]