• 検索結果がありません。

ダイナミック・ロジックへのタイミング・フォールト検出手法の適用

N/A
N/A
Protected

Academic year: 2021

シェア "ダイナミック・ロジックへのタイミング・フォールト検出手法の適用"

Copied!
8
0
0

読み込み中.... (全文を見る)

全文

(1)Vol.2014-ARC-210 No.18 Vol.2014-OS-129 No.18 2014/5/15. ৘ใॲཧֶձ‫ڀݚ‬ใࠂ IPSJ SIG Technical Report. μΠφϛοΫɾϩδοΫ΁ͷ λΠϛϯάɾϑΥʔϧτ‫ݕ‬ग़ख๏ͷద༻ ਆอ ை1. ࢁా ३ೋ1. ‫ޒ‬ౡ ਖ਼༟2. ࡔҪ मҰ1. ֓ཁɿۙ೥ɼ൒ಋମϓϩηεͷඍࡉԽʹ൐ͬͯνοϓ্ͷૉࢠ஗Ԇͷ͹Β͖͕ͭ૿Ճ͓ͯ͠ΓɼาཹΓΛ Ұఆʹͨ͠৔߹ɼϫʔετɾέʔεઃ‫Ͱܭ‬͸ੑೳ͕޲্͠ͳ͘ͳΔ‫ڪ‬Ε͕͋Δɽճ࿏ͷϫʔετͳ஗ԆͰ ͳ࣮͘ࡍͷ஗Ԇʹ‫ͮ͘ج‬ಈ࡞ΛՄೳʹ͢ΔͨΊʹɼಈతͳλΠϛϯάɾϑΥʔϧτ‫ݕ‬ग़ख๏Ͱ͋Δ Razor ͕ఏҊ͞Ε͍ͯΔɽ͔͠͠ɼRazor ΛϨδελɾϑΝΠϧ΍ΩϟογϡΛߏ੒͢ΔμΠφϛοΫɾϩδοΫ ΁ͷద༻͢Δख๏͸ߟྀ͞Ε͍ͯͳ͔ͬͨɽ͜Ε͸μΠφϛοΫɾϩδοΫʹ͓͚ΔϓϦνϟʔδಈ࡞͕ λΠϛϯάɾϑΥʔϧτΛϚεΫͯ͠͠·͏ͨΊͰ͋ΔɽຊߘͷఏҊख๏͸ɼμΠφϛοΫɾϩδοΫͷ ධՁ݁ՌʹԠͯ͡ɼϓϦνϟʔδͷ༗ແΛ੍‫͢ޚ‬Δ͜ͱʹΑͬͯద༻ΛՄೳʹ͢ΔɽఏҊख๏Λద༻ͨ͠ ϨδελɾϑΝΠϧΛτϥϯδελɾϨϕϧͰઃ‫͠ܭ‬ɼSPICE γϛϡϨʔγϣϯ্ͰλΠϛϯάɾϑΥʔ ϧτΛ‫ݕ‬ग़Ͱ͖Δ͜ͱΛ֬ೝͨ͠ɽ. 1. ͸͡Ίʹ ൒ಋମϓϩηεͷඍࡉԽʹ൐ͬͯɼૉࢠ஗Ԇͷ͹Βͭ ͖͕େ͖ͳ໰୊ͱͳΓͭͭ͋Δɽૉࢠ‫ܗ‬ঢ়ͳͲʹґଘ͢Δ ‫ܥ‬౷తͳ͹Β͖ͭ͸ઃ‫Ͱܭ‬ճආ͢Δ͜ͱ͕ՄೳͰ͋Δ͕ɼ. 䠍ୡ௦๓䛾䝥䝻䝉䝇 䠄䜀䜙䛴䛝ᑠ䠅. 㻌. ෆ७෺ΏΒ͗΍ LER(Line Edge RoughnessɺϥΠϯΤοδ. ᚤ⣽໬䛧䛯䝥䝻䝉䝇 (䜀䜙䛴䛝኱). 吟 呀 呉 吏 吐 吖 ᩘ. 㐜ᘏ. 㻜 㻜. ͷΏΒ͗) ͳͲʹ‫ى‬Ҽ͢ΔϥϯμϜͳ͹Β͖ͭ͸ճආͰ͖. ඾ᆺ್䛾๐ῶ. ͳ͍ɽ·ͨɼ͜ΕΒͷ͹Β͖ͭ͸τϥϯδελ΍഑ઢͷେ. ᭱ᝏ್䛾๐ῶ. ͖͕͞‫ࢠݪ‬ͷେ͖͞ʹۙͮ͘ʹैͬͯ૿Ճ͢Δ͜ͱ͕༧ଌ ඾ᆺ್. ͞Ε͍ͯΔ [1]ɽ. ᭱ᝏ್. ͹Β͖͕ͭ૿େ͍ͯ͘͠ͱɼैདྷͷ࠷ѱ஋ʹ‫͍ͨͮج‬ઃ ‫ܭ‬ख๏͸൵‫؍‬తʹͳΓ͗͢Δɽ͜ͷ༷ࢠΛਤ 1 ʹࣔ͢ɽඍ. ㏿ ᗘ. ࡉԽ͕ਐΉʹͭΕͯ஗Ԇͷయ‫ܕ‬஋͕޲্͢ΔҰํɼ͹Βͭ ͖ͷ૿େʹΑΓ࠷ѱ஋͸య‫ܕ‬஋΄Ͳ޲্͠ͳ͍ɽ͕ͨͬ͠ ͯɺ࠷ѱ஋ʹ‫͍ͨͮج‬ઃ‫Ͱܭ‬͸ LSI ͷಈ࡞଎౓͕޲্͠ͳ. 䝥䝻䝉䝇ᢏ⾡䛾ୡ௦. ͘ͳΔ‫ڪ‬Ε͕͋Δɽ ͜ͷ໰୊ʹରॲ͢ΔͨΊʹɼϫʔετɾέʔεͷ஗ԆͰ ͸ͳ࣮͘ࡍͷ஗Ԇʹ‫͍ͨͮج‬ಈ࡞ͷ࣮‫ݱ‬Λ໨తͱ͢Δख๏ ͕਺ଟ͘ఏҊ͞Ε͍ͯΔɽͦͷΑ͏ͳख๏ͷҰछͱͯ͠ɼ ಈ࡞࣌ʹλΠϛϯάɾϑΥʔϧτΛ‫ݕ‬ग़͠ճ෮͢Δख๏͕ ͋Δɽ λΠϛϯάɾϑΥʔϧτʢTiming Fault : TFʣͱ͸ɼ஗Ԇ 1. 2. ౦‫ژ‬େֶେֶӃ৘ใཧ޻ֶ‫ڀݚܥ‬Պ Graduate School of Information Science and Technology, The University of Tokyo ࠃཱ৘ใֶ‫ॴڀݚ‬ National Institute of Informatics. ⓒ 2014 Information Processing Society of Japan. ਤ1. ϓϩηοαͷయ‫ܕ‬஋ͱ࠷ѱ஋ͷ޲্෯ͷဃ཭. ͷಈతͳมԽʹΑΓઃ‫ऀܭ‬ͷҙਤͱ͸ҟͳΔಈ࡞͕Ҿ͖‫ى‬ ͜͞ΕΔա౉‫ނ‬োͰ͋Δɽϫʔετɾέʔεઃ‫Ͱܭ‬͸ɼ૝ ఆͨ͠ಈ࡞৚݅಺ͷϫʔετɾέʔεʹ͓͚Δ஗ԆΛ‫ੵݟ‬ ΋Γɼͦͷ৚݅಺Ͱ TF ͕ൃੜ͠ͳ͍Α͏ʹઃ‫͢ܭ‬Δɽ ҰํͰɼTF ͷൃੜࣗମ͸‫ڐ‬༰͠ɼTF ͷ‫ݕ‬ग़ɾճ෮Λ ߦ͏ख๏͕ߟ͑ΒΕΔɽ2.2 અͰड़΂Δ Razor [2], [3], [4] ͸ɼͦͷ୅දྫͰ͋Δɽ͜ͷΑ͏ͳख๏ͱ DVFS (Dynamic. Voltage and Frequency Scaling) [5] Λ૊Έ߹ΘͤΔͱɼҎԼ ͷΑ͏ʹɼ‫ੵݟ‬΋ΓͰ͸ͳ͍ɼ࣮ࡍͷ஗ԆʹԠͨ͡ಈ࡞Λ. 1.

(2) Vol.2014-ARC-210 No.18 Vol.2014-OS-129 No.18 2014/5/15. ৘ใॲཧֶձ‫ڀݚ‬ใࠂ IPSJ SIG Technical Report. Πϛϯά੍໿ͷ؇࿨ʹ͍ͭͯड़΂Δɽͳ͓ຊߘͰ͸ɼλΠ. 嵈崢. ϛϯά੍໿Λٞ࿦͢Δ্ͰαΠΫϧɾλΠϜͷΈʹண໨. 嵖嵈崢৅ে崢. ͠ɼFF/ϥονͷηοτΞοϓ/ϗʔϧυɾλΠϜɼΫϩο ৢଞभ嵆嵘嵈嵕崢भ崢 ৿੿ਡ崢. Ϋ–σʔλ஗Ԇɼ͓ΑͼɼΫϩοΫɾεΩϡʔͳͲͷߟྀʹ ͍ͭͯ͸ɼઆ໌͕൥ࡶʹͳΔͨΊলུ͍ͯ͠ΔɽඞཁͰ͋ Ε͹ɼ͜ΕΒΛٞ࿦ʹ௥Ճ͢Δ͜ͱ͸༰қͰ͋Δɽ. 嵘崢. ਤ2. λΠϛϯάɾϑΥʔϧτ‫ݕ‬ग़ɾճ෮ͱ DVFS ͷ૊Έ߹Θͤ. 2.1 λΠϛϯάɾμΠΞάϥϜ ਤ 3ʢ্ʣͷճ࿏ʹ͓͍ͯɼ৴߸͕఻ΘΔ༷ࢠΛಉਤʢԼʣ. ࣮‫͢ݱ‬Δ͜ͱ͕Ͱ͖Δɽ ਤ 2 ʹͦͷ༷ࢠΛࣔ͢ɽਤ 2 தʷҹ͸ϫʔετɾέʔε. ʹࣔ͢ɽຊߘͰ͸ɼਤ 3 ԼਤΛλΠϛϯάɾμΠΞάϥϜ ͱ‫ͼݺ‬ɼҎԼͰ͸ t-diagram ͱ‫͢ه‬ɽ௨ৗͷλΠϛϯάɾ. ͰఆΊΒΕͨ DVFS ͷ VʢVoltageɿి‫ిݯ‬ѹʣͱ FʢFre-. νϟʔτ͕࿦ཧ஋-࣌ؒͷ 2 ࣍‫ݩ‬Λ࣋ͭʹରͯ͠ɼt-diagram. quencyɿಈ࡞प೾਺ʣͷ૊Λද͍ͯ͠Δɽϫʔετɾέʔ. ͸࣌ؒ-ۭؒͷ 2 ࣍‫ݩ‬Λ࣋ͭɽ. εઃ‫Ͱܭ‬͸ɼ͜ͷΑ͏ʹ TF ͕ൃੜ͠ͳ͍Α͏े෼ͳϚʔ. ௨ৗͷλΠϜɾνϟʔτͰ͸ɼӈํ޲͕࣌ؒΛɼ্Լํ. δϯΛऔͬͯ V–F ͕ઃఆ͞ΕΔɽҰํ Razor ͷΑ͏ʹ TF. ޲͕࿦ཧ஋Λද͢ɽλΠϜɾνϟʔτ͸ɼ࿦ཧ஋ͷ࣌ؒత. ‫ݕ‬ग़ɾճ෮Λߦ͏ख๏Ͱ͸ɼϚʔδϯΛͱΔඞཁ͕ͳ͍ɽ. มԽΛද‫͢ݱ‬Δ͕ɼ1 ຊͷ೾‫Ͱܗ‬ද͢͜ͱ͕Ͱ͖Δͷ͸ճ. ਤ 2 த˓ҹͰද͞ΕΔ‫ݕ‬ग़௚લͷ V–F ͕ɼ‫ੵݟ‬΋ΓͰ͸ͳ. ࿏ͷಛఆͷ 1 ఺ͷৼΔ෣͍ʹ‫ݶ‬ΒΕΔɽෳ਺ͷ఺ʹ·͕ͨ. ͍ɼͦͷνοϓͷͦͷ࣌ͷಈ࡞‫͚͓ʹڥ؀‬Δ࣮ࡍͷ஗Ԇʹ. Δಈ͖Λ೺Ѳ͢ΔͨΊʹ͸ɼෳ਺ͷ೾‫ܗ‬Λฒ΂ͳ͚Ε͹ͳ. Ԡͨ͡ V–F Ͱ͋ΔɽTF ‫ݕ‬ग़ʹΑͬͯɼ͜ͷ V–F ෇ۙʹ͓. Βͳ͍ɽ. ͚Δಈ࡞͕ՄೳʹͳΔɽ͕ͨͬͯ͠ɼ࣮ࡍͷ஗ԆʹԠͨ͡. ͦΕʹରͯ͠ t-diagram ͸ɼԼํ޲͕࣌ؒΛɼӈํ޲͕ճ. ΑΓ௿ిѹɼߴप೾ଆͷ‫͍͠ݫ‬৚݅Ͱͷಈ࡞ʹ౿ΈࠐΉ͜. ࿏தΛ৴߸͕఻Θͬͯߦ͘ํ޲Λද͠ɼ࣌ؒͷ‫ܦ‬աʹͭΕ. ͱ͕Ͱ͖ɼੑೳ޲্ɼ͋Δ͍͸ফඅిྗ࡟‫͕ݮ‬ՄೳʹͳΔɽ. ͯ৴߸͕఻Θ͍༷ͬͯ͘ࢠΛ၆ᛌ͢Δ͜ͱ͕Ͱ͖Δɽਤ 3. ҰํɼRazor ʹΑΔ TF ‫ݕ‬ग़ΛϨδελɾϑΝΠϧ΍Ωϟο. ʢ্ʣʹࣔ͢ճ࿏Ͱɼ࣌ࠁ t = 0 ʹ 3 ͭͷ FF ͷग़ྗ (x, y, z). γϡΛߏ੒͢ΔμΠφϛοΫɾϩδοΫ΁ద༻͢Δख๏ʹ. ͕ (1, 1, 0) ͔Β (0, 0, 1) ʹભҠͨ͠ͱ͢Δɽxɼy ɼz ͔Β d. ͍ͭͯ͸ߟྀ͞Ε͍ͯͳ͔ͬͨɽࠓ೔ɼSRAM ͸഑ઢ஗. ʹࢸΔύεͷ஗ԆΛͦΕͧΕ tx ɼty ɼtz ͱ͢Δͱɼϩδο. Ԇ͕ଟ͘Λ઎Ί͓ͯΓɼ͜Ε͸εέʔϦϯάʹΑͬͯ‫ݮ‬Δ ͜ͱ͕ͳ͍ͨΊɼͦͷ૬ରతͳ஗Ԇ͕૿େ͍ͯ͠Δɽͨ͠ ͕ͬͯɼϨδελɾϑΝΠϧ΍ΩϟογϡͷΞΫηε͕ճ ࿏શମʹ͓͍ͯΫϦςΟΧϧͳ஗ԆΛ΋ͭ͜ͱ͸ආ͚ΒΕ ͳ͍ɽμΠφϛοΫɾϓϦνϟʔδɾϩδοΫ΁ Razor Λ ద༻͢Δ͜ͱ͕Ͱ͖Ε͹ɼճ࿏શମͷੑೳ޲্ɼ͋Δ͍͸ ফඅిྗ࡟‫͕ݮ‬ՄೳʹͳΔͱߟ͑Β͑Δɽ ຊߘͰ͸ɼRazor ͷ TF ‫ݕ‬ग़‫ߏػ‬ͷμΠφϛοΫɾϓϦ νϟʔδɾϩδοΫ΁ͷద༻ΛՄೳʹ͢ΔͨΊͷख๏Λఏ Ҋ͢Δɽ·ͨɼఏҊख๏Λద༻ͨ͠ϨδελɾϑΝΠϧΛ τϥϯδελɾϨϕϧͰઃ‫͠ܭ‬ɼSPICE γϛϡϨʔγϣϯ ্ͰλΠϛϯάɾϑΥʔϧτΛ‫ݕ‬ग़Ͱ͖Δ͜ͱΛ֬ೝͨ͠ɽ ҎԼɼୈ 2 અͰ͸ɺλΠϛϯάɾμΠΞάϥϜͱ‫ݺ‬͹Ε ΔਤΛಋೖ͠ɼ࣮ޮ஗Ԇͱ‫֓Ϳݺ‬೦Λઆ໌͢Δɽͦͷ্Ͱɼ. Razor ʹΑΔλΠϛϯά੍໿ͷ؇࿨ʹ͍ͭͯड़΂Δɽଓ͘ ୈ 3 અͰ͸ɺμΠφϛοΫɾϓϦνϟʔδɾϩδοΫͷ୅ දͱͯ͠ SRAM ΛऔΓ্͛ɼͦͷλΠϛϯά੍໿ͷಛ௃ Λड़΂Δɽୈ 4 અͰ͸ఏҊख๏ʹ͍ͭͯৄड़͢Δɽୈ 5 અ Ͱ͸ఏҊख๏ͷಈ࡞֬ೝʹ͍ͭͯड़΂ɼ࠷‫ʹޙ‬ୈ 6 અͰ· ͱΊΔɽ. 2. Razor: λΠϛϯάɾϑΥʔϧτ‫ݕ‬ग़‫ߏػ‬ ຊઅͰ͸ɼRazor ͷճ࿏ߏ੒ͱͦΕʹΑΔ࿦ཧճ࿏ͷλ. ⓒ 2014 Information Processing Society of Japan. ਤ3. λΠϛϯάɾμΠΞάϥϜ (t-diagram) ͱ࣮ޮ஗Ԇ. 2.

(3) Vol.2014-ARC-210 No.18 Vol.2014-OS-129 No.18 2014/5/15. ৘ใॲཧֶձ‫ڀݚ‬ใࠂ IPSJ SIG Technical Report. Ϋͷग़ྗ d ͸ɼ࣌ࠁ tx ɼty ʹ͓͍ͯ 0 → 1 → 0 ͱભҠ͢. main. Δɽz ͔Β d ʹࢸΔύεͷ৴߸͸ɼy ͔Β d ʹࢸΔύεͷ. d. ৴߸ʹΑͬͯมԽ͕ϚεΫ͞ΕΔͨΊɼ࣌ࠁ tz ʹ͸ग़ྗ. ݈ܿ݇ . Main cycle. error. ͸มԽ͠ͳ͍͜ͱʹ஫ҙ͞Ε͍ͨɽಉਤͷӈ୺ʹ͋Δ೾‫ܗ‬ ͕ɼd ʹ͓͚Δ௨ৗͷλΠϜɾνϟʔτʢΛӈʹ 90◦ ճస ͨ͠΋ͷʣͰ͋Δɽ. Detect. ݈ܿ݇  ݈ܿ݇ . ݈݀݁ܽ‫݀݁ݕ‬. shadow. ݈ܿ݇ . d. ݈݀݁ܽ‫݀݁ݕ‬ main:0. ύεͷ‫׆‬ੑԽͱλΠϛϯάɾμΠΞάϥϜ ਤ4. ಉਤͷΑ͏ʹ t-diagram Ͱ͸ɼϩδοΫͷೖྗʹ͓͍ͯ. shadow:1. Razor ͷճ࿏ߏ੒. ࿦ཧ஋͕มԽͨ࣌͠ࠁ͔Βɼग़ྗʹ͓͍ͯ࿦ཧ஋͕มԽ͠ ͳ‫݁݃ܽݐݏ‬. ͳ‫݁݃ܽݐݏ‬. ͨ࣌ࠁ·ͰΛ௚ઢ໼ҹͰ݁Ϳ͜ͱʹΑͬͯɼ৴߸ͷ఻ΘΔ ༷ࢠΛද͢͜ͱ͕Ͱ͖Δɽ. ߙ. ͳܿ‫݈݁ܿݕ‬. ͱʹΑͬͯɼ֤εςʔδͷ஗Ԇ͸ɼt-diagram ্ͷεςʔδ. ͳܿ‫݈݁ܿݕ‬. ͳ͓ t-diagram Ͱ͸ɼ֤εςʔδͷΫϦςΟΧϧɾύε ʹରԠ͢Δ௚ઢ໼ҹͷ֯౓Λ 45◦ ͱ͍ͯ͠Δɽ͜͏͢Δ͜ ͷԣ෯ʹΑͬͯද‫͢ݱ‬Δ͜ͱ͕Ͱ͖Δɽ࣮ࡍͷϩδοΫͰ ͸ɼ͹Β͖ͭͷͨΊɼ஗Ԇ͸࿈ଓతʹมԽ͢ΔɽͦͷͨΊɼ ໼ҹͷଘࡏൣғ͸ɼϩδοΫͷ࠷খ஗Ԇͷ໼ҹͱΫϦςΟ Χϧɾύεͷ஗Ԇͷ੺໼ҹʹ্ԼΛ‫·ڬ‬ΕͨྖҬͱͳΔɽ. t-diagram Ͱ͸ɼ໢ֻ͚Λࢪͯ͜͠ͷྖҬΛࣔ͢ɽ ਤ 3 ʹࣔͨ͠ྫͰ͸ɼલड़ͨ͠Α͏ʹɼz ͔Β d ʹࢸΔ ύεΛ௨Δ৴߸͸్தͰϚεΫ͞ΕΔͨΊɼ࣌ࠁ t = tz ʹ ͓͍ͯ͸ग़ྗ d ͸มԽ͠ͳ͍ɽύεΛ௨ͬͨ৴߸ʹΑͬͯ ࣮ࡍʹϩδοΫͷग़ྗ͕มԽͨ͠ͱ͖ɼͦͷ৴߸ʹΑͬͯ ͦͷύε͕‫׆‬ੑԽͨ͠ͱ‫͏ݴ‬ɽ. t-diagram Ͱ͸ɼύεΛ‫׆‬ੑԽͨ͠৴߸ͷ఻ୡΛ࣮ઢ໼ҹ. time. ਤ5. ୯૬ FFʢࠨʣͱ Razorʢӈʣͷ t-diagram. 2.2 Razor ͷߏ੒ͱλΠϛϯά੍໿ ਤ 4ʢ্ʣʹɼRazor FF ͷճ࿏ߏ੒Λࣔ͢ [2]ɽRazor FF ͸ɼ௨ৗͷ FFʢMain FFʣͱɼShadow Latch ʹΑͬͯߏ੒͞ ΕΔɽShadow Latch ʹ͸ɼMain FF ΁ͷͦΕΑΓҐ૬ͷ஗. Ͱද͢ɽ‫׆‬ੑԽ͠ͳ͔ͬͨ৔߹ʹ͸ɼ్தͰϚεΫ͞Εͨ. ΕͨΫϩοΫ͕‫͞څڙ‬Ε͓ͯΓɼMain FF ͱ Shadow Latch. ஈ֊Ͱ৴߸͸෺ཧతʹ͸ফࣦ͍ͯ͠Δ͕ɼԾ૝తʹ఺ઢ໼. Ͱ 2 ճɼ৴߸ͷαϯϓϦϯάΛߦ͏ɽͦΕΒͷ஋Λൺֱ͠. ҹͰද͢͜ͱʹ͢Δɽ. ͯɼҟͳ͍ͬͯΕ͹ TF ͱͯ͠‫ݕ‬ग़͢Δɽͳ͓ɼTF ‫ݕ‬ग़‫ޙ‬. ࣮ޮ஗Ԇ. ͸ɼύΠϓϥΠϯɾϑϥογϡͳͲɼΞʔΩςΫνϟɾϨ. ͋ΔϩδοΫʹ͓͍ͯ࠷‫׆ʹޙ‬ੑԽ͞Εͨύεͷ஗Ԇ Λɼ͜ͷϩδοΫͷ࣮ޮ஗Ԇͱ‫͢ʹͱ͜Ϳݺ‬Δɽਤ 3 ͷ৔ ߹ɼ࣌ࠁ t = tz ʹ͓͍ͯΫϦςΟΧϧɾύεΛ௨ͬͨ৴߸ ͕౸ண͢Δ͸͕ͣͩɼϚεΫ͞ΕͨͨΊɼϩδοΫͷग़ྗ. d ͸มԽ͠ͳ͍ɽ͜ͷ৔߹ɼ࣮ޮ஗Ԇ͸ ty ͱͳΔɽ. ϕϧͷख๏ʹΑͬͯ TF ͔Βͷճ෮͕ߦΘΕΔ [2], [7]ɽ ਤ 5 ͸ TF ‫ݕ‬ग़‫ߏػ‬Λ༻͍ͳ͍୯૬ FF ํࣜͱ Razor ͷ. t-diagram Λൺֱͨ͠΋ͷͰ͋Δɽ ಉਤ (ࠨ) ʹ͓͍ͯɼFF ͷԼʹ͋Δ࣮ઢ͸ɼϥον͕ด ͍ͯ͡Δঢ়ଶΛද͍ͯ͠Δɽ৴߸ͷઢ͕͜ͷ࣮ઢʹԊͬͯ. ࣌ࠁ t = ty ʹ͓͍ͯग़ྗ d ͕มԽͨ࣌͠ʹ͸࣮ޮ஗Ԇ. ఻͏༷ࢠ͸ɼͦͷؒϥον͕஋Λอ͍࣋ͯ͠Δ͜ͱΛද. ͕ ty Ͱ͋Δ͜ͱ͸෼͔Βͳ͍ɽ࣌ࠁ t = tz ʹ͓͍ͯ d ͕. ͢ɽΤοδɾτϦΨಈ࡞͸ɼϚελ–εϨʔϒͷϥονΛ. มԽ͠ͳ͔ͬͨ͜ͱΛ‫ॳͯݟ‬Ίͯ ty Ͱ͋ͬͨ͜ͱ͕෼͔. ‫͍ޓ‬ҧ͍ʹ‫ه‬ड़͢Δ͜ͱͰੜ͡Δ͔ܺؒΒ৴߸͕఻೻͢Δ. Δɽ͜ͷΑ͏ʹɼ࣮ޮ஗Ԇ͸ࣄ‫ޙ‬తʹ෼͔Δ͜ͱʹ஫ҙ͞. ༷ࢠͰද͢͜ͱ͕Ͱ͖ΔɽΫϩοΫͷ্ཱ͕ͪΓ·Ͱʹ৴. Ε͍ͨɽ. ߸͕ؒʹ߹͍ͬͯΕ͹Α͍ͷͰɼ୯૬ FF ͷ࠷େ஗Ԇ੍໿. t-diagram Ͱ͸࣮ޮ஗ԆʹରԠ͢Δ໼ҹΛଠ࣮ઢͰද͢ɽ. ͸ 1cycle/1stage ͱͳΔɽ. ϩδοΫ΁ͷೖྗͷมԽͷ࢓ํʹΑͬͯग़ྗͷมԽͷ࢓. ಉਤ (ӈ) ʹ͓͚Δ Razor Ͱ͸ɼMain FF ͔Β൒प‫ظ‬஗Ε. ํ΋༷ʑͰ͋ΓɼͲͷύε͕࠷‫׆ʹޙ‬ੑԽ͞ΕΔ͔͸ຖप. ͨΫϩοΫΛ Shadow Latch ʹ‫͍ͯ͠څڙ‬Δɽt-diagram ্. ‫ʹͱ͝ظ‬ҟͳΔɽ࣮ޮ஗Ԇ͸ग़ྗ͕௚લͷप‫͔ظ‬ΒมԽ͠. ʹ͓͚Δ FF ͷԼͷᒵ৭ͷ࣮ઢ͸ɼTF ͷ‫ݕ‬ग़΢Οϯυ΢Λ. ͳ͔ͬͨ৔߹ʹ͸࣮࣭ 0 ͱͳΓɼΫϦςΟΧϧɾύε͕‫׆‬. ද͍ͯ͠Δɽ‫ݕ‬ग़΢Οϯυ΢ͷɼ্୺Ͱ Main FF ͕ɼԼ୺. ੑԽ͞Εͨ৔߹ʹ͸࠷େͱͳΔɽ͜ͷΑ͏ʹ࣮ޮ஗Ԇ͕ɼ. Ͱ Shadow Latch ͕৴߸ͷαϯϓϦϯάΛߦ͍ɼͦͷ஋Λ. ೖྗͷมԽͷ࢓ํʹΑͬͯେ͖͘͹Βͭ͘͜ͱΛೖྗґଘ. ൺֱ͢ΔͨΊɼ‫ݕ‬ग़΢Οϯυ΢ʹ࣮ޮ஗ԆʹରԠ͢Δ৴߸. ͹Β͖ͭͱ‫Ϳݺ‬ɽೖྗґଘ͹Β͖ͭ͸ଞͷ͹Β͖ͭʹൺ΂. ͕౸ண͍ͯ͠Ε͹ɼTF ͕‫ݕ‬ग़͞ΕΔɽΫϦςΟΧϧɾύε. ͯඇৗʹେ͖͍ [6]ɽ. ͷ஗ԆʹରԠ͢Δ 45◦ ͷ੺ઢ͕‫ݕ‬ग़΢Οϯυ΢ͷԼ୺·Ͱ. ⓒ 2014 Information Processing Society of Japan. 3.

(4) Vol.2014-ARC-210 No.18 Vol.2014-OS-129 No.18 2014/5/15. ৘ใॲཧֶձ‫ڀݚ‬ใࠂ IPSJ SIG Technical Report. ʹ౸ண͢Ε͹ɼϫʔετɾέʔεʹ͓͍ͯ΋ TF ͱͯ͠ॲ ཧ͢Δ͜ͱ͕Ͱ͖Δɽ͕ͨͬͯ͠ɼαΠΫϧɾλΠϜʹର ͢Δ‫ݕ‬ग़΢Οϯυ΢ͷׂ߹Λ α ͱ͢Δͱɼ࠷େ஗Ԇ੍໿͸. ‫Ͳܮܤ‬. ‫ͳܮܤ‬. ܹ‫Ͳܮ‬. (1 + α)τ /1 εςʔδ ͱͳΓɼ୯૬ FF ํࣜΑΓ ατ ͚ͩվ ળ͞ΕΔɽ. 3. μΠφϛοΫɾϩδοΫ. ܹ‫ͳܮ‬. 䝯䝰䝸䝉䝹 䜰䝺䜲. Razor Ͱ͸ɼͦͷద༻ͷର৅ͱͯ͠ελςΟοΫɾϩδο Ϋ͕҉໧ͷ͏ͪʹ૝ఆ͞Ε͓ͯΓɼಛʹμΠφϛοΫɾϓ ϦνϟʔδɾϩδοΫ (dynamic precharged logic) ʹରͯ͠ ͦͷ··ద༻͢Δ͜ͱ͸Ͱ͖ͳ͍ɽ ͜ͷ͜ͱ͸ɼಛʹ SRAM Ͱ໰୊ͱͳΔɽSRAM ͷಡग़ ͠͸௨ৗɼμΠφϛοΫɾϓϦνϟʔδɾϩδοΫͱͯ͠. 䝥䝸䝏䝱䞊䝆 pMOS. ࣮૷͞ΕΔͨΊɼRazor Λͦͷ··ద༻͢Δ͜ͱ͕Ͱ͖ͳ ͍ɽSRAM ͸ɼࠓ೔ͷ LSI ʹ͓͍ͯܽ͘΂͔Β͟ΔཁૉͰ. ܵ‫ܣ‬. ͋ΓɼSRAM ʹద༻Ͱ͖ͳ͍͜ͱ͸ Razor ͷॏେͳܽ఺Ͱ ͋Δͱ‫͑ݴ‬Δɽ. ݄ܲܿ݃. ຊઅͰ͸ɼSRAM ͷμΠφϛοΫͳಈ࡞ʹ͍ͭͯड़΂ɼ. ‫Ͳܦ‬. Razor Λద༻͢Δࡍͷ໰୊఺ʹ͍ͭͯ࿦͡Δɽ 3.1 SRAM ͷߏ੒ͱಈ࡞ ߏ੒. 䝉䞁䝇䞉䜰䞁䝥. ܵ‫ܣ‬. ‫ݐݑ݋ܦ‬0. ‫ͳݐݑ݋ܦ‬. ‫ͳܦ‬. ਤ6. FF. SRAM ͷߏ੒. Ϩϕϧ͸ high ʹอͨΕΔɽ ͜ͷΑ͏ʹɼμΠφϛοΫɾϓϦνϟʔδɾϩδοΫʹ. Ұൠతͳ SRAM ͷϝϞϦɾηϧ͸ 6 τϥϯδελ͔Βͳ. ͓͍ͯ͸ϓϦνϟʔδ͞Εͨϊʔυͷిՙ͕σΟενϟʔ. ΓɼϏοτɾϥΠϯͱΠϯόʔλͷϧʔϓ͕ΞΫηεɾτ. δ͞ΕΔ͔൱͔ʹΑͬͯධՁΛߦ͏ɽϏοτɾϥΠϯͷి. ϥϯδελΛհͯͭ͠ͳ͕Δߏ੒ΛͱΔɽ͔͠͠͹Β͖ͭ. ҐϨϕϧ͕ low ͷঢ়ଶͰਖ਼͍͠ධՁΛߦ͏͜ͱ͸Ͱ͖ͳ͍. ͷ૿Ճʹ൐͍, ͜ͷߏ੒ʹ͓͍ͯಡग़҆͠ఆੑΛอͭ͜ͱ͸. ͔ΒɼධՁΛਖ਼ৗʹߦ͏ͨΊʹ͸ɼϏοτɾϥΠϯͷϓϦ. ࠔ೉ʹͳ͖͍ͬͯͯΔͨΊɼ8 τϥϯδελ͔ΒͳΓϏο. νϟʔδ͕ࣄલʹͳ͞Ε͍ͯͳ͚Ε͹ͳΒͳ͍ɽ. τɾϥΠϯͱΠϯόʔλͷϧʔϓͱ͕෼཭͞Εͨߏ੒ͷϝ ϞϦɾηϧΛ΋ͭ SRAM ΋ଟ͘࢖ΘΕ͖͍ͯͯΔ [8], [9]ɽ. 3.2 Razor ద༻࣌ͷ໰୊. ਤ 6 ʹɼͦͷߏ੒Λࣔ͢ɽͳ͓ɼಉਤͰ͸֤ϝϞϦɾηϧ. 2.2 અͰड़΂ͨΑ͏ʹɼRazor ʹΑΔ TF ‫ݕ‬ग़ͷਖ਼͠͞Λ. ΁ͷॻ͖ࠐΈϙʔτΛলུ͍ͯ͠ΔͨΊτϥϯδελ͕ 2. อূ͢ΔͨΊʹ͸ɼShadow Latch ΁ͷೖྗ͕αϯϓϦϯά. ͭগͳ͍͜ͱʹ஫ҙ͞Ε͍ͨɽ. ࣌఺Ͱਖ਼͘͠ͳͯ͘͸ͳΒͳ͍ɽ͔͠͠μΠφϛοΫɾϓ. ಡΈग़͠ಈ࡞. ϦνϟʔδɾϩδοΫʹ͓͍ͯ͜ΕΛอূ͢Δ͜ͱ͸ࠔ೉. SRAM ͷಡΈग़͠ಈ࡞͸ɼϓϦνϟʔδͱධՁ͕ަ‫ʹޓ‬ ߦΘΕΔ͜ͱͰ࣮‫͞ݱ‬Ε͍ͯΔɽਤ 6 Ͱ͸৴߸ P chg ͕ͦ ͷ੾Γସ͑Λ੍‫͍ͯ͠ޚ‬Δɽ ·ͣ P chg ͕ low Ͱ͋Δؒ͸ɼϓϦνϟʔδ pMOS ͕Φ. Ͱ͋Δɽͦͷ͜ͱΛਤ 7 Λ༻͍ͯड़΂Δɽ ਤ 7ʢࠨʣ͸ର৅ͱ͢Δ SRAM ͷηϯεɾΞϯϓͱ FF ෦෼Λදͨ͠΋ͷͰ͋ΔɽBL ͸ϏοτɾϥΠϯΛදͯ͠ ͍Δɽ. ϯʹͳΔ͜ͱͰɼϏοτɾϥΠϯ BL ͕ high ʹϓϦνϟʔ. ಉਤʢӈʣͷ্ԼͷλΠϛϯάɾνϟʔτ͸ɼͲͪΒ΋. δ͞ΕΔɽ͜ͷ͕ؒϓϦνϟʔδ‫͋Ͱؒظ‬ΓɼධՁͷ‫ؒظ‬. ಉਤʢࠨʣͷճ࿏ͷϏοτɾϥΠϯ BL ͱͦΕΛ૿෯ͨ͠. ͱ۠ผ͞ΕΔɽ. P chg ͕ high Ͱ͋Δؒ͸ධՁͷ‫͋Ͱؒظ‬Δɽ͜ͷ‫Ͱؒظ‬ ͸ϫʔυɾϥΠϯ W L ͕ 1 ߦ͚ͩΞαʔτ͞Εɼͦͷߦͷ ϝϞϦɾηϧͷΞΫηεɾτϥϯδελ͕͢΂ͯΦϯʹͳ. ৴߸ D ͷભҠΛද͍ͯ͠Δɽ্ԼͷλΠϛϯάɾνϟʔτ ͸ͦΕͧΕΫϩοΫप೾਺͕௿͍৔߹ͱߴ͍৔߹ͷಈ࡞ʹ ରԠ͍ͯ͠Δɽ ·ͣɼಉਤʢӈʣ্ଆʹ͓͍ͯɼධՁͷ‫ؒظ‬தʹ BL ͕. ΔɽΞΫηεɾτϥϯδελ͕Φϯʹͳͬͨ΋ͷͷ͏ͪɼ. σΟενϟʔδ͞ΕͯిҐ͕ηϯεɾΞϯϓͷᮢ஋ΛԼճ. ஋͕ high Ͱ͋ΔϝϞϦɾηϧͰ͸ɼϏοτɾϥΠϯΛυ. ΓɼηϯεɾΞϯϓͷग़ྗ D ͕ high ͔Β low ʹભҠͯ͠. ϥΠϒ͢ΔͨΊͷ nMOS ΋Φϯʹͳ͍ͬͯΔͨΊɼରԠ͢. ͍ΔɽҰํɼಉਤʢӈʣԼଆʹ͓͍ͯ͸ɼ্ଆͱಉ࣌ؒ͡Λ. ΔϏοτɾϥΠϯ͕υϥΠϒ͞ΕΔɽҰํɼ஋͕ low Ͱ͋. ͔͚ͯϏοτɾϥΠϯ͕ભҠ͢ΔͱɼMain FF ͷαϯϓϦ. ΔϝϞϦɾηϧʹରԠ͢ΔϏοτɾϥΠϯ͸ු༡͠ɼిҐ. ϯά࣌఺Ͱ͸৴߸ D ͸ high Ͱ͋Δɽ͔͠͠ਖ਼͘͠͸ low. ⓒ 2014 Information Processing Society of Japan. 4.

(5) Vol.2014-ARC-210 No.18 Vol.2014-OS-129 No.18 2014/5/15. ৘ใॲཧֶձ‫ڀݚ‬ใࠂ IPSJ SIG Technical Report. ൈ ͳǤͲ ‫ܮܤ‬. ݄ܲܿ݃. ϏοτɾϥΠϯ͕σΟενϟʔδ͞Ε͍ͯΔ৔߹. SRAM ͷಡग़͠ಈ࡞ʹ͓͍ͯϏοτɾϥΠϯͷมԽ͕‫ى‬ ͜Δ৔߹͸ɼσΟενϟʔδ͞Ε͍ͯͳ͍ঢ়ଶ͔ΒɼσΟ. ‫ܮܤ‬. ܵ‫ܣ‬. Main Latch ͷαϯϓϦϯά࣌఺ͰϏοτɾϥΠϯ͕σΟε. ‫ܦ‬. νϟʔδ͞Ε͍ͯΔ৔߹͸ͦΕҎ߱ͷมԽ͸ੜ͡ͳ͍ͨ. ݉ܽ݅݊ ‫ݐݑ݋ܦ‬. ‫ܦ‬. ‫ݎ݋ݎݎܧ‬. ݄ܲܿ݃. ενϟʔδ͞Εͨঢ়ଶ΁ͷมԽ͔͋͠Γ͑ͳ͍ɽͦͷͨΊɼ. main:0. ൈ ͳǤͷ ݄ܲܿ݃. ‫ݓ݋݄݀ܽݏ‬. ‫ܮܤ‬. 䝍䜲䝭䞁䜾䞉䝣䜷䞊䝹䝖᳨ฟ. ‫ܦ‬. ΊɼͦΕ͕ਖ਼͍͠ධՁͰ͋Δͱ‫ͯ͠ͳݟ‬Α͍ɽͦͯ͠ɼ͜ ͷ৔߹ʹ͸ TF ‫ݕ‬ग़ͷඞཁ͕ͳ͍ͷͰɼShadow Latch ͷα ϯϓϦϯά࣌఺·Ͱ଴ͭඞཁ͕ͳ͘ɼ͙͢ʹϓϦνϟʔδ Λ։࢝͢Δ͜ͱ͕Ͱ͖Δɽ ϏοτɾϥΠϯ͕σΟενϟʔδ͞Ε͍ͯͳ͍৔߹. main:1. shadow:1. Main Latch ͷαϯϓϦϯάͷ࣌఺ͰϏοτɾϥΠϯ͕ σΟενϟʔδ͞Ε͍ͯͳ͍৔߹͸ɼͦͷप‫͍͓ͯʹظ‬ ϏοτɾϥΠϯ͕ຊདྷσΟενϟʔδ͞ΕΔ΂͖͔൱͔ͱ. ਤ 7 Razor ͷμΠφϛοΫɾϩδοΫ΁ͷద༻ͷ໰୊ɿ ʢࠨʣSRAM ͷηϯεɾΞϯϓपลճ࿏ɼ(ӈ) ࠨਤͷλΠϛϯάɾνϟʔτ. ͍͏ਖ਼͍݁͠Ռ͸ɼShadow Latch ͷαϯϓϦϯάͷ࣌఺· Ͱ͸൑அͰ͖ͳ͍ɽ ͔࣮͠͠ࡍʹ͸ɼਖ਼͍݁͠Ռ͕ͲͪΒͰ͋Ζ͏ͱɼ͜ͷ. Λ఻ൖ͠ͳ͚Ε͹ͳΒͣɼTF ͕ൃੜ͢Δ͜ͱ͕෼͔Δɽ. ৔߹͸ϓϦνϟʔδΛඞཁͱ͠ͳ͍͜ͱ͕อূ͞ΕΔɽͳ. ͜ͷ TF ͕‫ݕ‬ग़͞ΕΔͨΊʹ͸ɼShadow Latch ͕αϯϓϦ. ͥͳΒ͹ɼਖ਼͍݁͠Ռ͕σΟενϟʔδ͞ΕΔঢ়ଶͰ͋Δ. ϯά͢Δ݁Ռ͕ਖ਼͍͠ඞཁ͕͋Δɽ͔͠͠ɼShadow Latch. ৔߹ɼσΟενϟʔδ͞Εͳ͍ঢ়ଶͰ͋Δ৔߹ͷͦΕͧΕ. ͷαϯϓϦϯά࣌఺·Ͱʹ D ͕ਖ਼͍݁͠ՌͰ͋Δ low ʹ. ʹ͍ͭͯҎԼͰड़΂ΔΑ͏ʹɼϓϦνϟʔδΛ͠ͳͯ͘΋. ભҠ͢Δ͜ͱ͸ͳ͍ɽͳͥͳΒ͹ɼMain FF ͷαϯϓϦϯ. ໰୊͕ͳ͍͔ΒͰ͋Δɽ. άͷ௚‫ʹޙ‬ϓϦνϟʔδ͕ߦΘΕɼϏοτɾϥΠϯͷిҐ ͕ high ʹҾ্͖͛ΒΕΔ͔ΒͰ͋Δɽ. • ϏοτɾϥΠϯ͕σΟενϟʔδ͞ΕΔ৔߹ɼRazor ʹΑͬͯ Shadow Latch ͷαϯϓϧ࣌఺ʹ͓͍ͯ TF ͕. ͜ͷΑ͏ʹϓϦνϟʔδಈ࡞ͷ΋ͱͰ͸ Shadow Latch. ‫ݕ‬ग़͞ΕɼΞʔΩςΫνϟɾϨϕϧͷख๏ʹΑͬͯ TF. ͕ϩδοΫͷਖ਼͍݁͠ՌΛαϯϓϦϯά͢Δ͜ͱ͕อূ͞. ͔Βͷճ෮͕ߦΘΕΔɽ͕ͨͬͯ͜͠ͷ৔߹ʹ࣍αΠ. Εͳ͍ɽ͢ͳΘͪϓϦνϟʔδಈ࡞͕ TF ΛϚεΫͯ͠͠. ΫϧͷධՁͷͨΊʹଈ࠲ʹϓϦνϟʔδΛߦ͏ඞཁ͸. ·͏ͨΊɼRazor ʹΑͬͯ΋ TF Λ‫ݕ‬ग़Ͱ͖ͳ͍ͱ͍͏໰. ͳ͍ɽ. ୊͕͋Δɽ. 4. ఏҊख๏. • ϏοτɾϥΠϯ͕σΟενϟʔδ͞Εͳ͍৔߹ɼShadow Latch ͷαϯϓϧ࣌఺Ͱ Main FF ͕αϯϓϧͨ͠஋͕ ਖ਼͍͜͠ͱ͕อূ͞ΕΔɽҰํϏοτɾϥΠϯ͸σΟ. ຊઅͰ͸ɼϓϦνϟʔδ৴߸ͷ੍‫ʹޚ‬Αͬͯ SRAM ʹ. ενϟʔδ͞Ε͍ͯͳ͍ͨΊɼϓϦνϟʔδͷඞཁ͕. ର͠ Razor ʹΑΔ TF ‫ݕ‬ग़ͷద༻ΛՄೳʹ͢Δख๏ΛఏҊ. ͳ͘ɼͦͷ··࣍αΠΫϧͷධՁΛ։࢝͢Δ͜ͱ͕Ͱ. ͢ΔɽຊఏҊʹΑͬͯɼSRAM ͷಡग़͠ʹ͓͚Δ࠷େ஗Ԇ. ͖Δɽ. ੍໿Λେ෯ʹ؇࿨͢Δ͜ͱ͕Ͱ͖Δɽ. Ҏ্͔ΒɼMain FF ͷαϯϓϧ࣌఺ͰϏοτɾϥΠϯ͕ σΟενϟʔδ͞Ε͍ͯΕ͹ଈ࠲ʹϓϦνϟʔδ͠ɼͦ͏. 4.1 ఏҊ. Ͱͳ͚Ε͹ϓϦνϟʔδΛ͠ͳ͍Α͏ʹ੍‫͢ޚ‬Δ͜ͱʹ. ఏҊख๏͸ɼMain Latch ͷαϯϓϦϯά࣌఺ͰɼSRAM. ΑͬͯɼRazor ͷద༻͕ՄೳʹͳΔɽ͜ͷ੍‫ޚ‬͸Ϗοτɾ. ͷ֤ϏοτɾϥΠϯ͕σΟενϟʔδ͞Ε͍ͯΔ͔൱͔ʹ. ϥΠϯ஋Λ੍‫ޚ‬৴߸ͱͯ͠ϓϦνϟʔδ৴߸ͷήʔςΟϯ. ΑͬͯɼϓϦνϟʔδΛҎԼͷΑ͏ʹ੍‫͢ޚ‬Δ΋ͷͰ͋Δɽ. άΛߦ͑͹Α͍ɽ. • ϏοτɾϥΠϯ͕σΟενϟʔδ͞Ε͍ͯΔ৔߹ɼϓ ϦνϟʔδΛߦ͏ɽ. • ϏοτɾϥΠϯ͕σΟενϟʔδ͞Ε͍ͯͳ͍৔߹ɼ ϓϦνϟʔδΛߦΘͳ͍ɽ ͜ͷఏҊख๏͕ Razor ͷద༻ΛՄೳʹ͢Δ͜ͱΛɼMain. Latch ͷαϯϓϦϯά࣌఺ͰͷϏοτɾϥΠϯͷධՁͱɼ. ճ࿏ߏ੒ ਤ 8 ʹఏҊख๏ͷճ࿏ߏ੒Λࣔ͢ɽಉਤʢࠨʣͷճ࿏ʹ ରͯ͠ɼಉਤʢӈʣ͸ Razor ͷ TF ‫ݕ‬ग़‫ߏػ‬Λ෇Ճ͠ɼಡ Έग़͠஋ʹԠͨ͡ϓϦνϟʔδ৴߸ͷ੍‫ޚ‬ͷͨΊͷήʔτ ͕෇Ճ͞Ε͍ͯΔɽ ͳ͓ɼಉਤ͸࿦ཧతͳߏ੒Λද͢΋ͷͰ͋Γɼճ࿏࣮૷. ಡग़͠ର৅ͷϝϞϦɾηϧͷ஋ͱͷ૊Έ߹Θͤ͝ͱʹઆ໌. ʹ͓͍ͯ͸ͳΔ΂͘໘ੵ͕গͳ͘ͳΔΑ͏ʹ࣮૷͢Δඞཁ. ͢Δɽ. ͕͋ΔɽFF ΋͘͠͸ϥον͸ɼೖྗͷભҠ͕‫ݶ‬ఆ͞Εͯ. ⓒ 2014 Information Processing Society of Japan. 5.

(6) Vol.2014-ARC-210 No.18 Vol.2014-OS-129 No.18 2014/5/15. ৘ใॲཧֶձ‫ڀݚ‬ใࠂ IPSJ SIG Technical Report. ͍Δ৔߹ɼμΠφϛοΫɾϩδοΫʹΑͬͯ୅༻͢Δ͜ͱ ͕Ͱ͖ɼτϥϯδελ਺Λେ෯ʹ࡟‫͢ݮ‬Δ͜ͱ͕ՄೳͰ͋ Δ [10]ɽ ܵ‫ܣ‬. ͜ͷ͜ͱΛར༻ͯ͠τϥϯδελ਺Λ཈͑ͨճ࿏࣮૷Λ ਤ 9 ʹࣔ͢ɽಉਤʹ͸ਤ 8 ͷ֤‫ػ‬ೳͱͷେ·͔ͳରԠΛࣔ. ‫ݐݑ݋ܦ‬. ‫ܦ‬. ͍ͯ͠Δɽͨͩ͠ Main FF ΍ Shadow Latch ͷ‫ػ‬ೳ͸ͦΕ. ݄ܲܿ݃. ͧΕʹ෼ࢄ͞Ε͍ͯΔͨΊɼͦΕΒͷରԠ͸͍ࣔͯ͠ͳ͍ɽ ‫ܮܤ‬. ྫ͑͹ɼಉਤͷηϯεɾΞϯϓ෦෼ͷ࠷΋ೖྗʹ͍ۙΠ ϯόʔλ͸ɼग़ྗ͕ high Ͱ͋Ε͹ɼͦΕΛϓϦνϟʔδ‫ظ‬ 吮 呁 吘 吺 呎 吏 ไ ᚚ. ؒͷؒೖྗ͔Β੾Γ཭ͯ͠อ࣋Ͱ͖Δɽ͜Ε͸ɼΠϯόʔ λͷग़ྗϊʔυͱ‫ݩ‬ʑ͋Δ nMOS ͷؒʹ෇Ճ͞Εͨ nMOS ʹΑΔ΋ͷͰ͋Δɽ͜ͷ nMOS ͸ϓϦνϟʔδ੍‫ޚ‬৴߸Ͱ ͋Δ N P chg Λήʔτೖྗͱ͓ͯ͠ΓɼN P chg ͕ low ͷ ‫ؒظ‬ɼ͢ͳΘͪϓϦνϟʔδ‫͚͓ʹؒظ‬ΔΠϯόʔλͷग़. ܵ‫ܣ‬. ݉ܽ݅݊. ‫ܦ‬. ‫ݐݑ݋ܦ‬ ‫ݎ݋ݎݎܧ‬. ‫ݓ݋݄݀ܽݏ‬. ྗ஋͕ϏοτɾϥΠϯͷϓϦνϟʔδʹ൐ͬͯ low ʹͳͬ. ݄ܲܿ݃. 䝍䜲䝭䞁䜾䞉䝣䜷䞊䝹䝖᳨ฟ. ͯ͠·͏͜ͱΛ๷͍Ͱ͍Δɽ ·ͨɼμΠφϛοΫɾϩδοΫʹΑΔ஋อ࣋ͷ෦෼͸ɼ. Main FF ͷαϯϓϦϯά࣌఺ͰͷϏοτɾϥΠϯঢ়ଶͷ‫ه‬. ਤ8. ʢ্ʣSRAM ͷηϯεɾΞϯϓपลճ࿏ɼʢԼʣఏҊख๏Λద ༻ͨ͠ճ࿏. Աʹ༻͍ΒΕ͍ͯΔɽMain FF ͷαϯϓϦϯά࣌఺ʹ͓͍ ͯϏοτɾϥΠϯ͕σΟενϟʔδ͞Ε͍ͯͳ͍ͱ͖ɼ͜ ‫ܮܤ‬. ͷμΠφϛοΫɾϩδοΫͷग़ྗϊʔυ͸σΟενϟʔ δ͞ΕΔɽͦͷ‫ޙ‬ϏοτɾϥΠϯͷిҐ͕Լ͕ͬͨͱͯ͠. 䝉䞁䝇䞉䜰䞁䝥. ΋࠶ͼग़ྗϊʔυͷిҐ͕ high ʹ໭Δ͜ͱ͸ͳ͍ͨΊɼ. ܵ‫ܧܣ‬. Main FF ͷαϯϓϦϯά࣌఺ʹ͓͍ͯͷϏοτɾϥΠϯͷ. ‫ܦ‬. ঢ়ଶΛอ࣋Ͱ͖͍ͯΔͱ‫͑ݴ‬Δɽ. ‫ݐݑ݋ܦ‬. ݄ܰܲܿ݃. TF ‫ݕ‬ग़෦෼Ͱ͸ɼμΠφϛοΫɾϩδοΫʹΑΔϥο. 吮 呁 吘 吺 呎 吏 ไ ᚚ. νͷ෦෼Ͱ‫ه‬Ա͞Ε͍ͯΔ Main FF ͷαϯϓϦϯά஋ͱɼ ηϯεɾΞϯϓ͕༩͑ΔϏοτɾϥΠϯ஋ͱΛೖྗͱ͠ɼ. Error ৴߸ͷධՁΛߦ͍ͬͯΔɽ ·ͨɼಉਤʹ໌͍ࣔͯ͠ͳ͍͕ɼDout ͱ Error ͸ͱ΋ʹ. ‫ݎ݋ݎݎܧ‬. 咥 咗 ᳨ ฟ ݄ܲܿ݃. ϓϦνϟʔδ͞ΕΔඞཁ͕͋Δɽͨͩ͜͠ΕΒ͸ଞͷϏο τɾϥΠϯͱ‫ڞ‬༗͢ΔͨΊɼඞͣ͠΋ϏοτɾϥΠϯ͝ͱ. 䝎䜲䝘䝭䝑䜽䞉䝻䝆䝑䜽䛻䜘䜛್ಖᣢ. ʹ pMOS ͕ඞཁͱͳΔ΋ͷͰ͸ͳ͍ɽ ಉਤͷ࣮૷ʹ͓͍ͯ͸ɼμΠφϛοΫɾϩδοΫʹΑΔ ϥον‫ػ‬ೳͷ୅༻ʹΑͬͯಉ༷ʹল໘ੵԽ͞Εͨηϯεɾ ΞϯϓΛϕʔεͱͯ͠ɼτϥϯδελ਺Λ 3.5 ഒͷ૿Ճʹ. ਤ 9 ఏҊख๏ͷճ࿏࣮૷. ·Ͱ཈͑Δ͜ͱ͕Ͱ͖Δɽ ࠷େ஗Ԇ੍໿ͷ؇࿨ ఏҊख๏Λద༻ͨ͠ SRAM ΞΫηεεςʔδͷ t-diagram. ͋ Δ ɽͦ ͷ ࠷ େ ஗ Ԇ ੍ ໿ ͸ (1.0 − β)τ /0.5 εςʔδ ʹ. 0.5τ /0.5 εςʔδ Ͱ͋Δɽ. Λਤ 10 ʹࣔ͢ɽ͜͜ͰͷධՁ͸ϫʔυɾϥΠϯ͕Ξαʔ. ҰํɼఏҊख๏Λద༻ͨ͠ਤ 10(ӈ) Ͱ͸ɼTF ͱͯ͠ॲ. τ͞Ε͔ͯΒɺσΟενϟʔδ͕‫ྃ͢׬‬Δ·Ͱͱ͢Δɽͭ. ཧͰ͖Δ‫ݶ‬ք·Ͱ࠷େ஗Ԇ੍໿Λ؇࿨͢Δ͜ͱ͕Ͱ͖Δɽ. ·Γ৴߸ͷ఻ୡΛද͢໼ҹͷɼ։࢝఺͸σίʔμʹΑΔ. εςʔδͷ࠷େ஗ԆʹରԠ͢Δ 45˃ͷ੺ઢ͕‫ݕ‬ग़΢Οϯ. ϫʔυɾϥΠϯͷΞαʔτͷ։࢝Λද͠ɼऴண఺͸σΟε. υ΢ͷԼ୺·Ͱʹ౸ண͢Ε͹ TF ͱͯ͠ॲཧ͢Δ͜ͱ͕Ͱ. νϟʔδ͕ͳ͞ΕΔ৔߹͸σΟενϟʔδͷ‫ྃ׬‬Λද͢ɽ. ͖ΔͨΊɼαΠΫϧɾλΠϜʹର͢Δ‫ݕ‬ग़΢Οϯυ΢ͷׂ. σΟενϟʔδ͕ͳ͞Εͳ͍࣌ʹ͸໼ҹ͸ਫฏͰ͋Δɽ·. ߹Λ α ͱ͢Δͱɼ࠷େ஗Ԇ੍໿͸ (0.5 + α)τ /0.5 εςʔδ. ͨɼ1 प‫͚͓ʹظ‬ΔϓϦνϟʔδ‫ؒظ‬ͷׂ߹ β Λ 0.5 ͱ͠. ͱͳΔɽԾʹ α = 0.5 ͱ͢ΔͱɼಉਤͰࣔ͞ΕΔΑ͏ʹɼ. ͍ͯΔɽ. ΫϩοΫप೾਺Λ࠷େ 2.0 ഒ·ͰҾ্͖͛Δ͜ͱ͕Մೳͱ. ਤ 10(ࠨ) ͸ ɼఏ Ҋ ख ๏ ద ༻ લ ͷ ճ ࿏ ͷ t-diagram Ͱ. ⓒ 2014 Information Processing Society of Japan. ͳΔɽ. 6.

(7) Vol.2014-ARC-210 No.18 Vol.2014-OS-129 No.18 2014/5/15. ৘ใॲཧֶձ‫ڀݚ‬ใࠂ IPSJ SIG Technical Report. 1.0[V]. ൈ ʹǤͲ. ݈ܿ݇. ܹ‫Ͳܮ‬ǡ ܹ‫ͳܮ‬. ݄ܲܿ݃. 250[ps]. 1.00V 0.90V 0.85V 0.80V. ‫ܮܤ‬. ݄ܰܲܿ݃ ‫ݐݑ݋ܦ‬. ‫ݎ݋ݎݎܧ‬. ਤ 12 SPICE γϛϡϨʔγϣϯ೾‫ܗ‬ɽαΠΫϧλΠϜ:250[ps]ɼԹ ౓:55[ˆ]ɽ ਤ 10. SRAM ͷ t-diagramɿ ʢࠨʣఏҊख๏ద༻લ ɼ ʢӈʣఏҊख๏ద ༻‫ޙ‬. 5.3 ݁Ռ ਤ 12 ͸ SPICE γϛϡϨʔγϣϯʹΑΔ݁ՌͰ͋Δɽ͜ ද 1 ධՁʹ༻͍ͨιϑτ΢ΣΞ‫ڥ؀‬. ͜ͰɼW L0ɼW L1 ͸ͦΕͧΕϝϞϦɾηϧ 0 ͱ 1 ͷΞΫ. ճ࿏ɾϨΠΞ΢τΤσΟλ. Virtuoso Version IC6.1.5 ISR15. RC நग़. Calibre xACT3D Version 2012.3.31 26. γϛϡϨʔγϣϯ. HSPICE Version H-2013.03. ϓϦνϟʔδ৴߸ɼBL ͸ϏοτɾϥΠϯɼN P chg ͸ήʔ. ϥΠϒϥϦ. FreePDK45nm [11]. ςΟϯά‫ޙ‬ͷϓϦνϟʔδ৴߸ɼDout ͱ Error ͸ਤ 11 Ͱ. ηεɾτϥϯδελΛ੍‫͢ޚ‬ΔϫʔυϥΠϯ৴߸ɼP chg ͸. ද͞ΕΔΑ͏ʹͦΕͧΕಡΈग़͠஋ͷग़ྗͱ TF ‫ݕ‬ग़৴߸ Ͱ͋Δɽ͜ΕΒͷ৴߸೾‫͕ܗ‬ɼి‫ిݯ‬ѹͷҟͳΔ৔߹ʹͭ ݄ܲܿ݃. ͍ͯॏͶͯදࣔ͞Ε͍ͯΔɽ. ‫݁݊݅ܮݐ݅ܤ‬ ‫ܮ ʹܯ‬Τܵ ൌ ͹Ͳ݊݉Τ͹Ͳ݊݉ ൈ ʹͳǤ͸ߤ݉. ܹ‫Ͳܮ‬. ܹ‫ͳܮ‬. AMP. ܹ‫݅ܮ‬. ‫ݐݑ݋ܦ‬ ‫ݎ݋ݎݎܧ‬. ‫݁݊݅ܮݐ݅ܤ‬ ‫݁ܿ݊ܽݐ݅ܿܽ݌ܽܥ‬ሺͶǤ͸݂‫ܨ‬ሻ. ࠷ॳͷप‫Ͱظ‬͸ W L0 ͕Ξαʔτ͞ΕɼϝϞϦɾηϧ 0 ͕ಡΈग़͞ΕΔɽͲͷి‫ిݯ‬ѹ৚݅ʹ͓͍ͯ΋ɼP chg ͕. high ʹͳ͔ͬͯΒϏοτɾϥΠϯ͕؇΍͔ʹσΟενϟʔ δ͞Ε͸͡Ί͍ͯΔ͜ͱ͕‫ͯݟ‬औΕΔɽి‫ిݯ‬ѹ͕ 0.85[V]. ͳ͸݉݁݉‫ݏ݈݈݁ܿݕݎ݋‬. ਤ 11 SPICE γϛϡϨʔγϣϯ༻ճ࿏. ΑΓߴ͍ঢ়ଶͰ͸ɼಡΈग़͠͸໰୊ͳ͘ߦΘΕ͍ͯΔɽҰ ํి‫ిݯ‬ѹ͕ 0.80[V] ͷঢ়ଶͰ͸ɼಡΈग़͠σʔλͷ৴߸. Dout ͸ᮢ஋ۙ͘ʹ͋Γɼ͜ΕΑΓ௿͍ి‫ిݯ‬ѹͷ‫ڥ؀‬Լʹ ͓͍ͯ TF ͕ൃੜ͢Δ֬཰͸ߴ͍ͱ͍͑Δɽ͔͜͠͠ͷͱ. 5. ධՁ 4 અͰड़΂ͨఏҊख๏ʹରͯ͠ɼSPICE γϛϡϨʔγϣ ϯʹΑͬͯಈ࡞֬ೝΛߦͬͨɽ. ͖Τϥʔ৴߸ Error ͕Ξαʔτ͞ΕɼTF Λ‫ݕ‬ग़͢Δ͜ͱ ͕Ͱ͖͍ͯΔɽ࣮ࡍ͸ TF ͕‫ݕ‬ग़͞Εͨ৔߹͸ TF ͔Βͷ ճ෮ॲཧΛߦ͏͜ͱʹͳΔɽ ࣍ͷप‫Ͱظ‬͸ɼW L1 ͕Ξαʔτ͞ΕɼϝϞϦɾηϧ 1 ͕ಡΈग़͞ΕΔɽ͜ͷͱ͖͸ BL ͷσΟενϟʔδ͸‫͜ى‬. 5.1 ධՁ‫ڥ؀‬ ද 1 ʹධՁʹ༻͍ͨιϑτ΢ΣΞͱςΫϊϩδɾϥΠϒ ϥϦΛࣔ͢ɽ. Βͳ͍ɽP chg ͕ low ʹͳΔλΠϛϯάͰɼN P chg ͸ήʔ ςΟϯά͞Εͯ high ͷ··Ͱ͋ΓɼධՁ͕‫ܧ‬ଓ͍ͯ͠Δɽ ࣍ͷप‫ظ‬ͷ։࢝·Ͱʹ BL ͸σΟενϟʔδ͞Εͳ͔ͬͨ ͨΊɼTF ͸ൃੜ͍ͯ͠ͳ͍ɽ·ͨɼBL ͸ిҐ͕ high Ϩ. 5.2 ධՁखॱ ਤ 11 ʹࣔ͢ճ࿏Λ༻͍ͯධՁΛߦ͏ɽ ϏοτɾϥΠϯʹରͯ͠ϝϞϦɾηϧ 16 ‫ͱݸ‬ɼ 4.6f [F ] ͷ഑ઢ༰ྔΛ෇Ճ͢Δɽ഑ઢ༰ྔʹ͍ͭͯ͸ɼಉ‫ن‬໛ͷϨ. ϕϧʹ͋ΔͷͰɼධՁΛߦ͏͜ͱ͕Ͱ͖Δɽ. 6. ·ͱΊͱࠓ‫ޙ‬ͷ՝୊ ຊߘͰ͸ɼμΠφϛοΫɾϩδοΫʹΑ࣮ͬͯ૷͞Εͨ. δελɾϑΝΠϧͷϨΠΞ΢τઃ‫ܭ‬σʔλ͔Βநग़Λߦ͍ɼ. SRAM ͷಡग़͠ʹରͯ͠ɼϓϦνϟʔδ৴߸ͷ੍‫ʹޚ‬Αͬ. ϏοτɾϥΠϯ͢΂ͯʹ͍ͭͯฏ‫ۉ‬ΛͱΔ͜ͱͰࢉग़ͨ͠ɽ. ͯ Razor Λద༻͠ɼ࠷େ஗Ԇ੍໿Λ؇࿨͢Δख๏ͷఏҊͱ. ϝϞϦɾηϧ 0 ͱ 1 ʹ͋Β͔͡Ί 0, 1 Λॻ͖ࠐΈɼి‫ݯ‬. ͦͷಈ࡞֬ೝΛߦͬͨɽࠓ‫ޙ‬͸ɼਖ਼ৗͳ݁Ռ͕ಘΒΕΔಈ. ిѹΛมಈͤͭͭ͞ʢ1.00[V]ɼ0.90[V]ɼ0.85[V]ɼ0.80[V]ʣ ɼ. ࡞৚݅ͷൣғͱɼΤϥʔ͕ग़ྗ͞ΕΔಈ࡞৚݅ͷൣғͷॏ. ަ‫ʹޓ‬ಡग़͠ΞΫηεΛߦͬͨɽ. ͳΓΛे෼খ͘͢͞ΔΑ͏ʹճ࿏ͷ࠷దԽΛߦ͍ɼఏҊख. ⓒ 2014 Information Processing Society of Japan. 7.

(8) Vol.2014-ARC-210 No.18 Vol.2014-OS-129 No.18 2014/5/15. ৘ใॲཧֶձ‫ڀݚ‬ใࠂ IPSJ SIG Technical Report. ๏Λద༻ͨ͠ϨδελɾϑΝΠϧͷੑೳ޲্ʹ͍ͭͯධՁ ͢Δɽ·ͨɼυϛϊͷΑ͏ͳଟஈͷμΠφϛοΫɾϩδο Ϋʹରͯ͠ຊߘͰఏҊͨ͠ख๏Λద༻͢Δࡍʹൃੜ͢Δ໘ ੵ૿ՃΛ௿‫͢ݮ‬Δख๏ͷߟ࡯Λߦ͏ɽ. [11] [12]. University, N. C. S.: NCSU EDA Wiki. http://www. eda.ncsu.edu/wiki/NCSU_EDA_Wiki. Ashish, S., Dennis, S. and David, B.: Statistical Analysis and Optimization for VLSI: Timing and Power, ISBN: 9780-387-25738-9 (2005).. ँࣙ ϨδελɾϑΝΠϧͷ഑ઢ༰ྔͷநग़ʹ͓͍ͯɼ Ԙ୩ ྄ଠ ॿ‫ݹ໊( ڭ‬԰େֶେֶӃɾ޻ֶ‫ڀݚ‬Պ) ʹσʔλ Λ͍͍ͨͩͨɽ͜ͷ৔Λ͓आΓͯ͠‫ँײ‬ਃ্͛͠Δɽ ຊ‫ڀݚ‬ͷҰ෦͸ɼJST CRESTʮσΟϖϯμϒϧ VLSI γ εςϜͷ‫ج‬൫ٕज़ʯʮΞʔΩςΫνϟͱ‫ࣜܗ‬త‫ʹূݕ‬ΑΔ ௒σΟϖϯμϒϧ VLSIʯ ɼ͓ΑͼՊֶ‫ڀݚ‬අิॿۚ‫ج‬൫‫ݚ‬ ‫( ڀ‬B)ɾ26280012ʮϨδϦΤϯεࢦ޲ίϯϐϡʔλγες Ϝʹؔ͢Δ‫ڀݚ‬ʯͷࢧԉʹΑΓߦΘΕͨɽ·ͨɼຊ‫ڀݚ‬ͷ Ұ෦͸౦‫ژ‬େֶେ‫ن‬໛ूੵγεςϜઃ‫ڭܭ‬ҭ‫ڀݚ‬ηϯλʔ Λ௨͠ɼγϊϓγε‫ࣜג‬ձࣾɼ೔ຊέΠσϯε‫ࣜג‬ձࣾɼ ϝϯλʔ‫ࣜג‬ձࣾͷ‫ߦͰྗڠ‬ΘΕͨ΋ͷͰ͋Δɽ ࢀߟจ‫ݙ‬ [1]. [2]. [3]. [4]. [5]. [6]. [7]. [8]. [9]. [10]. ฏຊढ़࿠, ஛಺ܿ, ੢ాজஉ: MOS τϥϯδελͷεέʔ Ϧϯάʹ൐͏ಛੑ͹Β͖ͭ, ిࢠ৘ใ௨৴ֶձࢽ, Vol. 92, No. 6 (2009). D.Ernst, N.Kim, S.Das, S.Pant, T.Pham, R.Rao, C.Ziesler, D.Blaauw, T.Austin and T.Mudge: Razor: A Low-Power Pipeline Based on Circuit-Level Timing Speculation, Int’l Symp. on Microarchitecture (MICRO), pp. 7–18 (2003). Blaauw, D., Kalaiselvan, S., Lai, K., Ma, W.-H., Pant, S., Tokunaga, C., Das, S. and Bull, D.: Razor II: In Situ Error Detection and Correction for PVT and SER Tolerance, Int’l Symp. on Solid-State Circuits Conference (ISSCC), pp. 32 – 48 (2008). Bull, D., Das, S., Shivshankar, K., Dasika, G., Flautner, K. and Blaauw, D.: A power-efficient 32b ARM ISA processor using timing-error detection and correction for transient-error tolerance and adaptation to PVT variation, Solid-State Circuits Conference Digest of Technical Papers (ISSCC), 2010 IEEE International, pp. 284 –285 (2010). Mallik, A., Cosgrove, J., Dick, R. P., Memik, G. and Dinda, P.: PICSEL: Measuring User-Perceived Performance to Control Dynamic Frequency Scaling, Int’l Conf. on Architectural Support for Programming Languages and Operating Systems (ASPLOS), pp. 70–79 (2008). ‫ت‬ଟ‫و‬৴, Ԙ୩྄ଠ, ‫ޒ‬ౡਖ਼༟, ࡔҪमҰ: λΠϛϯά੍໿ Λ؇࿨͢ΔΫϩοΩϯάํࣜ, ઌਐత‫جࢉܭ‬൫γϯϙδ΢ Ϝ SACSIS, pp. 347–354 (2010). ‫ޒ‬ౡਖ਼༟, ૔ా੒‫ݾ‬, Ԙ୩྄ଠ, ࡔҪमҰ: λΠϛϯάɾ ϑΥʔϧτ଱ੑΛ࣋ͭ Out-of-Order ϓϩηοα, ৘ใॲཧ ֶձ࿦จࢽɿίϯϐϡʔςΟϯάγεςϜ, Vol. 53, No. 41, pp. 17–30 (2012). Chang, L., Fried, D. M., Hergenrother, J., Sleight, J. W., Dennard, R. H., Montoye, R. K., Sekaric, L., McNab, S. J., Topol, A. W., Adams, C. D., Guarini, K. W. and Haensch, W.: Stable SRAM cell design for the 32 nm node and beyond, VLSI Technology, 2005. Digest of Technical Papers. 2005 Symposium on, pp. 128–129 (2005). ID: 1. Kumar, R. and Hinton, G.: A family of 45nm IA processors, Solid-State Circuits Conference - Digest of Technical Papers, 2009. ISSCC 2009. IEEE International, pp. 58–59 (2009). ID: 1. Harris, D.: Skew-tolerant Circuit Design, Morgan Kaufmann Publishers, pp. 12–14 (2001).. ⓒ 2014 Information Processing Society of Japan. 8.

(9)

参照

関連したドキュメント

This paper derives a priori error estimates for a special finite element discretization based on component mode synthesis.. The a priori error bounds state the explicit dependency

The set of families K that we shall consider includes the family of real or imaginary quadratic fields, that of real biquadratic fields, the full cyclotomic fields, their maximal

Vovelle, “Existence and uniqueness of entropy solution of scalar conservation laws with a flux function involving discontinuous coefficients,” Communications in Partial

If in the infinite dimensional case we have a family of holomorphic mappings which satisfies in some sense an approximate semigroup property (see Definition 1), and converges to

In Section 3 using the method of level sets, we show integral inequalities comparing some weighted Sobolev norm of a function with a corresponding norm of its symmetric

As is well-known, this is an ill-posed problem Using the Tikhonov method, the authors give a regularized solution, and assuming the (unknown) exact solution is in H(R),a > 0

Tuan, Regularization and error estimate for the nonlinear backward heat problem using a method of integral equation., Nonlinear Anal., Volume 71, Issue 9, 2009, pp.. Trong

Easy to see that in this case the direction of B should be purely rational such that the orthogonal plane (B) contains two different reciprocal lattice vectors. It is evident also