Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
TCP Offload through Connection Handoffを読んだ
Search
Sponsored
·
Your Podcast. Everywhere. Effortlessly.
Share. Educate. Inspire. Entertain. You do you. We'll handle the rest.
→
Yuuki Tsubouchi (yuuk1)
November 01, 2012
Research
0
2.2k
TCP Offload through Connection Handoffを読んだ
論文紹介スライド
Yuuki Tsubouchi (yuuk1)
November 01, 2012
Tweet
Share
More Decks by Yuuki Tsubouchi (yuuk1)
See All by Yuuki Tsubouchi (yuuk1)
AIスーパーコンピュータにおけるLLM学習処理性能の計測と可観測性 / AI Supercomputer LLM Benchmarking and Observability
yuukit
1
650
SREはサイバネティクスの夢をみるか? / Do SREs Dream of Cybernetics?
yuukit
3
390
SREのためのテレメトリー技術の探究 / Telemetry for SRE
yuukit
13
3k
AIスパコン「さくらONE」の オブザーバビリティ / Observability for AI Supercomputer SAKURAONE
yuukit
2
1.2k
AIスパコン「さくらONE」のLLM学習ベンチマークによる性能評価 / SAKURAONE LLM Training Benchmarking
yuukit
2
970
とあるSREの博士「過程」 / A Certain SRE’s Ph.D. Journey
yuukit
11
5.5k
eBPFを用いたAIネットワーク監視システム論文の実装 / eBPF Japan Meetup #4
yuukit
3
1.7k
クラウドのテレメトリーシステム研究動向2025年
yuukit
4
1.2k
博士論文公聴会: Scaling Telemetry Workloads in Cloud Applications: Techniques for Instrumentation, Storage, and Mining / PhD Defence
yuukit
1
520
Other Decks in Research
See All in Research
大規模言語モデルにおけるData-Centric AIと合成データの活用 / Data-Centric AI and Synthetic Data in Large Language Models
tsurubee
1
500
[Devfest Incheon 2025] 모두를 위한 친절한 언어모델(LLM) 학습 가이드
beomi
2
1.4k
生成AI による論文執筆サポート・ワークショップ 論文執筆・推敲編 / Generative AI-Assisted Paper Writing Support Workshop: Drafting and Revision Edition
ks91
PRO
0
120
Earth AI: Unlocking Geospatial Insights with Foundation Models and Cross-Modal Reasoning
satai
3
480
Self-Hosted WebAssembly Runtime for Runtime-Neutral Checkpoint/Restore in Edge–Cloud Continuum
chikuwait
0
330
世界モデルにおける分布外データ対応の方法論
koukyo1994
7
1.5k
SkySense V2: A Unified Foundation Model for Multi-modal Remote Sensing
satai
3
500
ローテーション別のサイドアウト戦略 ~なぜあのローテは回らないのか?~
vball_panda
0
280
20年前に50代だった人たちの今
hysmrk
0
140
第66回コンピュータビジョン勉強会@関東 Epona: Autoregressive Diffusion World Model for Autonomous Driving
kentosasaki
0
350
Agentic AI フレームワーク戦略白書 (2025年度版)
mickey_kubo
1
120
学習型データ構造:機械学習を内包する新しいデータ構造の設計と解析
matsui_528
6
3.2k
Featured
See All Featured
Gemini Prompt Engineering: Practical Techniques for Tangible AI Outcomes
mfonobong
2
280
Lessons Learnt from Crawling 1000+ Websites
charlesmeaden
PRO
1
1.1k
Tips & Tricks on How to Get Your First Job In Tech
honzajavorek
0
440
Rebuilding a faster, lazier Slack
samanthasiow
85
9.4k
The Curse of the Amulet
leimatthew05
1
8.7k
What’s in a name? Adding method to the madness
productmarketing
PRO
24
3.9k
Claude Code どこまでも/ Claude Code Everywhere
nwiizo
61
52k
Agile that works and the tools we love
rasmusluckow
331
21k
VelocityConf: Rendering Performance Case Studies
addyosmani
333
24k
Building a Modern Day E-commerce SEO Strategy
aleyda
45
8.7k
GitHub's CSS Performance
jonrohan
1032
470k
How to make the Groovebox
asonas
2
1.9k
Transcript
5$10GqPBEUISPVHI $POOFDUJPO)BOEPGG )ZPOHZPVC,JNBOE4DPUU3JYOFS3JDF6OJWFSTJUZ *O1SPDFFEJOHTPG&VSP4ZT JEZ@VVLJ!Z@VVLJ
*OUSPEVDUJPO ‣ 5$1ॲཧʹ͓͍ͯɼϝϞϦΞΫηε͕ϘτϧωοΫ ‣ ύέοτ͋ͨΓͷॲཧʹཁ͢Δ໋ྩͷ͕ϝϞϦΞΫηε ‣ ;FSP$PQZ*0νΣοΫαϜܭࢉͷΦϑϩʔυͷΑ͏ͳٕज़Ͱɹɹɹɹɹ zύέοτσʔλzʹର͢ΔϝϞϦΞΫηε͑ΒΕΔ ‣ ;FSP$PQZ*0Χʔωϧۭ͔ؒΒϢʔβۭؒʹσʔλίϐʔΛݮͯ͠ޮԽ
‣ ݱࡏͰɼzίωΫγϣϯσʔλߏମzͷΞΫηε͕ϘτϧωοΫ ‣ ίωΫγϣϯ૿ՃʹΑΓσʔλߏମ͕$16ΩϟογϡʹͷΒͳ͘ͳΔ ‣ /*$ͷΦϑϩʔυʹΑΓϝϞϦΞΫηεݮΒͤΔ ‣ ͔͠͠ɼίωΫγϣϯ͕૿Ճ͢Δͱ/*$ͷϝϞϦʹͷΒͳ͘ͳΔ
*OUSPEVDUJPO ‣ 5$1ίωΫγϣϯϋϯυΦϑ ‣ /*$ʹશͯͷίωΫγϣϯΛΦϑϩʔυͤͣʹɼҰ෦ͷίωΫγϣ ϯΛΦϑϩʔυ͢Δ ‣ 04͍ͭͰΦϑϩʔυͨ͠ίωΫγϣϯΛऔΓͤΔͨΊɼ/*$ ͱ$16ؒͷλεΫΛશʹίϯτϩʔϧͰ͖Δɽ ‣
/*$Ϧιʔε͕গͳ͘ͳͬͨΒίωΫγϣϯΛ04ʹฦͤΔ ‣ /*$௨ৗ*1ϔομΛॲཧ͢Δ͕ɼϋϯυΦϑΛ͏ͱɼ͢Ͱʹ֬ ཱ͞ΕͨίωΫγϣϯਖ਼͍͠ϙʔτͱϧʔςΟϯάใΛͭͨ Ίɼ/*$ϧʔςΟϯά͠ͳͯ͘Α͍
*OUSPEVDUJPO ‣ 5$1ίωΫγϣϯϋϯυΦϑطଘͷఏҊͰ͋Γɼɹɹɹɹɹɹɹɹɹ ຊจͰ5$1ίωΫγϣϯϋϯυΦϑͷઃܭͱ࣮ΛఏҊ͢Δ ‣ ಠࣗͷมߋΛՃ͑ͨ'SFF#4%ͱɹɹɹɹɹɹɹɹɹɹɹɹɹɹɹɹɹɹ ɹ ɹ ϓϩάϥϚϒϧͳΪΨϏοτΠʔαωοτίϯτϩʔϥʹΑΓ࣮ ‣
࣮ݧͷ݁Ռɼ֤ύέοτॲཧʹཁ͢Δ$16αΠΫϧΛݮͨ͠ ‣ ໋ͳίωΫγϣϯʹ͔͔ΘΒͣɼ41&$XFCΣϒαʔόͷɹ εϧʔϓοτ্͕ ‣ 04ͷมߋιέοτҎԼͷωοτϫʔΫελοΫʹݶఆ͞Ε͍ͯΔ ‣ ଞͷ#4%ܥ04Ͱ͜ͷϋϯυΦϑΠϯλϑΣʔεΛ༻͢Δͷ༰қ
/FUXPSL4UBDL1FSGPSNBODF ‣ 5$1ෳࡶͳϓϩτίϧͰ͋Γɼඞཁͳ$16αΠΫϧ͕େ͖͍ ‣ (I[$16ɼ,#-ΩϟογϡͰ#Λ.CTͰૹ৴ ‣ 5$1ϝϞϦΞΫηε໋ྩ͕ଟ͍ʢʣ ‣ ίωΫγϣϯ͕ଟ͍ͱ͖ʢ%ͱ&ʣʹ4UBMMTʢͪ࣌ؒʣ͕ଟ͍ ‣
5$1"$$&-&3"5*0/ ‣ طଘख๏Ͱ͋Δϑϧ5$1Φϑϩʔυͷ ‣ ܭࢉࢿݯ ‣ (CTͰϑϧʹ௨৴͢ΔͱඵؒͰ(ݸͷ໋ྩ ‣ ϓϩάϥϚϒϧͳϓϩηοαͰͳ͘ઐ༻ͷϋʔυΣΞ͕ඞཁ ‣
ίετ͕͔͔Δ͚ͩͰͳ͘ɼγεςϜͷॊೈੑଛͳ͏ ‣ ϝϞϦ༰ྔ ‣ /*$ͷϝϞϦओهԱΑΓߴ͕ͩ༰ྔ͕খ͍͞ ‣ ಉ͡ίετͰ༰ྔΛେ͖͘͢ΔͱʹͳΓɼΦϑϩʔυ͢Δҙຯ͕ͳ͍ ‣ ιϑτΣΞΞʔΩςΫνϟͷෳࡶ͞ ‣ ϙʔτ൪߸ͷׂΓৼΓͱ*1ϧʔςΟϯά୯ҰͷΠϯλϑΣʔε͚ͩͰܾΊΒΕͳ͍ ‣ ෳͷΠϯλϑΣʔεΛ·͍ͨͩάϩʔόϧͳઃܭ͕ඞཁ
$POOFDUJPO)BOEPGG ‣ ίωΫγϣϯϋϯυΦϑͷϓϩηε ‣ 04͕ίωΫγϣϯΛཱ֬͢Δ ‣ 04͕ओهԱ͔Β/*$ͱίωΫγϣϯͷঢ়ଶΛసૹ͢Δ ‣ 04ͦͷίωΫγϣϯʹର͢Δ5$1ॲཧΛҰ࣌ఀࢭ͢Δ ‣
Φϑϩʔυޙʹ04ΞϓϦέʔγϣϯ͔Β/*$ͷϦΫΤετΛதܧ͢Δ ‣ 04͕ίωΫγϣϯΛཱ֬͢ΔͨΊɼ04ϙʔτ൪߸ͷ֬อҎ্ͷ੍ޚ ΛͯͨΓɼϧʔςΟϯάͷܾఆ͕Ͱ͖Δ
"SDIJUFDUVSF ‣ /*$&UIFSOFUʹՃ͑ͯιέοτ5$1*1ΛؚΉ ‣ -PPLVQͰड͚औͬͨύέοτʹରԠ͢ΔίωΫγϣϯ͕͋Δ͔Ͳ ͏͔ΛνΣοΫ ‣ ͋Ε/*$্Ͱͯ͢ॲཧ͞ΕΔ ͳ͚Εͦͷ࣌Ͱ04ʹ͞ΕΔ ‣
#ZQBTT04ͷιέοτͱ/*$ͷ ιέοτΛհ͢Δ ‣ -PPLVQ͕ͳ͚Ε/*$্ͷ5$1 ͰνΣοΫ͢ΔͨΊඇޮ
$POOFDUJPO%BUB4USVDUVSF ‣ ϋϯυΦϑͷͨΊͷίωΫγϣϯσʔλߏ ‣ ϋϯυΦϑͰɼ04ͱ/*$ͷ྆ํ͕5$1ॲཧ͢Δ͜ͱ͕͋ΔͨΊɼ5$1ॲཧ ʹඞཁͳσʔλʢ4PDLFUɼ4PDLFU#V⒎FSʣΛ྆ํͰͭ ‣ /*$ʹϙΠϯλΛஔ͘ͱ;FSP$PQZ*0͕ɹɹɹɹɹɹɹɹɹɹɹɹɹɹɹɹɹ ༰қʹͳΔͨΊɼ/*$ͷ4PDLFU#V⒎FSɹɹɹɹɹɹɹɹɹɹɹɹɹɹɹɹɹ ʹϝϞϦΞυϨεͱσʔλΛஔ͖ɺɹɹɹɹɹɹɹɹɹɹɹɹɹɹɹɹɹɹ
ɹ ɹ ɹ ඞཁͳͱ͖ʹϗετ͔ΒϑΣον͢Δ ‣ /*$ͷϝϞϦۭ͖༰ྔ͕খ͍͞ͱ͖/*$ɹɹɹɹɹɹɹɹɹɹɹɹɹɹɹɹɹ ͷ4PDLFU#V⒎FSʹԿஔ͔ͳ͍
&YQFSJNFOUBM.FUIPE ‣ ڥ ‣ ϗετɿ"UIMPO91 ()[ɼ,#-DBDIFɼ%3".(# ‣ ϓϩάϥϚϒϧ/*$ɿίΞɼ.)[ɼ.#43". ‣ ϋϯυΦϑΠϯλϑΣʔευϥΠόͰ࣮
‣ /*$্ͷ5$1*1࣮'SFF#4%ϕʔε ‣ ϗετͷ5$1*1࣮ͷมߋϋϯυΦϑΠϯλϑΣʔεͷݺͼग़͠෦ͷΈ ‣ /*$ͷίΞͷ͏ͪยํϓϩϑΝΠϧͷΈߦ͏ ‣ ࠷େͷ5$1εϧʔϓοτ.CT͕ͩɼ/*$ΛίΞ͑.CT
&YQFSJNFOUBM3FTVMUT $ZDMFT ‣ #ͷσʔλΛίωΫγϣϯΛมԽͤ͞ͳ͕Βଌఆ ‣ /*$ʹΦϑϩʔυ͢Δ͜ͱʹΑΓੑೳ্ʢ੨ʣ ‣ ϋϯυΦϑ࣌ʢʣΦϑϩʔυແ͠ʢʣΑΓੑೳ্ ‣ -Ωϟογϡώοτʹ͍ͭͯಉ༷
ͷ ‣ /*$ʹͯ͢Φϑϩʔυ͢Δͱ͖ ϝϞϦʹΒͳͯ͘ੑೳམͪΔͷͰʣ Cycles Per Packet
$PODMVTJPO ‣ 5$1ॲཧͰϝϞϦੑೳ͕ϘτϧωοΫ ‣ /*$ͷϝϞϦߴͳͨΊɼ/*$ͷΦϑϩʔσΟϯάʹΑΓɼϝϞϦ ʹର͢ΔෛՙΛ؇ ‣ /*$ͷϦιʔεݶΒΕ͍ͯΔͷͰͯ͢ͷίωΫγϣϯΛΦϑϩʔυ ͢Δͱ͔͑ͬͯύϑΥʔϚϯε͕Լ ‣
ίωΫγϣϯϋϯυΦϑʹΑΓɼ04͕ඞཁͳͱ͖ʹίωΫγϣϯΛ /*$ʹͨ͠ΓɼऔΓͨ͠Γ͢Δ͜ͱͰϦιʔεͷόϥϯεΛ੍ޚ ‣ ࣮ݧͷ݁ՌɼϓϩτλΠϓ൛$16αΠΫϧΛݮ ‣ γϛϡϨʔγϣϯ൛ݮ
Yet Another Handoff