引用: "4th Quarter Earnings Presentation" @ https://www.intc.com/ https://d1io3yog0oux5.cloudfront.net/_2cead9b6413a1a91de449423742eea20/intel/db/887/8894/earnings_presentation/Q4%272022+Earnings+Deck_Final+PDF.pdf Intel オフィシャルコメント: https://download.intel.com/newsroom/2023/corporate/q422-investor-call-remarks.pdf Earning Comments from CEO Pat Gelsinger and CFO Dave Zinsner We are making tough decisions to right-size the organization, and we further sharpened our business focus within our BUs (business units) by rationalizing product roadmaps and investments. NEX continues to do well and is a core part of our strategic transformation, but we will end future investment on our network switching product line, while still fully supporting existing products and customers. Since my return, we have exited seven businesses, providing in excess of $1.5 billion in savings. We are also well underway to integrating AXG (Accelerated Computing Systems and Graphics) into CCG (Client Computing Group) and DCAI (Data Center and AI Group), respectively, to drive a more effective go-to-market capability, accelerating the scale of these businesses while further reducing costs.
to Operate PSG as Standalone Business https://www.intel.com/content/www/us/en/newsroom/news/intel-oct-2023-news.html EE Times Japan > FPGA > Intelが旧Altera部門を分離、数年以内にIPOへ ... https://eetimes.itmedia.co.jp/ee/articles/2310/04/news142.html
d'etre, missing features and works done Andy Fingerhut, Intel The Power of Fully-Specified Data Planes Rob Sherwood, NEX Cloud Networking Group, CTO 2023 P4 Workshop @SanJose ~ Keynote 1 & 2
the elephant https://sketchplanations.com/the-blind-and-the-elephant データプレーン(完全)定義言語 fully-specified data plane language データプレーン 抽象化レイヤ (Fixed Function ASIC への適用) テストの自動生成 検証可能な形式言語 Formal Language データプレーン プログラミング言語 data plane programming language 高速パケット処理デバイス のプログラミング データプレーン APIを自動生成 (P4Runtime) ターゲット非依存の プログラミング (ASIC, FPGA) P4の存在理由 ~ P4’s raison d’être ~ 群盲象を評す
P4から探る ~ サーバーサイド高速パケット処理の現状と展望|海老澤健太郎@トヨタ自動車 SNA ... Smart Network Appliance P4を用いた独自実装 2022 P4 Workshop "The Journey towards Predictable Network in Alibaba Cloud" Dennis Cai, Head of Network Infrastructure https://opennetworking.org/2022-p4-workshop-gated/ 2022 P4 Workshop
Vladimir Gurevich ☺ https://www.linkedin.com/in/vladimir-gurevich-63b887/ https://support.p4ica.com/hc/en-us/articles/18767647670299-Intel-Connectivity-Academy-ICA-XFG-by-P4ICA Intel® Connectivity Academy (ICA-XFG) by P4ICA Intel Connectivity Academy (ICA-XFG) is a set of instructor-led courses that cover all aspects of data and control plane development using Intel Tofino ASIC Family, P4_16 language and Intel P4 Studio SDE. These courses can be delivered online or in-person together with the hands-on labs. Starting September 1, 2023 these courses are delivered by an independent company, P4ICA (P4ICA, LTD) under the special license from Intel. This site contains all the materials for the classes, including the presentation materials, labs and tools. If you've ever attended Barefoot Academy or Intel Connectivity Academy, you should have an account on this site that allows you to access the materials for the courses you attended. In addition to that, you can find new course prospectuses and the upcoming class schedule. For the latter we also recommend that you follow us on Eventbrite. Intel® Connectivity Academy (ICA-XFG) by P4ICA
East-West Traffic 中心 サーバーに近い場所での ネットワーク機能提供が必要に アプリケーション構成の変化により、ネットワーク機能の場所も変化 Firewall Load Balancer Server Server Server Server Internet Server Server Server Internet micro-service
• They found that the diversity of workloads would benefit from flexible architectures, and identify a “datacenter tax” in the lower layers of the software stack that comprises nearly 30% of cycles across jobs and that are prime candidates for hardware specialization and acceleration. • 2020 Facebook: Accelerometer: Understanding Acceleration Opportunities for Data Center Overheads at Hyperscale • https://dl.acm.org/doi/10.1145/3373376.3378450 • Another study by Facebook published in 2020 found that microservices spend as few as 18% of CPU cycles executing core application logic. The remaining cycles were spent in common operations not core to the application logic, including I/O processing, logging, and compression. Facebook believed that accelerating standard building blocks can significantly improve data center performance and they built a model to project possible hardware speedup in microservices. アプリケーションが占めるCPU利用率の低下 ~ P4から探る ~ サーバーサイド高速パケット処理の現状と展望|海老澤健太郎@トヨタ自動車
done", Andy Fingerhut, Intel, 2023 P4 Workshop @SanJose SmartNIC で必要な P4 機能(スイッチと比較) ~ P4から探る ~ サーバーサイド高速パケット処理の現状と展望|海老澤健太郎@トヨタ自動車 Now part of the PNA specification
per second to large tables. • コントロールプレーン非依存なテーブル更新 • Table lookup miss 時のエントリ追加 • Timeout したエントリの削除 • ステートフルなプロトコルへの対応 • Data-plane-writable action data • e.g. maintain expected TCP sequence numbers independently for each table entry, in TCP connection tracking. SmartNIC で必要な P4 機能(スイッチと比較) ~ P4から探る ~ サーバーサイド高速パケット処理の現状と展望|海老澤健太郎@トヨタ自動車
from Switch to DPU https://opennetworking.org/wp-content/uploads/2022/05/Alan-Lo-and-Milind-Chabbi-Final-Slide-Deck-1.pdf Shared-memory for local controller Bulk table updates for remote controller
infrap4d Infrap4d integrates Stratum, the Kernel Monitor (krnlmon), Switch Abstraction Interface (SAI), Table Driven Interface (TDI), and a P4 target driver into a separate process (daemon). The IPDK Networking Recipe (originally P4-OVS Split Architecture) modularizes P4-OVS and reduces coupling between its components, making the code easier to maintain and more suitable for upstreaming. It moves the P4-specific components of the integrated architecture of P4-OVS to a separate process called infrap4d. https://github.com/ipdk-io/networking-recipe