Hard real-time processing Created late 1990s Deterministic la - - PowerPoint PPT Presentation

hard real time processing
SMART_READER_LITE
LIVE PREVIEW

Hard real-time processing Created late 1990s Deterministic la - - PowerPoint PPT Presentation

In Integrated Device Technology Partner sin ince 2015 HQ: : San Jose, Cali lifornia ~1500 employees Hard real-time processing Created late 1990s Deterministic la latency ~ 100 ns Combines strengths of PCIe Ie and Ethernet Supports


slide-1
SLIDE 1
slide-2
SLIDE 2

In Integrated Device Technology Partner sin ince 2015 HQ: : San Jose, Cali lifornia ~1500 employees Hard real-time processing

slide-3
SLIDE 3

Created late 1990’s Deterministic la latency ~ 100 ns Combines strengths of PCIe Ie and Ethernet Supports heterogeneous systems Open standard – rapidio.org

slide-4
SLIDE 4

Seminars Hands-on la labs 4 4 nodes Each wit ith a RapidIO NIC IC 1 1 RapidIO Swit itch

slide-5
SLIDE 5

Use cases Focus on dif ifferent areas Evaluate suitability

slide-6
SLIDE 6
slide-7
SLIDE 7

Explore lo low le level APIs Is Messaging API I – protocol communication DMA API I – data transfer Unic icast - Mult lticast Hadoop deployed on 4-node la lab clu luster Usin ing RIO IONET driver – emula lating TCP/IP over RapidIO Nodes communicating solely using RapidIO

slide-8
SLIDE 8

RO ROOT hcli lient (R (ROOT macro) TUnixSystem TRa RapidIOSystem FS FS and

  • th

ther ser ervices TCP/IP Ra RapidIO hserv (R (ROOT macro) RO ROOT TUnixSystem TRa RapidIOSystem TCP/IP FS FS and

  • th

ther ser ervices Ra RapidIO TUnixSystem TRa RapidIOSystem DMA Messaging Ra RapidIO

ROOT over RIO IONET Native port of ROOT to RapidIO Override core cla lass in in ROOT Add RapidIO capabilities Reuse base cla lass functionality for file ilesystem and other services PROOF – dis istributed ROOT

slide-9
SLIDE 9

Vis isit ID IDT Ottawa offices Project review In In-depth training Target clu luster Twik iki Git it repo JIR IRA - Kanban

slide-10
SLIDE 10

Fin inalize ROOT proof of concept PROOF port to RapidIO RapidIO wit ith enterprise analy lytics framework (H (Hadoop/Spark) Benchmark appli lications Set up 19 in inch rack servers and top of rack swit itch

slide-11
SLIDE 11