OpenMPI OpenIB

Open MPI: 

Open MPI Overview and Current Status Tim Woodall - LANL Galen Shipman - LANL/UNM

Overview: 

Overview Point-to-Point Architecture OpenIB Implementation Results Future Work

Point-to-Point Architecture: 

Point-to-Point Architecture Component Architecture: “Plug-ins” for different capabilities (e.g. different networks) Tunable run-time parameters Three component frameworks: Point-to-point messaging layer (PML) implements MPI semantics Byte Transfer Layer (BTL) abstracts network interfaces Memory Pool (mpool) provides for memory management/registration

PML Framework: 

PML Framework Single PML manages multiple BTL modules Maintains set of BTLs on a per-peer basis Message fragmentation and scheduling Implements MPI semantics Synchronous / buffered / ready / normal sends Persistent requests / Request completion Eager/Rendezvous protocol Eager send of short messages Configurable threshold (short vs. long) Multiple long protocols

PML Protocols: 

PML Protocols Send / receive pipeline to / from pre-registered buffers (non-contiguous data) MPI_Alloc_mem support Red/black tree of memory registrations BTL associated with registration is used by scheduler Xfer of contiguous data with 1 RDMA (after match) “Leave pinned” run-time parameter Registration on first-use MRU cache (configurable size) of registrations Bandwidth equivalent to pre-registered buffers (MPI_Alloc_mem)

PML Protocols (Continued): 

PML Protocols (Continued) Dynamic memory registration/deregistration Fragment message and build pipeline of RDMA requests Overlap [de-]registration with RDMA Bandwidth 97% of pre-registered memory at large message sizes (8Mbytes) Performance impacted by bus type/bandwidth

BTL Framework: 

BTL Framework MPI agnostic Provides simple API to upper layers Tagged send/receive primitives One-sided put/get operations Access to data type engine for zero copy data transfer BTL modules natively support commodity networks: Current (self, shared memory, myrinet GM/MX, Infiniband mvapi/OpenIB, Portals, TCP) Planned (LAPI, Quadrics Elan4)

OpenIB BTL: 

OpenIB BTL BTL module initialization Resources allocation Connection management Small message Xfer Large message Xfer OpenIB Issues Future Work

BTL module initialization: 

BTL module initialization A separate BTL module is initialized for each port on each HCA The PML schedules across these BTL modules just as any other interconnect When multiple BTL modules exist peers establish QP connections by matching subnets

Resource Allocation: 

Resource Allocation

SRQ Scalability : 

SRQ Scalability K- multiplier based on number of nodes

Connection management: 

Connection management Addressing information is exchanged dynamically via an OOB channel This greatly improves scalability but at the cost of increased first message latency Connections are established with peers in the same subnet (local subnet routing only)

Small Message Xfer: 

Small Message Xfer Maintain list of pre-registered fragments for send and recv List grows dynamically in chunks (more efficient to register) Small messages are copied to/from pre-registered buffers Recv descriptors are posted as needed based on min/max thresholds

Small Message Performance: 

Small Message Performance * Send/Recv based protocol

Large Message Xfer: 

Large Message Xfer RDMA Write and RDMA Read are both supported RDMA Read provides better performance than RDMA Write - control messages are reduced RDMA pipeline protocol performance highly dependent on I/O Bus performance

Results: 

Results OpenMPI/OpenIB - All

Results: 

Results OpenMPI/OpenIB - All - Log

Results: 

Results OpenMPI/OpenIB - Eager limit

Results: 

Results Combined Results

Results: 

Results Combined Results - Log

OpenIB Opportunities: 

OpenIB Opportunities User level notification of VM activity Caching of memory registrations can be dangerous Need the ability to detect VM changes that effect memory registrations (such as sbrk and munmap) Reliable Multicast for collectives SRQ performance, 2/10 usec penalty, but who’s counting?

Future Work: 

Future Work Small message RDMA (using working set of peers) - optional Dynamic connection management using Unreliable Datagrams Dynamic connection teardown - optional

Source Code Access: 

Source Code Access Subversion repository Download client from: http://subversion.tigris.org/ v1.2.1 or later Check out with: svn co http://svn.open-mpi.org/svn/ompi/trunk ompi Anonymous, read-only access

Questions?: 

Questions? Tim Woodall Email: twoodall@lanl.gov Phone: 505-665-5224 Galen Shipman Email: gshipman@lanl.gov

Hardware Specs: 

Hardware Specs Dual Intel Xeon 3.2 GHz 1024 KB Cache 2 Gbytes memory Bus: Intel Corp. E7525/E7520/E7320 PCI Express Mellanox Technologies MT25208 InfiniHost III Ex 288 Port Voltaire switch