A Critique of RDMA

By Patrick Geoffray, Ph.D.

August 18, 2006

Do you remember VIA, the Virtual Interface Architecture? I do. In 1998, according to its promoters — Intel, Compaq, and Microsoft — VIA was supposed to change the face of high-performance networking. VIA was a buzzword at the time; Venture Capital was flowing, and startups multiplying. Many HPC pundits were rallying behind this low-level programming interface, which promised scalable, low-overhead, high-throughput communication, initially for HPC and eventually for the data center. The hype was on and doom was spelled for the non-believers.

It turned out that VIA, based on RDMA (Remote Direct Memory Access, or Remote DMA), was not an improvement on existing APIs to support widely used application-software interfaces such as MPI and Sockets. After a while, VIA faded away, overtaken by other developments.

VIA was eventually reborn into the RDMA programming model that is the basis of various InfiniBand Verbs implementations, as well as DAPL (Direct Access Provider Library) and iWARP (Internet Wide Area RDMA Protocol). The pundits have returned, VCs are spending their money, and RDMA is touted as an ideal solution for the efficiency of high-performance networks.

However, the evidence I'll present here shows that the revamped RDMA model is more a problem than a solution. What's more, the objective that RDMA pretends to address of efficient user-level communication between computing nodes is already solved by the two-sided Send/Recv model in products such as Quadrics QsNet, Cray SeaStar (implementing Sandia Portals), Qlogic InfiniPath, and Myricom's Myrinet Express (MX).

Send/Recv versus RDMA

The difference between these two paradigms, Send/Receive (Send/Recv) and RDMA, resides essentially in the way to determine the destination buffer of a communication.

  • In the Send/Recv model, the source issues a Send that describes the location of the data to be sent, the destination posts a Receive that similarly indicates where the data is going to be written, and a matching capability is used to associate a posted Recv to an incoming Send. Each side has part of the information required for the completion of the communication; this is a “two-sided” interface.
  • In the RDMA model, both origin and destination buffers must be registered prior to any operations. This memory registration returns a handle that can be used in RDMA operations, Read or Write (a.k.a. Get or Put), to describe the origin or destination buffer. Only one side needs to have all of the information required for the completion of the communication; this is a “one-sided” interface.

In fact, most high-speed interconnects support both Send/Recv and RDMA (Put/Get) programming models, but the Send/Recv implementations of RDMA-based fabrics often lack a rich matching capability, which is a key element.

Matching

MPI is a two-sided interface with a large matching space: a MPI Recv is associated with a MPI Send according to several criteria such as Sender, Tag, and Context, with the first two possibly ignored (wildcard). Matching is not necessarily in order and, worse, a MPI Send can be posted before the matching MPI Recv is ready, creating the need to handle unexpected messages.

It is trivial to layer a Send/Recv interface on top of another Send/Recv interface as long as its matching space is wide enough. MPI requires 64 bits of matching information, and MX, Portals, and QsNet provide such a matching capability.

InfiniBand Verbs and other RDMA-based APIs do not support matching at all: they associate incoming Sends with outstanding Recvs in the order in which the Recvs were posted. If only one process can send to a specific receive queue, then you need to send messages in the same order you want to receive them. In the case of multiple senders (Shared Receive Queue), then you have no guarantee about order between senders. Worse, not having a Recv posted at the time an incoming message arrives is considered a fatal error in reliable mode. This lack of support for unexpected messages requires a specific flow-control mechanism to avoid this situation.

When implementing MPI on top of a RDMA programming model, you need to do the matching explicitly by sending a message containing the matching information to the receive side.

Overhead (zero-copy)

One of the key alleged benefits of RDMA is to avoid intermediate memory copies. How does that work with MPI? Not well, for two reasons: memory registration and synchronization.

Memory registration

Zero-copy requires both origin and destination buffers to be registered to allow safe access by the network device. Memory registration and deregistration are expensive operations. They involve a system call and an iteration over each virtual page in an OS-specific way. This registration overhead is actually similar to and sometimes exceeds the cost of a simple memory copy, going against the original objective of avoiding the overhead of intermediate copies.

So, is it possible to do zero-copy in MPI? Yes, but in very specific situations or with some unpleasant tricks:

  • Using lightweight kernels without virtual memory (Cray XT3, IBM Blue Gene).
  • Patching the various Linux kernels to implement safe long-term registration (Quadrics).
  • Hijacking the Malloc library to implement a non-portable and potentially unsafe user-level registration cache (MPICH-GM, MVAPICH, OpenMPI)
  • Optimize the registration code with specific NIC support so that the overhead is cheaper than the memory copy for large messages (MX).

In short, zero-copy, thus RDMA, is practical and efficient in the common case only when the memory registration can occur out of the critical path, which is incompatible with MPI semantics as well as Sockets semantics.

Synchronization

To do zero-copy, you also need to know where you are going to read/write the data on the remote side. Inasmuch as MPI is a two-sided interface, you need to do the matching prior to the RDMA operation, thus requiring an explicit synchronization of both the send and receive sides. The most dramatic side effect is that the MPI sender has to wait for the MPI receiver to be ready before completing the communication and reusing the send buffer. In some situations, the receive side can be very late and the matching synchronization propagates the delay to the send side. For this reason, it is not worth avoiding an intermediate copy on the send side for small/medium messages. Once the message is buffered, the MPI_Send can complete and the application can effectively move on, including reusing the send buffer.

So, it may be surprising for outsiders that most MPI implementations that claim zero-copy actually do a memory copy on both send and receive sides for small/medium messages, typically up to 16K or 32K messages, to avoid synchronization and memory-registration overhead. The RDMA model does not help in any way. The following graph, generated by my Myricom colleague Dr. Loïc Prylli, illustrates this trade-off in MX between using registration and deregistration, which provides the lowest overhead for large messages, versus memory copies, which provides the lowest overhead for short/medium messages.

Low Latency

Inasmuch as the RDMA operations are faster in native RDMA interconnects than the implementation of Send/Recv over RDMA, it is logical to try to use RDMA for small messages, even with intermediate copies. The protocol is straightforward: the send buffer is copied into a pre-registered internal buffer, RDMA-Written into a pre-registered internal buffer on the receive side, and then copied out to the corresponding MPI receive buffer. The problem lies in the one-sided characteristic of the RDMA Write: only one process can safely write into a remote buffer at a time. So the solution is to allocate a pre-registered internal buffer for each possible sender and look for messages in all of these separate buffers. Sounds nice? Not really. This strategy does not scale in time or in space. With only two processes, the receive side has only one buffer to poll. With N processes in the MPI job, the receive side has (N-1) locations to poll. Inasmuch as the polling time increases linearly with the number of processes, so does the latency. That's why most MPI implementations on RDMA-based fabrics use an RDMA method for small jobs, and fall back to a slower, more scalable, Send/Recv method for larger MPI jobs. For example, here are the short-message latencies reported in “InfiniBand Scalability in Open MPI“, Shipman, et al, IPDPS, May 2006, for the two modes of MVAPICH over InfiniBand.

What does it mean practically? Point-to-point micro-benchmarks (i.e., marketing benchmarks) are artificially better on small configurations, but performance is different in networks of useful scale. By contrast, all Send/Recv interconnects have consistent latencies, independent of the number of processes in the MPI job. Also, even the 4.19 us latency above compares unfavorably with the short-message latencies of QsNet, Portals/SeaStar, InfiniPath, or MX/Myri-10G.

Overlap of Communication and Computation (Progression)

Inasmuch as MPI is a two-sided interface requiring matching, it is important that this matching can occur independently of the MPI application, particularly for large messages. Without such asynchronous progress, it is not possible to overlap communication with computation, even when doing zero-copy.

The two following graphs are reproduced by permission from “Measuring MPI Send and Receive Overhead and Application Availability in High Performance Network Interfaces,” Doerfler, et al, EuroPVM/MPI, September 2006, a paper I highly recommend. Rather than showing overhead, the graphs show the send and receive overlap capability — the fraction of the processor time available for computation — with several recent HPC systems, listed here with their interconnects:

  • Red Storm: Cray SeaStar/Portals
  • Tbird: Cisco InfiniBand/MVAPICH
  • CBC-B: Qlogic InfiniPath
  • Odin: Myricom Myri-10G/MPICH-MX
  • Squall: Quadrics QsNetII

Asynchronous progress can be implemented by matching completely in the NIC (Quadrics QsNet and Cray SeaStar) or in the host with specific NIC support (Myricom MX). Some sort of progression could be done on top of a RDMA programming model, but would have a significant impact on short-message latency. Because matching is an integral part of the Send/Recv model, asynchronous implementation is easier in this context. Note that InfiniPath, even though it uses the Send/Recv model, always does a copy on the receive side for large messages, so it cannot overlap.

The graphs above were produced using the Sandia MPI Micro-Benchmark Suite (SMB).

Scalability

The biggest can of worms that the RDMA programming model brings to the table relates to scalability. RDMA is a connected model, i.e., one software entity called a Queue Pair (QP) in one process space is associated with one and only one such entity in a different process space. So, if you want to be able to communicate with all N processes in your MPI job, you need to use a different QP for each process except yourself. That's (N-1) QPs per process. Even with multiple processes per nodes, the number of QPs per RDMA-based NIC is large enough to accommodate all of them. However, this connected model directly affects the host memory footprint: each QP consumes host memory, so the total memory footprint grows linearly with the number of QPs. For example, it was not uncommon to see several gigabytes of memory required for large InfiniBand clusters. The following graph, which well illustrates this point, is from “InfiniBand Scalability in Open MPI“, Shipman et al, IPDPS, May 2006.

By contrast, the memory footprint of all Send/Recv-based interconnects is constant and independent of the number of processes in the MPI job.

Recently, the addition of the Shared Receive Queue (SRQ) concept, noted in the “Open MPI — SRQ” in the graph above, alleviates some of this scalability nightmare: a receive queue can be shared among several connections, thus consolidating the posted buffers used for the limited Send/Recv functionality. It is interesting to note that a shared receive queue is inherently a Send/Recv programming model feature, perhaps a sign that the native RDMA interconnects are slowly migrating to the more efficient Send/Recv programming model. However, SRQ does not address the overhead of initially establishing the connections, so a connection-on-demand mechanism is needed to hide the problem. This approach may work with MPI microbenchmarks, in which processes typically do not have to communicate with each peer in the MPI application, but this approach will not work well for many real-world MPI applications.

Conclusions

The RDMA programming model does not fit well with the two-sided semantic constraints of MPI or the Sockets interface. It's nothing new; it has been demonstrated before with previous incarnations of RDMA-based low-level communication layers such as VIA. However, the hype is on and the marketing message is wrongly promoting RDMA as the ultimate unifying programming model. Is the RDMA model completely useless? Of course not. It can be leveraged successfully for one-sided programming paradigms such as ARMCI (Aggregate Remote Memory Copy Interface) or UPC (Universal Parallel C), but these paradigms have yet to show better performance and ease of use than MPI.

History always repeats itself, so we can expect the current RDMA programming model to continue to struggle alongside the efficient Send/Recv model, and eventually to fade away.

The author would like to thank Charles Seitz and Loïc Prylli for their help in writing this article.

—–

Patrick Geoffray earned his Ph.D. at the University of Lyon, France, in 2001. His interests lie in high-performance computing. He is currently a Senior Software Architect at Myricom, in the Software Development Lab in Oak Ridge, TN. He is responsible for the design of the Myrinet Express (MX) message-passing system and its associated firmware implementations, and was earlier in charge of various middleware running on GM/Myrinet, including MPICH-GM and VIA-GM.

Subscribe to HPCwire's Weekly Update!

Be the most informed person in the room! Stay ahead of the tech trends with industy updates delivered to you every week!

Intel, Micro Debut Quad-Level Cell NAND Flash

May 22, 2018

Chipmakers continue to gear designs toward AI and other demanding cloud workloads that take advantage of datacenter flash storage capacity. To that end, memory specialist Micron Technology Inc. began shipping compact sol Read more…

By George Leopold

Japan Meteorological Agency Takes Delivery of Pair of Crays

May 21, 2018

Cray has supplied two identical Cray XC50 supercomputers to the Japan Meteorological Agency (JMA) in northwestern Tokyo. Boasting more than 18 petaflops combined peak computing capacity, the new systems will extend the a Read more…

By Tiffany Trader

ASC18: Final Results Revealed & Wrapped Up

May 17, 2018

It was an exciting week at ASC18 in Nanyang, China. The student teams braved extreme heat, extremely difficult applications, and extreme competition in order to cross the cluster competition finish line. The gala awards ceremony took place on Wednesday. The auditorium was packed with student teams, various dignitaries, the media, and other interested parties. So what happened? Read more…

By Dan Olds

HPE Extreme Performance Solutions

HPC and AI Convergence is Accelerating New Levels of Intelligence

Data analytics is the most valuable tool in the digital marketplace – so much so that organizations are employing high performance computing (HPC) capabilities to rapidly collect, share, and analyze endless streams of data. Read more…

IBM Accelerated Insights

Mastering the Big Data Challenge in Cognitive Healthcare

Patrick Chain, genomics researcher at Los Alamos National Laboratory, posed a question in a recent blog: What if a nurse could swipe a patient’s saliva and run a quick genetic test to determine if the patient’s sore throat was caused by a cold virus or a bacterial infection? Read more…

ASC18: Tough Applications & Tough Luck

May 17, 2018

The applications at the ASC18 Student Cluster Competition were tough. Tougher than the $3.99 steak special at your local greasy spoon restaurant. The apps are so tough that even Chuck Norris backs away from them slowly. Read more…

By Dan Olds

Japan Meteorological Agency Takes Delivery of Pair of Crays

May 21, 2018

Cray has supplied two identical Cray XC50 supercomputers to the Japan Meteorological Agency (JMA) in northwestern Tokyo. Boasting more than 18 petaflops combine Read more…

By Tiffany Trader

ASC18: Final Results Revealed & Wrapped Up

May 17, 2018

It was an exciting week at ASC18 in Nanyang, China. The student teams braved extreme heat, extremely difficult applications, and extreme competition in order to cross the cluster competition finish line. The gala awards ceremony took place on Wednesday. The auditorium was packed with student teams, various dignitaries, the media, and other interested parties. So what happened? Read more…

By Dan Olds

Spring Meetings Underscore Quantum Computing’s Rise

May 17, 2018

The month of April 2018 saw four very important and interesting meetings to discuss the state of quantum computing technologies, their potential impacts, and th Read more…

By Alex R. Larzelere

Quantum Network Hub Opens in Japan

May 17, 2018

Following on the launch of its Q Commercial quantum network last December with 12 industrial and academic partners, the official Japanese hub at Keio University is now open to facilitate the exploration of quantum applications important to science and business. The news comes a week after IBM announced that North Carolina State University was the first U.S. university to join its Q Network. Read more…

By Tiffany Trader

Democratizing HPC: OSC Releases Version 1.3 of OnDemand

May 16, 2018

Making HPC resources readily available and easier to use for scientists who may have less HPC expertise is an ongoing challenge. Open OnDemand is a project by t Read more…

By John Russell

PRACE 2017 Annual Report: Exascale Aspirations; Industry Collaboration; HPC Training

May 15, 2018

The Partnership for Advanced Computing in Europe (PRACE) today released its annual report showcasing 2017 activities and providing a glimpse into thinking about Read more…

By John Russell

US Forms AI Brain Trust

May 11, 2018

Amid calls for a U.S. strategy for promoting AI development, the Trump administration is forming a senior-level panel to help coordinate government and industry research efforts. The Select Committee on Artificial Intelligence was announced Thursday (May 10) during a White House summit organized by the Office of Science and Technology Policy (OSTP). Read more…

By George Leopold

Emerging Advanced Scale Tech Trends Focus of Annual Tabor Conference

May 9, 2018

At Tabor Communications' annual Advanced Scale Forum (ASF) held this week in Austin, the focus was on enterprise adoption of HPC-class technologies and high performance data analytics (HPDA). It’s a confab that brings together end users (CIOs, IT planners, department heads) and vendors and encourages... Read more…

By the Editorial Team

MLPerf – Will New Machine Learning Benchmark Help Propel AI Forward?

May 2, 2018

Let the AI benchmarking wars begin. Today, a diverse group from academia and industry – Google, Baidu, Intel, AMD, Harvard, and Stanford among them – releas Read more…

By John Russell

How the Cloud Is Falling Short for HPC

March 15, 2018

The last couple of years have seen cloud computing gradually build some legitimacy within the HPC world, but still the HPC industry lies far behind enterprise I Read more…

By Chris Downing

Russian Nuclear Engineers Caught Cryptomining on Lab Supercomputer

February 12, 2018

Nuclear scientists working at the All-Russian Research Institute of Experimental Physics (RFNC-VNIIEF) have been arrested for using lab supercomputing resources to mine crypto-currency, according to a report in Russia’s Interfax News Agency. Read more…

By Tiffany Trader

Nvidia Responds to Google TPU Benchmarking

April 10, 2017

Nvidia highlights strengths of its newest GPU silicon in response to Google's report on the performance and energy advantages of its custom tensor processor. Read more…

By Tiffany Trader

Deep Learning at 15 PFlops Enables Training for Extreme Weather Identification at Scale

March 19, 2018

Petaflop per second deep learning training performance on the NERSC (National Energy Research Scientific Computing Center) Cori supercomputer has given climate Read more…

By Rob Farber

AI Cloud Competition Heats Up: Google’s TPUs, Amazon Building AI Chip

February 12, 2018

Competition in the white hot AI (and public cloud) market pits Google against Amazon this week, with Google offering AI hardware on its cloud platform intended Read more…

By Doug Black

US Plans $1.8 Billion Spend on DOE Exascale Supercomputing

April 11, 2018

On Monday, the United States Department of Energy announced its intention to procure up to three exascale supercomputers at a cost of up to $1.8 billion with th Read more…

By Tiffany Trader

Lenovo Unveils Warm Water Cooled ThinkSystem SD650 in Rampup to LRZ Install

February 22, 2018

This week Lenovo took the wraps off the ThinkSystem SD650 high-density server with third-generation direct water cooling technology developed in tandem with par Read more…

By Tiffany Trader

Leading Solution Providers

HPC and AI – Two Communities Same Future

January 25, 2018

According to Al Gara (Intel Fellow, Data Center Group), high performance computing and artificial intelligence will increasingly intertwine as we transition to Read more…

By Rob Farber

Researchers Measure Impact of ‘Meltdown’ and ‘Spectre’ Patches on HPC Workloads

January 17, 2018

Computer scientists from the Center for Computational Research, State University of New York (SUNY), University at Buffalo have examined the effect of Meltdown Read more…

By Tiffany Trader

Google Chases Quantum Supremacy with 72-Qubit Processor

March 7, 2018

Google pulled ahead of the pack this week in the race toward "quantum supremacy," with the introduction of a new 72-qubit quantum processor called Bristlecone. Read more…

By Tiffany Trader

HPE Wins $57 Million DoD Supercomputing Contract

February 20, 2018

Hewlett Packard Enterprise (HPE) today revealed details of its massive $57 million HPC contract with the U.S. Department of Defense (DoD). The deal calls for HP Read more…

By Tiffany Trader

CFO Steps down in Executive Shuffle at Supermicro

January 31, 2018

Supermicro yesterday announced senior management shuffling including prominent departures, the completion of an audit linked to its delayed Nasdaq filings, and Read more…

By John Russell

Deep Learning Portends ‘Sea Change’ for Oil and Gas Sector

February 1, 2018

The billowing compute and data demands that spurred the oil and gas industry to be the largest commercial users of high-performance computing are now propelling Read more…

By Tiffany Trader

Nvidia Ups Hardware Game with 16-GPU DGX-2 Server and 18-Port NVSwitch

March 27, 2018

Nvidia unveiled a raft of new products from its annual technology conference in San Jose today, and despite not offering up a new chip architecture, there were still a few surprises in store for HPC hardware aficionados. Read more…

By Tiffany Trader

Hennessy & Patterson: A New Golden Age for Computer Architecture

April 17, 2018

On Monday June 4, 2018, 2017 A.M. Turing Award Winners John L. Hennessy and David A. Patterson will deliver the Turing Lecture at the 45th International Sympo Read more…

By Staff

  • arrow
  • Click Here for More Headlines
  • arrow
Share This