Visit additional Tabor Communication Publications
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 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.
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.
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.
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:
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.
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.
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:
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).
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.
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.
Jun 18, 2013 |
The world's largest supercomputers, like Tianhe-2, are great at traditional, compute-intensive HPC workloads, such as simulating atomic decay or modeling tornados. But data-intensive applications--such as mining big data sets for connections--is a different sort of workload, and runs best on a different sort of computer.
Jun 18, 2013 |
Researchers are finding innovative uses for Gordon, the 285 teraflop supercomputer housed at the San Diego Supercomputer Center (SDSC) that has a unique Flash-based storage system. Since going online, researchers have put the incredibly fast I/O to use on a wide variety of workloads, ranging from chemistry to political science.
Jun 17, 2013 |
The advent of low-power mobile processors and cloud delivery models is changing the economics of computing. But just as an economy car is good at different things than a full size truck, an HPC workload still has certain computing demands that neither the fastest smartphone nor the most elastic cloud cluster can fulfill.
Jun 14, 2013 |
For all the progress we've made in IT over the last 50 years, there's one area of life that has steadfastly eluded the grasp of computers: understanding human language. Now, researchers at the Texas Advanced Computing Center (TACC) are utilizing a Hadoop cluster on its Longhorn supercomputer to move the state of the art of language processing a little bit further.
Jun 13, 2013 |
Titan, the Cray XK7 at the Oak Ridge National Lab that debuted last fall as the fastest supercomputer in the world with 17.59 petaflops of sustained computing power, will rely on its previous LINPACK test for the upcoming edition of the Top 500 list.
05/10/2013 | Cleversafe, Cray, DDN, NetApp, & Panasas | From Wall Street to Hollywood, drug discovery to homeland security, companies and organizations of all sizes and stripes are coming face to face with the challenges – and opportunities – afforded by Big Data. Before anyone can utilize these extraordinary data repositories, however, they must first harness and manage their data stores, and do so utilizing technologies that underscore affordability, security, and scalability.
04/15/2013 | Bull | “50% of HPC users say their largest jobs scale to 120 cores or less.” How about yours? Are your codes ready to take advantage of today’s and tomorrow’s ultra-parallel HPC systems? Download this White Paper by Analysts Intersect360 Research to see what Bull and Intel’s Center for Excellence in Parallel Programming can do for your codes.
Join HPCwire Editor Nicole Hemsoth and Dr. David Bader from Georgia Tech as they take center stage on opening night at Atlanta's first Big Data Kick Off Week, filmed in front of a live audience. Nicole and David look at the evolution of HPC, today's big data challenges, discuss real world solutions, and reveal their predictions. Exactly what does the future holds for HPC?
Join our webinar to learn how IT managers can migrate to a more resilient, flexible and scalable solution that grows with the data center. Mellanox VMS is future-proof, efficient and brings significant CAPEX and OPEX savings. The VMS is available today.