TotalView Debugger Offers Instant Replay

By Michael Feldman

September 8, 2008

Although they don’t enjoy the celebrity of compilers, debuggers are one of the most critical tools that affect programmer productivity. There’s a good deal of anecdotal evidence that debugging code, not writing it, is the most time-consuming phase of software development. In “The Elements of Programming Style,” Brian Kernighan wrote: “Everyone knows that debugging is twice as hard as writing a program in the first place. So if you’re as clever as you can be when you write it, how will you ever debug it?” And unlike code development, there is no formal methodology to debugging — it’s essentially a trial and error activity. Developers run blindly through the program, setting breakpoints and watching for bad things to happen.

The problem, as Kernighan and Rob Pike describe in “The Practice of Programming,” is that “[d]ebugging involves backwards reasoning, like solving murder mysteries. Something impossible occurred, and the only solid information is that it really did occur. So we must think backwards from the result to discover the reasons.”

TotalView Technologies’ ReplayEngine is aimed squarely at this cognitive challenge. It allows the developer to execute the application in reverse. So instead of trying to sneak up on the bad behavior, it allows the program to fail, at which point the developer can take control and start backing up through the execution, examining the program state to see where the code went wrong.

“We’re very happy to be bring this to market,” said TotalView Technologies President and CEO Rich Collier. “We believe it’s going to be breakthrough technology not only in terms of how to do debugging, but also possibly how you go about doing development.”

They currently have a number of beta customers for ReplayEngine and, according to Collier, it’s already showing its worth in the field. The early users have reported finding a lot of bugs in a very short period of time. The common theme is that ReplayEngine eliminates the most repetitive parts of debugging — restarting the application, examining the breakpoints, waiting for the error to occur, and then setting new breakpoints and repeating the process.

“This is particularly useful in scalable, multicore environments in which errors can be hard to track down or reproduce,” said Addison Snell, vice president and general manager of Tabor Research. “TotalView is helping to enable more people to effectively develop scalable applications. Our research shows this is a prime area of concern for the industry.”

One of the most difficult scenarios for debugging is a program crash, where the developer is left with an incoherent program state. Since the crash has stomped on memory, the stack trace is destroyed or at least can’t be trusted. In this case, ReplayEngine can be used to reverse the crash and help the developer figure out what led up to it. A replay can also be used to track down race conditions, bad pointers, array bounds overwrites, memory leaks, and non-robust I/O handling.

The way ReplayEngine works is by recording program context at critical points during execution. The libraries are instrumented to capture all function and system calls, recording the program state at each point. When the user chooses to back up to a specific location in the code, the recorded execution is rerun in a simulated replay environment, starting at the closest point to where the user wants to return, and executing forward to that instruction. At that point, the user can continue to back up or go forward again.

Since ReplayEngine controls I/O, it is able to rerun these external events as they originally occurred. For example, when the program makes a read() call during program execution, the input actually happens and ReplayEngine captures the data. During replay mode, the program may be asked to run through that execution again one or more times. In those cases, the read() call will be intercepted by ReplayEngine and the recorded input data will be used. This ensures deterministic behavior during backward debugging.

ReplayEngine buttons allow you step backward over or into functions. You can also back up out of the current function to the point before the call. Another button lets you move back to a specified source line. The normal breakpoints, watchpoints, and some conditional breakpoints can be used when running forward in replay mode. A live button allows you to jump back into normal debugging mode.

The initial ReplayEngine add-on works on standard Linux distributions on x86 (32-bit and 64-bit) targets, so if you’re looking for BlueGene or Itanium support, you’re out of luck. MPI is supported, specifically MPICH 2.0. According to Collier, at this point the engine can handle an application spread across tens of MPI processes, so if your software is using thousands of compute nodes, the application will have to be scaled down to use the replay feature. Future versions may support larger jobs and expand target coverage, but the initial version is aimed at the heart of TotalView’s market.

But backing up through an MPI application of any size is quite an achievement. Since each MPI process is recorded by ReplayEngine, the developer is able to examine the execution history of each one. For example, a crash on process 20 due to an invalid numeric value (say a divide-by-zero operation) might be traced to a variable that got its value from an incoming message via process 8. The developer could then switch the debug focus to process 8, running on a completely different node. The developer could then locate the point in the execution history where that message was sent. Then the developer could work their way backwards in the execution history of this new process to discover how the invalid data came to be.

The ReplayEngine add-on starts at $1,295 for a single token. Buying in quantity reduces the price significantly: $545/token for 10 tokens and $150/token for 100. There’s also TotalView’s 75 percent off discount for educational use, an area where the company expects a lot of interest. “It is ideally suited for a variety of instructional purposes in the classroom,” said Paul Gray, Supercomputing Conference Education Program Chair in a prepared statement for the press release. “From a software engineering perspective, being able to record and replay program execution will go a long way in improving the effectiveness and efficiency of the troubleshooting phase of software development, which is a crucial component of high-performance computing education.”

There are a handful of other efforts out there aimed at replay debugging, including an open source version of gdb, currently under development. A debugger called UNDOdb, which is written on top of gdb, is available today, but only supports 32-bit x86 and in not thread-aware, much less MPI-aware. Some replay engines are available for Java, but that’s outside of TotalView’s purview. To date, the difficulty of implementing general-purpose replay debugging has prevented most tool makers from bringing this technology to market. “It took a debugger company to come up with something like this,” said Collier. “We live it and breathe it every day.”

But it was no snap for the TotalView engineers either. Not only did they have to implement the session record and simulation replay features, but it all had to be integrated seamlessly into their live debugger foundation. TotalView has been working on the new technology for at least a few years, demonstrating a prototype at SC07 in Reno last November. Collier admitted some engineers expressed doubts it could even be done. He said it actually helped that they were able to use some of the initial versions of ReplayEngine to debug the technology itself.

“It took baseball 108 years to come up with the replay,” he laughed, “so I guess we’re not so bad.”

Subscribe to HPCwire's Weekly Update!

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

MLPerf Inference 4.0 Results Showcase GenAI; Nvidia Still Dominates

March 28, 2024

There were no startling surprises in the latest MLPerf Inference benchmark (4.0) results released yesterday. Two new workloads — Llama 2 and Stable Diffusion XL — were added to the benchmark suite as MLPerf continues Read more…

Q&A with Nvidia’s Chief of DGX Systems on the DGX-GB200 Rack-scale System

March 27, 2024

Pictures of Nvidia's new flagship mega-server, the DGX GB200, on the GTC show floor got favorable reactions on social media for the sheer amount of computing power it brings to artificial intelligence.  Nvidia's DGX Read more…

Call for Participation in Workshop on Potential NSF CISE Quantum Initiative

March 26, 2024

Editor’s Note: Next month there will be a workshop to discuss what a quantum initiative led by NSF’s Computer, Information Science and Engineering (CISE) directorate could entail. The details are posted below in a Ca Read more…

Waseda U. Researchers Reports New Quantum Algorithm for Speeding Optimization

March 25, 2024

Optimization problems cover a wide range of applications and are often cited as good candidates for quantum computing. However, the execution time for constrained combinatorial optimization applications on quantum device Read more…

NVLink: Faster Interconnects and Switches to Help Relieve Data Bottlenecks

March 25, 2024

Nvidia’s new Blackwell architecture may have stolen the show this week at the GPU Technology Conference in San Jose, California. But an emerging bottleneck at the network layer threatens to make bigger and brawnier pro Read more…

Who is David Blackwell?

March 22, 2024

During GTC24, co-founder and president of NVIDIA Jensen Huang unveiled the Blackwell GPU. This GPU itself is heavily optimized for AI work, boasting 192GB of HBM3E memory as well as the the ability to train 1 trillion pa Read more…

MLPerf Inference 4.0 Results Showcase GenAI; Nvidia Still Dominates

March 28, 2024

There were no startling surprises in the latest MLPerf Inference benchmark (4.0) results released yesterday. Two new workloads — Llama 2 and Stable Diffusion Read more…

Q&A with Nvidia’s Chief of DGX Systems on the DGX-GB200 Rack-scale System

March 27, 2024

Pictures of Nvidia's new flagship mega-server, the DGX GB200, on the GTC show floor got favorable reactions on social media for the sheer amount of computing po Read more…

NVLink: Faster Interconnects and Switches to Help Relieve Data Bottlenecks

March 25, 2024

Nvidia’s new Blackwell architecture may have stolen the show this week at the GPU Technology Conference in San Jose, California. But an emerging bottleneck at Read more…

Who is David Blackwell?

March 22, 2024

During GTC24, co-founder and president of NVIDIA Jensen Huang unveiled the Blackwell GPU. This GPU itself is heavily optimized for AI work, boasting 192GB of HB Read more…

Nvidia Looks to Accelerate GenAI Adoption with NIM

March 19, 2024

Today at the GPU Technology Conference, Nvidia launched a new offering aimed at helping customers quickly deploy their generative AI applications in a secure, s Read more…

The Generative AI Future Is Now, Nvidia’s Huang Says

March 19, 2024

We are in the early days of a transformative shift in how business gets done thanks to the advent of generative AI, according to Nvidia CEO and cofounder Jensen Read more…

Nvidia’s New Blackwell GPU Can Train AI Models with Trillions of Parameters

March 18, 2024

Nvidia's latest and fastest GPU, codenamed Blackwell, is here and will underpin the company's AI plans this year. The chip offers performance improvements from Read more…

Nvidia Showcases Quantum Cloud, Expanding Quantum Portfolio at GTC24

March 18, 2024

Nvidia’s barrage of quantum news at GTC24 this week includes new products, signature collaborations, and a new Nvidia Quantum Cloud for quantum developers. Wh Read more…

Alibaba Shuts Down its Quantum Computing Effort

November 30, 2023

In case you missed it, China’s e-commerce giant Alibaba has shut down its quantum computing research effort. It’s not entirely clear what drove the change. Read more…

Nvidia H100: Are 550,000 GPUs Enough for This Year?

August 17, 2023

The GPU Squeeze continues to place a premium on Nvidia H100 GPUs. In a recent Financial Times article, Nvidia reports that it expects to ship 550,000 of its lat Read more…

Shutterstock 1285747942

AMD’s Horsepower-packed MI300X GPU Beats Nvidia’s Upcoming H200

December 7, 2023

AMD and Nvidia are locked in an AI performance battle – much like the gaming GPU performance clash the companies have waged for decades. AMD has claimed it Read more…

DoD Takes a Long View of Quantum Computing

December 19, 2023

Given the large sums tied to expensive weapon systems – think $100-million-plus per F-35 fighter – it’s easy to forget the U.S. Department of Defense is a Read more…

Synopsys Eats Ansys: Does HPC Get Indigestion?

February 8, 2024

Recently, it was announced that Synopsys is buying HPC tool developer Ansys. Started in Pittsburgh, Pa., in 1970 as Swanson Analysis Systems, Inc. (SASI) by John Swanson (and eventually renamed), Ansys serves the CAE (Computer Aided Engineering)/multiphysics engineering simulation market. Read more…

Choosing the Right GPU for LLM Inference and Training

December 11, 2023

Accelerating the training and inference processes of deep learning models is crucial for unleashing their true potential and NVIDIA GPUs have emerged as a game- Read more…

Intel’s Server and PC Chip Development Will Blur After 2025

January 15, 2024

Intel's dealing with much more than chip rivals breathing down its neck; it is simultaneously integrating a bevy of new technologies such as chiplets, artificia Read more…

Baidu Exits Quantum, Closely Following Alibaba’s Earlier Move

January 5, 2024

Reuters reported this week that Baidu, China’s giant e-commerce and services provider, is exiting the quantum computing development arena. Reuters reported � Read more…

Leading Solution Providers

Contributors

Comparing NVIDIA A100 and NVIDIA L40S: Which GPU is Ideal for AI and Graphics-Intensive Workloads?

October 30, 2023

With long lead times for the NVIDIA H100 and A100 GPUs, many organizations are looking at the new NVIDIA L40S GPU, which it’s a new GPU optimized for AI and g Read more…

Shutterstock 1179408610

Google Addresses the Mysteries of Its Hypercomputer 

December 28, 2023

When Google launched its Hypercomputer earlier this month (December 2023), the first reaction was, "Say what?" It turns out that the Hypercomputer is Google's t Read more…

AMD MI3000A

How AMD May Get Across the CUDA Moat

October 5, 2023

When discussing GenAI, the term "GPU" almost always enters the conversation and the topic often moves toward performance and access. Interestingly, the word "GPU" is assumed to mean "Nvidia" products. (As an aside, the popular Nvidia hardware used in GenAI are not technically... Read more…

Shutterstock 1606064203

Meta’s Zuckerberg Puts Its AI Future in the Hands of 600,000 GPUs

January 25, 2024

In under two minutes, Meta's CEO, Mark Zuckerberg, laid out the company's AI plans, which included a plan to build an artificial intelligence system with the eq Read more…

Google Introduces ‘Hypercomputer’ to Its AI Infrastructure

December 11, 2023

Google ran out of monikers to describe its new AI system released on December 7. Supercomputer perhaps wasn't an apt description, so it settled on Hypercomputer Read more…

China Is All In on a RISC-V Future

January 8, 2024

The state of RISC-V in China was discussed in a recent report released by the Jamestown Foundation, a Washington, D.C.-based think tank. The report, entitled "E Read more…

Intel Won’t Have a Xeon Max Chip with New Emerald Rapids CPU

December 14, 2023

As expected, Intel officially announced its 5th generation Xeon server chips codenamed Emerald Rapids at an event in New York City, where the focus was really o Read more…

IBM Quantum Summit: Two New QPUs, Upgraded Qiskit, 10-year Roadmap and More

December 4, 2023

IBM kicks off its annual Quantum Summit today and will announce a broad range of advances including its much-anticipated 1121-qubit Condor QPU, a smaller 133-qu Read more…

  • arrow
  • Click Here for More Headlines
  • arrow
HPCwire