Jee-TC’13

Jee is headed out to the NVIDIA GPU Technology Conference (GTC) [link]. If you are attending, please drop by his poster, “P0159: A Roofline Model of Energy” [PDF].gtc-2013

Adieu, Shu!

(left-to-right) Rich, Shu, and Jee

(left-to-right) Rich, Shu, and Jee

We wish a hearty thanks and farewell to Shu Liu, whom we had the pleasure of hosting on a 6-month research internship. She heads back to Shanghai where she will finish up in the Georgia Tech-Shanghai Masters degree program. Best of luck, Shu — it was great having you visit!

SIAM CSE’13 talks

Aparna and Jee are at SIAM CSE 2013 in Boston this week, giving talks the fast multipole method and the energy roofline, respectively. (UPDATE: Click on the title for Jee’s talk to access the PDF slides.)

Tech report: Energy rooflines

Rooflines in time vs. "archlines" in energy

Rooflines in time vs. “archlines” in energy

Check out our new technical report, which presents a thought-experiment on the question of whether engineering an algorithm to optimize time differs from doing so with respect to energy (e.g., Joules).

  • Jee Whan Choi, Richard Vuduc. “A roofline model of energy.” Technical report no. GT-CSE-12-01, Georgia Institute of Technology, School of Computational Science and Engineering, Atlanta, GA, USA, December 2012. [PDF (452 KiB)]

Abstract

We describe an energy-based analogue of the time-based roofline model of Williams, Waterman, and Patterson (Comm. ACM, 2009). Our goal is to explain—in simple, analytic terms accessible to algorithm designers and performance tuners—how the time, energy, and power to execute an algorithm relate. The model considers an algorithm in terms of operations, concurrency, and memory traffic; and a machine in terms of the time and energy costs per operation or per word of communica- tion. We confirm the basic form of the model experimentally. From this model, we suggest under what conditions we ought to expect an algorithmic time-energy trade-off, and show how algorithm properties may help inform power management.

Subject area: High-Performance Computing
Keywords: performance analysis; power and energy modeling; computational intensity; machine balance; roofline model

Note (January 14, 2013): The above link to the PDF now points to the official version on the Georgia Tech Library’s report repository. If you downloaded an earlier version (dated December 24, 2012), you may wish to re-download this newer version. (The only difference is the addition of Appendix B, made on December 31, 2012.)

Post-semester disassembly

Just a random snapshot from The HPC Garage following the end of a busy semester.

Aparna "disassembles" following the semester's end. Assembly code courtesy of Marat.

Aparna “disassembles” following the semester’s end. Assembly code courtesy of Marat.

Bon voyage, Hana.

The HPC Garage bids a warm “adieu” to Hana Strakova, who visited from the University of Vienna this semester. She’s doing some interesting new work on algorithmic resilience using gossip-based methods. Best of luck upon your return, Hana, it was great having you around this semester!

Photo: Rich and Hana

Rich and fall semester intern, Hana (University of Vienna)

Synthesis Lecture on GPUs

Our Morgan & Claypool Synthesis Lecture on GPGPU performance analysis is now available! (Most libraries should have proxy access; the book is $30 otherwise, DRM-free.)


GT @ SC’12

The Georgia Tech both at Supercomputing 2012

The happy faces of Georgia Tech HPC at Supercomputing 2012. Please drop by!

The HPC Garage is at Supercomputing 2012! We are here with the rest of the Georgia Tech crew, which overall has a great presence at SC; see the list of talks and birds-of-a-feather sessions at sc12.gatech.edu

CSE Seminar: Energy “rooflines”

We are kicking off the CSE seminar this week, with a short “ideas” talk on the relationship among the time, energy, and power required by an algorithm. This talk features new work by Jee on the energy-based analogue of the time-based roofline model (of Williams, Patterson, and Waterman in CACM, 2009).

You can get the slides hereUpdate: You can also watch the talk here.


VECPAR’12 + K Supercomputer

Rich at the K Supercomputer facility in Kobe, Japan

Rich at the K Supercomputer facility in Kobe, Japan

 

Rich has the pleasure of being at the 10th International Meeting on High-Performance Computing for Computational Science (VECPAR), held this year at RIKEN AICS research facility in Kōbe (神戸市), Japan. RIKEN is home of the amazing K Supercomputer, shown here (only partly obstructed by Rich’s head, which he wishes had as much computing power).

Rich is especially grateful to the VECPAR organizers and other hosts (here, here, here, and here, among others) for the chance to speak, to see the K Computer, and to sample Kobe beef (yes, it’s that good!). For those interested in the talk, here is a link to Rich’s slides: [PDF].