addition to language frontends, we are planning support for
further backends. Because PTX serves only as an intermediate representation, it is possible to translate and execute compiled megakernels on machines other than NVIDIA GPUs.
OptiX has a CPU fallback path that employs this approach.
One downside of OptiX, like any compiler, is that performance of the compiled kernel does not always match a
hand-tuned kernel for a specific use-case. We continue to
explore optimization techniques to close that gap. The original paper discusses performance in more detail.
We have also discovered tradeoffs in the compile-time specialization of kernels that achieve high performance, but result
in small delays when assumptions are violated and a kernel
must be regenerated. In the future, the system may choose to
fall back to a generalized kernel to maintain slightly degraded
interactivity while a new specialized kernel is compiled.
acknowledgments
The car, frog, and engine model in Figure 1 are courtesy of
TurboSquid. The bunny model in Figures 3 and 9 is courtesy of
the Stanford University Graphics Lab. Phil Miller was instrumental in keeping the effort on track. The authors benefited greatly
from groundwork and numerous conversations on ray tracing
with members of NVIDIA Research and the SceniX team.
hanrahan, P. Interactive k-d tree gpu
raytracing. In I3D ‘07: Proceedings of
the 2007 Symposium on Interactive
3D Graphics and Games (2007), acm,
new york, ny, usa, 167–174.
4. kajiya, j.t. the rendering equation.
In Computer Graphics (Proceedings of
ACM SIGGRAPH) (1986), 143–150.
5. lattner, c., adve, v. llvm: a
compilation framework for lifelong
program analysis & transformation.
In CGO ‘04: Proceedings of the 2004
International Symposium on Code
Generation and Optimization (2004).
6. lindholm, e., nickolls, j., oberman, s.,
montrym, j. nvIDIa tesla: a unified
graphics and computing architecture.
IEEE Micro 28 (2008), 39–55.
7. mcguire, m., luebke, D. hardware-
accelerated global illumination
by image space photon mapping.
In Proceedings of the 2009 ACM
SIGGRAPH/Euro Graphics conference
on High Performance Graphics (2009).
8. nvIDIa. Ptx: Parallel thread
execution Isa version 2. 3 (2011).
http://developer.download.nvidia.com/-compute/DevZone/docs/html/c/doc/
ptx_isa_ 2. 3.pdf.
9. nvIDIa. nvIDIa optix ray tracing
engine Programming guide version 2. 5
(2012). http://www.nvidia.com/object/
optix.html.
Morgan McGuire (morgan@cs.williams.
edu), nvIDIa and Williams college.
1. aila, t., laine, s. understanding the
efficiency of ray traversal on gPus.
In Proceedings of High-Performance
Graphics 2009 (2009), 145–149.
2. gribble, c. P., ramani, k. coherent
References
ray tracing via stream filtering.
In Proceedings of the 2006 IEEE
Symposium on Interactive Ray Tracing
(2008), 59–66.
3. horn, D.r., sugerman, j., houston, m.,
Steven G. Parker, heiko Friedrich,
David Luebke, Keith Morley,
James Bigler, Jared hoberock,
David McAllister, Austin Robison,
Andreas Dietrich, Greg humphreys,
and Martin Stich ({sparker, hfriedrich,
dluebke, kmorley, jbigler, jhoberock,
davemc, arobison, adietrich, ghumphreys,
mstich}@ nvidia.com), nvIDIa,
santa clara, ca.
© 2013 acm 0001-0782/13/05
You’ve come a long way.
Share what you’ve learned.
ACM has partnered with MentorNet, the award-winning nonprofit e-mentoring network in engineering,
science and mathematics. MentorNet’s award-winning One-on-One Mentoring Programs pair ACM
student members with mentors from industry, government, higher education, and other sectors.
• Communicate by email about career goals, course work, and many other topics.
• Spend just 20 minutes a week - and make a huge difference in a student’s life.
• Take part in a lively online community of professionals and students all over the world.
Make a difference to a student in your field.
Sign up today at: www.mentornet.net
Find out more at: www.acm.org/mentornet
MentorNet’s sponsors include 3M Foundation, ACM, Alcoa Foundation, Agilent Technologies, Amylin Pharmaceuticals, Bechtel Group Foundation, Cisco
Systems, Hewlett-Packard Company, IBM Corporation, Intel Foundation, Lockheed Martin Space Systems, National Science Foundation, Naval Research
Laboratory, NVIDIA, Sandia National Laboratories, Schlumberger, S.D. Bechtel, Jr. Foundation, Texas Instruments, and The Henry Luce Foundation.