Mirror's Edge was one of my absolute favorite games! It's really cool to see it featured in the course slides :]
Fun fact about the game (that's relevant to this course): it was one of the first AAA games to use nVidia's then-new PhysX engine (this was 2009) and featured real-time physical simulation of cloth, shattering glass, steam, and other particle effects, which was pretty cool at the time (and made the game unbearably laggy).
A video demonstrating the technology: https://www.youtube.com/watch?v=uFio7wMTQ2k
keenan
and made the game unbearably laggy
Hey, you've got to push the envelope! :-)
I was actually working at NVIDIA a couple years before that; back in those days there was some talk of "physics processing units (PPUs)" with Ageia's PhysX being the prime candidate. The industry didn't go that direction, but we're now seeing some super interesting new hardware aimed at ray tracing/machine learning with the Turing architecture: NVIDIA RTX
lykospirit
Yes, I was watching the nVidia keynote at Gamescom and the ray tracing technology they showcased looked really impressive!
During the keynote, Jensen Huang mentioned that ray tracing at this level was "10 years in the making". What's special about the new RTX series that makes this possible? Is it sheer computing power, or something more? How does machine learning come into play?
More importantly, how much of this stuff will we learn in this course? :]
anonymous_panda
It will be great if ray tracing can be covered in this class. Also I am interested in the integrated deep learning model that accelerates ray tracing, like the model architecture, training pipeline, performance, comparison to SOTA and etc.
omegaiota
@anonymous_panda
Ray tracing will be covered in depth and in fact you'll be implementing your own pathtracer from scratch for the 3rd assignment!
keenan
@lykospirit I honestly having been following GPU hardware lately, but I'd love to hear a good explanation if you find one! It's definitely not just more transistors; features like these really have an impact on the architecture, and it wouldn't be surprising to me if a change like this was many years in the making (even something as simple as tessellation shaders saw many hardware revisions come and go before they actually made it into silicon). We'll talk a bit about GPUs, but not in any great depth. You can peruse the material from this CMU course: Visual Computing Systems but unfortunately it's no longer offered!
Mirror's Edge was one of my absolute favorite games! It's really cool to see it featured in the course slides :]
Fun fact about the game (that's relevant to this course): it was one of the first AAA games to use nVidia's then-new PhysX engine (this was 2009) and featured real-time physical simulation of cloth, shattering glass, steam, and other particle effects, which was pretty cool at the time (and made the game unbearably laggy).
A video demonstrating the technology: https://www.youtube.com/watch?v=uFio7wMTQ2k
Hey, you've got to push the envelope! :-)
I was actually working at NVIDIA a couple years before that; back in those days there was some talk of "physics processing units (PPUs)" with Ageia's PhysX being the prime candidate. The industry didn't go that direction, but we're now seeing some super interesting new hardware aimed at ray tracing/machine learning with the Turing architecture: NVIDIA RTX
Yes, I was watching the nVidia keynote at Gamescom and the ray tracing technology they showcased looked really impressive!
During the keynote, Jensen Huang mentioned that ray tracing at this level was "10 years in the making". What's special about the new RTX series that makes this possible? Is it sheer computing power, or something more? How does machine learning come into play?
More importantly, how much of this stuff will we learn in this course? :]
It will be great if ray tracing can be covered in this class. Also I am interested in the integrated deep learning model that accelerates ray tracing, like the model architecture, training pipeline, performance, comparison to SOTA and etc.
@anonymous_panda Ray tracing will be covered in depth and in fact you'll be implementing your own pathtracer from scratch for the 3rd assignment!
@lykospirit I honestly having been following GPU hardware lately, but I'd love to hear a good explanation if you find one! It's definitely not just more transistors; features like these really have an impact on the architecture, and it wouldn't be surprising to me if a change like this was many years in the making (even something as simple as tessellation shaders saw many hardware revisions come and go before they actually made it into silicon). We'll talk a bit about GPUs, but not in any great depth. You can peruse the material from this CMU course: Visual Computing Systems but unfortunately it's no longer offered!