Fernanda recently posted about her GTC journey and it got me thinking about mine. Our careers have been somewhat adjacent for many years but it wasn’t until I joined Voltron Data that we actually worked at the same company. I was taking a look trying to remember, and I see photos from GTC 2014 that I think was my first, then speaking at GTC 2015 and there was a tutorial I gave the following year.
It was GTC 2015 where I presented “Visualization Toolkit: Faster, Better, Open Scientific Rendering and Compute”. We were not recorded but Robert Maynard and I presented where he talked about all the advances we had made using CUDA in VTK-m, and I presentetd the rendering side. The slides are still available from that talk here (at least for now), and I think at that meeting I was approached about writing this up for a new open access software journal, and later that year it was published.
I had similar feelings about earlier iterations of the conference that were smaller and there was a lot more exchange of ideas whereas this year felt more like. I thought quite a few times about applying for a position at NVIDIA but I never saw the right opportunity for me, I did a lot of work on TITAN leveraging GPUs in the early days of supercomputing working with partners at NVIDIA, universities and national laboratories. After a few years at Brookhaven National Laboratory I too found myself attending GTC 2024 as a proud employee of a GPU-based software startup.
It was exciting to be there, and I went primarily to dive into what was happening in the NVIDIA ecosystem after being a little out of the loop for three or four years. My primary focus before was always on the hardware accelerated rendering, cross-platform development and open source ecosystems in that space. As you might see from the slides I handed over to my co-presenter to discuss advances in CUDA and GPGPU, and so I was eager to expand my horizons a little.
I am still a hardware nerd, having built a PC last year with hand-picked parts and I really got a lot out of the Grace Hopper sessions. The discussion of ARM reminded me of porting I did to amd64 in Gentoo when I bought one of the new 64 bit chips way back when. I was always a little lukewarm on CUDA as I hoped for more open approaches that worked across different hardware vendors, but the success of the approach is clear and I was always pragmatic about using the best tools available to me.
I got a lot from the conference and remain excited about the future. I hope to continue evolving and finding the best role that I can play in this evolving ecosystem and I feel like I am well positioned to do that. As a bonus a number of coworkers were present and being in a remote startup it was a great chance to meet a number of them in person along with seeing a few colleagues I have not seen since well before the pandemic.
It also happened that Rob (now at NVIDIA) was there and we managed to grab a few drinks at the place we presented so many years ago but I totally failed to immortalize this moment or the previous one in 2015!