|Currently at||NVIDIA: Tesla GPU Computing (CUDA) Group|
|Doing||General Manager, Tesla GPU Accelerated Computing Business Unit|
|sumitg at gmail.com|
Phd, CS (Embedded Systems), UC Irvine, 2003
B.Tech., EE, IIT Delhi, 1995
The gamut of my work experience ranges from product management, business management and product marketing to ASIC design to writing parallelizing compilers to EDA tools to FPGAs.
I currently manage the Tesla GPU Accelerated Computing business unit at NVIDIA. We promote the use of GPUs as very high performance accelerators that fit into standard x86 CPU based servers, workstations, and PCs. GPUs are massively parallel computing processors that accelerate applications that range from big data analytics to seismic processing in oil and gas, to sequence searches and alignment in bio-informatics, and risk analysis and options pricing in financial services.
NVIDIA's GPUs have evolved over the years to become massively parallel with 1000s of processor cores. This architecture is called the CUDA parallel computing architecture and enables GPU computing (also called GPGPU). The same floating point performance that is useful for graphics is also useful for scientific, technical, biological, imaging, and financial computing. The reason GPUs have become so popular as accelerators is that they are programmable using C, C++, Fortran using simple extensions or keywords, known as the CUDA extensions.
Prior to this, I was product manager at Tensilica responsible for the Xtensa and Diamond processor IP cores. I did both inbound product management (product definition, roadmap), and outbound product marketing (launches, sales assistance, collateral, sales tools).
Before this, I was an Entrepreneur-in-Residence at Tallwood Venture Capital, where I analyzed technological innovations and market forces in the reconfigurable, FPGA, and programmable spaces. I did market research and due diligence on business plans.
I worked for a few years in India after graduating from IIT Delhi. My first job was as a Software Engineer at IBM Global Services, in Bangalore India.
During the summer of 1998, I interned with the VLSI Systems Design and Methodologies group at IMEC in Leuven, Belgium. During the same summer, I also travelled through several parts of Europe (needless to say that was fun !).
Life as I remember it started in Kuwait (although I was born in Bhilai, Madhya Pradesh, India). I studied at the Indian School, Kuwait till the Iraqi invasion of Kuwait, after which I spent my "refugee" days in Delhi. I completed my high school at Mother's International School, Delhi in 1991.
I then went on to pick up a B.Tech in Electrical Engineering in 1995 from IIT Delhi (Indian Institute of Technology Delhi). (You can register yourself at this site, if you are an IIT Delhi Alumnus and at the IIT Delhi Alumni Association web page.)
I received a Phd from the School of Information and Computer science at UC, Irvine.
For my Ph.D., I worked on the crossroads of compiler technology and embedded system design and design automation. I (with a little/lot of help from my friends) built a C-based parallelizing high-level synthesis framework called SPARK.
The novel parallelizing high-level synthesis methodology that we developed for SPARK presents a leap forward in improving the quality of results for high-level synthesis.
My Ph.D. advisor was Prof Rajesh Gupta, who heads the Microlectronic Embedded System Lab at the Computer Science and Engineering Dept, at the University of California at San Diego.
Next, I did Post-doctoral research at the Center for Embedded Computer Systems on reconfigurable computing and system-level design and synthesis at UC San Diego & UC Irvine.
On the lighter side, I enjoy outdoor sports and am (was) into rock climbing, tennis, hiking. My last passion (before I became a family man) was Aikido with Sensei Haruo Matsuoka. I have also learnt some Wushu from Sifu Vadim Zukhov.