cuda_spotlight

CUDA Spotlight: GPU-Accelerated Quantum Chemistry

todd-martinez-stanford

This week’s Spotlight is on Professor Todd Martínez of Stanford.

Professor Martínez’ research lies in the area of theoretical chemistry, emphasizing the development and application of new methods which accurately and efficiently capture quantum mechanical effects.

Professor Martínez pioneered the use of GPU technology for computational chemistry, culminating in the TeraChem software package that uses GPUs for first principles molecular dynamics. He is a founder of PetaChem, the company that distributes this software.

The following is an excerpt from our interview (you can read the complete Spotlight here).

NVIDIA: Todd, tell us about TeraChem.
Todd: TeraChem simulates the dynamics and motion of molecules, solving the electronic Schrodinger equation to determine the forces between atoms. This is often called first principles molecular dynamics or ab initio molecular dynamics.

The primary advantage over empirical force fields (for example, often used for protein structure) is that chemical bond rearrangements and electron transfer can be described seamlessly.

NVIDIA: Who are the users of TeraChem?
Todd: The primary users are academic researchers (from fields such as chemistry, materials and biology), but we are seeing increased interest from industry. Even though TeraChem was designed for ab initio molecular dynamics, it can also carry out simpler tasks such as optimization of the geometry of a molecular structure and calculation of reaction mechanisms.

Many people use TeraChem for these purposes, often referred to as quantum chemistry. Some of the most enthusiastic users have realized that TeraChem makes many simulations possible that were otherwise intractable. They have used TeraChem to optimize molecular structures for proteins, nanotubes, and minerals and to explore chemical reactions happening in these structures.

The protein BPTI (Bovine Pancreatic Trypsin Inhibitor) in TeraChem
The protein BPTI (Bovine Pancreatic Trypsin Inhibitor) in TeraChem

NVIDIA: How does TeraChem leverage GPU computing?
Todd: TeraChem was written for GPUs. I believe it remains the only quantum chemistry and/or ab initio molecular dynamics code which was designed for the GPU. The payoff from this redesign was remarkable – TeraChem is up to two orders of magnitude faster than similar programs designed for CPUs. This is bringing us much closer to the holy grail of computer-aided molecular design.

NVIDIA: When did you first see the promise of GPU computing?
Todd: Our interest in stream processing began around 2000. Our first efforts were focused on the Emotion Engine and Graphics Synthesizer in the PlayStation 2. We wrote a few key routines in assembler and ported the GAMESS quantum chemistry code to run on the PS2.

Writing in assembler was very tedious and the idea of rewriting all the code in assembler was unattractive. Thus, we were keeping close tabs on developments in GPGPU computing, which might be a more sustainable approach. Indeed, we were evaluating Brook as a possible avenue to expand that work.

Fortunately, we found out about NVIDIA’s plans to release CUDA from Professor Wen-mei Hwu at UIUC and were able to redirect these efforts into CUDA. Our first CUDA electronic structure code was written in late 2007, using a beta release of CUDA.

NVIDIA: What is unique about the TeraChem code?
Todd: We reformulated all the algorithms of quantum chemistry to map well onto GPU architectures. This implies a fair bit of precomputing to arrange the data so that the GPU can process it effectively. Because of this, TeraChem takes advantage of spatial locality more efficiently than other quantum chemistry packages. It also spends almost all of its execution time on the GPU – the CPU does very little work.

NVIDIA: In what ways is your team using CUDA?
Todd: All of our code development is now done with an eye towards stream processors and most of this is in CUDA. Unlike many who view the GPU as an accelerator for the CPU, we tend to see the CPU as a coprocessor for the GPU. Most of our computers are now loaded with eight GPUs in each node and we would be delighted to increase the GPU density even further. Reformulating algorithms to run on the GPU as much as possible has clearly been a winning strategy so far.


[Editor's Note: Learn more about GPU-based quantum chemistry at the GPU Technology Conference in March. Stanford graduate student Nathan Luehr will speak on Virtual Molecular Modeling Kits: Playing Games with Quantum Chemistry, on Thurs., March 27.]


Read the full interview with Todd Martínez. Read more CUDA Spotlights.

∥∀

About Calisa Cole

Calisa Cole
Calisa joined NVIDIA in 2003 and currently focuses on marketing related to CUDA, NVIDIA’s parallel computing architecture. Previously she ran Cole Communications, a PR agency for high-tech startups. She majored in Russian Studies at Wellesley and earned an MA in Communication from Stanford. Calisa is married and the mother of three boys. Her favorite non-work activities are fiction writing and playing fast games of online scrabble.