eHiTS Lightning changes the game of Virtual Screening and Docking
When we announced eHiTS Lightning to the world to our users and at conferences we were greeted with a balance of scepticism and interest. Colleagues working with FPGAs and GPUs wondered why we would do the work to implement on the Cell processor rather than follow more tried and true tested methods. Well, we’re not a company to always follow the rules. Fortunately for us Zsolt Zsoldos,our Chief Technical Officer is of the nature to pursue the best solution. He is focused on delivering the best solutions for our users and doing the best science possible. Based on some of the news we are seeing floating around of late (and to be expanded upon in later posts) we made an appropriate decision in terms of choosing a processor that is becoming mainstream, is certainly outperforming FPGAs and GPUS and is offering us the ability to move away from some of the hype associated with what we promised to truly delivering the outstanding performance we expected. For now we are shipping to our beta testers the latest beta release of eHiTS Lightning for people to test. What have we done in the latest release?
During the development process to improve the system we have performed a lot of testing. These tests will form the basis of a series of future blog posts and/or technical notes in the near future. For now I’ll summarize the results here. In comparison to the previous beta-release, we have improved:
- Stability: the system stability has improved dramatically as we have performed tests and hit issues we have resolved immediately.
- The Pose Prediction Accuracy has improved significantly since the previous release. We have tested on a series of datasets which have served as our testing sets for our platform for a number of previous releases.
Relative to our previous eHiTS release on the Intel platform we have achieved:
- 40 x speed up in performance. This is a consistent speed improvement for large molecules containing more than seven rigid fragments (see image below).
- There is much better enrichment based on testing on 10 different test cases. For these particular test cases we see a 20-30% improvement in enrichment factors.
- We have been working on examining correlations with binding energy (pKi,pKd) in a number of our tests over the past few months. We are seeing improved performance over all historical correlations
In our hands we are seeing improvements across the board. The software is now in the hands of our users for their feedback. We will communicate some of the results summarized here in more detail shortly.
posted by Aniko
December 18th, 2008 at 5:28 pm
[…] We are always watching for innovative solutions in docking. We acknowledge those scientists pushing towards the edge of performance and excellence. When we saw the recent announcement regarding the DockStar solution from Silicon Informatics we were interested to see whether they had made some of the promised breakthroughs with their GPU-based solution. Their website promises “With the combined power of the DockStar™ Linux Workstation, NVIDIA’s® Tesla™ GPU’s and our proprietary software kernels, Silicon Informatics’ DockStar™ solution outperforms conventional workstations by 10 - 20+ times.” The system is based on the Autodock 4.0 software platform. As commented in my recent blogpost we have been doing a lot of work to validate the performance of eHiTS Lightning and gathering validation data for throughput, pose accuracy and enrichment so we were interested to compare our data with those of the GPU-based DockStar solution. We’ll report the data in much more detail in a Case Study note presently in development but our observations at present are based on comparing to information they have on the site. […]
January 7th, 2009 at 10:55 pm
“Thanks for provide me this informative info. Great! I am really waiting for your next post.
Thanks for sharing!”