Kaspersky Labs has begun experimenting with new detection software designed to leverage the power of NVIDIA’s Tesla GPUs.
Kaspersky uses a similarity algorithm in their search for new viruses on the Internet. The algorithm is designed to scrutinize files downloaded from the web and compare them to known viral strains in an attempt to isolate rogue chunks of code which could be viruses. When a virus is identified, it is added to a definition update which is later sent out to all end-user Kaspersky anti-virus clients when they synchronize with the server.
Kaspersky has rewritten their similarity engine in CUDA, and is now experimenting with the modified software on an NVIDIA Tesla S1070 setup. In initial testing, Kaspersky claims the CUDA-based engine achieved, “a 360-fold increase in the speed of the similarity-defining algorithm when compared to the popular Intel Core 2 Duo central processor running at a clock speed of 2.6GHz”.
The potential benefit of this approach to virus detection is enormous to Kaspersky’s customers. With such a substantial increase in detection speed of new viruses, customers using Kaspersky anti-virus can be assured that their AV protection will most likely be capable of recognizing new threats before they even have a chance to be infected. Kaspersky plans to build on this design in the future, aiming to eventually run their entire virus detection AI on Tesla-based multi-GPU computers.

Post a comment
Tweet this
Share on Facebook
Print this article


RSS Feeds