Faster video encoding on Penryn family of processors - Fall IDF Video Demo

By Aaron Tersteeg (Intel) (118 posts) on November 12, 2007 at 9:48 am

Darren Yee, Tech Marketing Engineer Intel Software Solutions Group, and I shot a video demo of TMPGEnc 4.0 Xpress, a popular video encoder that was optimized for Quad-core and Intel® Streaming SIMD Extensions 4 (Intel® SSE4) instructions on the Penryn family of Intel processors at the Fall 2007 Intel Developer Forum.

The demo in the video show how the same application encoding video on both a Core 2 Duo and a Wolfdale dual core processor in Penryn family. This optimized version resulted in an over 2X improvement over Core 2 Duo using the same application encoding the same video clip. The Core 2 Duo machine encoded the file in 1:06 seconds where the Penryn machine tool only 40 seconds.

The primary instruction used in this software optimization was MPSADBW for optimized Sum absolute difference calculations, a common operation used in motion estimation. Visit the Penryn and the SSE4 instruction set page for more information.

Click To Play