NUMA-Booster algorithm

Among Infoworld‘s Green 15 winners was the following item

Intel uses some 50,000 servers to perform silicon design and development tasks — operations that grow increasingly complex each year, thereby increasing annual compute demand by 45 percent. Rather than piling more machines into its data centers, the company launched an initiative to squeeze more work out of its current fleet.

To do so, the chipmaker developed an algorithm dubbed NUMA-Booster to schedule design workloads to run more optimally than would otherwise using default OS scheduling on multicore NUMA systems. The algorithm increased server performance by 17 percent, effectively boosting the overall capacity of its server clusters by 11 percent. The company expects to see even greater gains as it retires older machines and deploys newer multicore models.

Based on the success of the pilot portion of this project, the company then worked with its finance and user groups to reduce forecasted servers and energy footprint while still meeting computing demand.

What is this algorithm? Are they going to license it?



Tell me (anonymous OK)

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s