High Performance Computing Group - Industrial
Affiliates Program
Operating System Projects
Follows a list of the research projects that have been started in the
last 4 months or that are about to be launched.
OS Support for Application-Specific
Resource Management in Multiprocessors
This project is focused on the the development of microkernel-based
environments to support parallel applications. The main goal consists of
finding the best place to take scheduling decisions in order to get the
best performance. We address issues such as processor and physical memory
dynamic hardware partitioning, memory conscious scheduling and event notifications
to user-level. Policies are located at user-level servers and the kernel
mechanisms are tunable by the different applications.
...clik here for more information...
Exokernel-based tunable multiprocessor
operating systems
This project is focused on the the development of microkernel-based
environments to support parallel applications. The main goal consists of
finding the best place to take scheduling decisions in order to get the
best performance. We address issues such as processor and physical memory
dynamic hardware partitioning, memory conscious scheduling and event notifications
to user-level. Policies are located at user-level servers and the kernel
mechanisms are tunable by the different applications.
...clik here for more information...
Cooperative Caching and Prefetching
for Parallel/Distributed File System
The goal of this project is to design a new
generation in cooperative caches where physical
locality is not the main issue to achieve a high-performance
cache. In this new cache,
the coherence, fault-tolerance and load-distribution
problems will be solved in an efficient way.
Finally we will also present aggressive prefetching
algorithms that take advantage of cooperative caches.
...clik here for more information...
Hardware Conscious Operating
Systems
In this project we plan to propose new
algorithms for the operating system that take hardware
issues such as caches into account. For instance,
we plan to propose scheduling algorithms that will try to avoid switching
the context when a given process is taking
advantage of hardware elements such as cache, predictors, etc.
...clik here for more information...