Some interesting paper in the Nov.-Dec. Issue of Micro 2010:
- Rethinking Digital Design: Why Design Must Change, from the group of Mark Horowitz. It emphasizes on the importance of processor generator based on processor templates in the energy efficient application specific design. The Figure 6 of this paper, originally published in ISCA 2010, is now widely cited by people in their talks.
- Performance and Energy Implications of Many-Core Caches for Throughput Computing, from Intel. The design space of the memory hierarchy seems to be quite limited.
- Workload Reduction and Generation Techniques, by Luk van Ertvelde and Lieven Eeckhout. Synthesize a white box by sampling a black box.
IEEE Computer Graphics and Applications has a special issue on camera culture:
- Andy Witkin: From Computer Vision to Computer Graphics, by Tony De Rose from Pixar.
- A Digital Gigapixel Large-Format Tile-Scan Camera, a thorough discussion on the challenges of designing a high resolution camera.
- Depth Director: A System for Adding Depth to Movies, computer graphics aided by computer vision.
JSSC Feb 2011 is a special issue on ISSCC 2010:
- A 345 mW Heterogeneous Many-Core Processor With an Intelligent Inference Engine for Robust Object Recognition, from the BONE group of KAIST.
- POWER7™, a Highly Parallel, Scalable Multi-Core High End Server Processor
- A 45 nm SOI Embedded DRAM Macro for the POWER™ Processor 32 MByte On-Chip L3 Cache
- A Family of 32 nm IA Processors, on the Intel Westmere.
- A 48-Core IA-32 Processor in 45 nm CMOS Using On-Die Message-Passing and DVFS for Performance and Power Scaling, from Intel.
- A 40 nm 16-Core 128-Thread SPARC SoC Processor
- An x86-64 Core in 32 nm SOI CMOS, from AMD.
Some good old papers on coarse-grained reconfigurable architectures:
- Lessons Learned from Designing the Montium – a Coarse-Grained Reconfigurable Processing Tile, in ISSOC 2004.
- Low-Power Domain Specific Processors for Digital Signal Processing, PhD thesis of Arthur Abnous, in 2001.
Some interesting papers in Feb 2011 issue of TPAMI:
- A Laplacian Approach to Multi-Oriented Text Detection in Video
- 3D Face Reconstruction from a Single Image Using a Single Reference Face Shape
- Learning to Detect a Salient Object
- Topology Preserving Relaxation Labeling for Nonrigid Point Matching
Reconfigurable baseband processing architecture for communication, in IET-CDT Jan 2011.
GPU-Friendly Multi-View Stereo Reconstruction Using Surfel Representation and Graph Cuts, to appear in j.cviu.
A Mixed-Precision Algorithm for the Solution of Lyapunov Equations on Hybrid CPU-GPU Platforms, to appear in Journal of Parallel Computing. The size of the matrix seems to be too large for normal systems.
Some funny papers by Robert W. Numrich:
- Self-similarity of parallel machines, to appear in Journal of Parallel Computing.
- The computational energy spectrum of a program as it executes, in the Journal of Supercomputing.
- A metric space for computer programs and the principle of computational least action, in the Journal of Supercomputing 2007.
Clusters are assumed in the analysis of self-similarity. A single core processor with cache is assumed in the analysis of computational spectrum. These two metrics are universal, but could be tricky when applying them to many-core architectures, e.g. GPU, Cell, etc. Other metrics could be more effective for the analysis of many-core architectures. Modern many-core architectures also consist of heavily non-linear components, which makes a linear mechanical model difficult to approximate the real system.
Prediction and Compensation of Motion Accuracy in a Linear Motion Bearing Table, to appear in the Journal of Precision Engineering.
Two recent free books:
- Introduction to High Performance Scientific Computing , by Victor Eijkhout in the FLAME project. The FLAME project has some other interesting books released earlier.
- Computer Vision: Algorithms and Applications, by Richard Szeliski from Microsoft.
MOSIX has released the MOSIX Virtual OpenCL Cluster Platform, which is a run time environment for heterogeneous platform.
Breakthroughs of the Decade: Products and Technologies That Changed the World, from xbitlabs. The products and technologies are only in the fields of electronics and computers.
The computer vision group of Microsoft Research Cambridge has some interesting publications on stereo vision, tracking, and machine learning. The group was said to provide machine learning technologies for the Kinect.
VLM – The Vision-Learning-Mining Research Lab has some high quality publications on gesture recognition.
Some details on the Sandy Bridge from anandtech. There will be a version consisting of more than 2000 pins. That is likely reaching the physical limit of maximum pin counts of a conventional socket.