Communications of the ACM
Latest Research
From Communications of the ACM
Technical Perspective: FPGA Compute Acceleration Is First About Energy Efficiency
"A Reconfigurable Fabric for Accelerating Large-Scale Datacenter Services" presents a research deployment of Field Programmable Gate Arrays (FPGAs) in a Microsoft...
From Communications of the ACM
A Reconfigurable Fabric For Accelerating Large-Scale Datacenter Services
We describe a medium-scale deployment of a composable, reconfigurable hardware fabric on a bed of 1,632 servers, and measure its effectiveness in accelerating the...
From Communications of the ACM
Technical Perspective: If I Could Only Design One Circuit . . .
"DianNao Family: Energy-Efficient Hardware Accelerators for Machine Learning" shows a deep understanding of both neural net implementations and the issues in computer...
From Communications of the ACM
Diannao Family: Energy-Efficient Hardware Accelerators For Machine Learning
We introduce a series of hardware accelerators (i.e., the DianNao family) designed for Machine Learning (especially neural networks), with a special emphasis on...
From Communications of the ACM
Technical Perspective: Toward Reliable Programming For Unreliable Hardware
"Verifying Quantitative Reliability for Programs that Execute on Unreliable Hardware" by Carbin et al. addresses challenges related to a bug, how likely it is to...
From Communications of the ACM
Technical Perspective: Combining Logic and Probability
In "Probabilistic Theorem Proving," Gogate and Domingos suggest how PTP could be turned in a fast approximate algorithm by sampling from the set of children of...
From Communications of the ACM
Probabilistic Theorem Proving
Many representation schemes combining first-order logic and probability have been proposed in recent years. We propose the first method that has the full power...
From Communications of the ACM
Technical Perspective: High-Performance Virtualization: Are We Done?
"Bare-Metal Performance for Virtual Machines with Exitless Interrupts" shows how to enable a virtual machine to attain "bare metal" performance from high-speed...
From Communications of the ACM
Bare-Metal Performance For Virtual Machines with Exitless Interrupts
We present ExitLess Interrupts (ELI), a software-only approach for handling interrupts within guest virtual machines directly and securely.
From Communications of the ACM
Technical Perspective: Treating Networks Like Programs
"Software Dataplane Verification" takes existing static checking of networks to a new level by checking the real code in the forwarding path of a Click router using...
From Communications of the ACM
Software Dataplane Verification
We present the result of working iteratively on two tasks: designing a domain-specific verification tool for packet-processing software, while trying to identify...
From Communications of the ACM
Technical Perspective: The Specialization Trend in Computer Hardware
Specialization improves energy-efficiency in computing but only makes economic sense if there is significant demand. A balance can often be found by designing...
From Communications of the ACM
Convolution Engine: Balancing Efficiency and Flexibility in Specialized Computing
We present the Convolution Engine (CE) — a programmable processor specialized for the convolution-like data-flow prevalent in computational photography, computer...
From Communications of the ACM
Technical Perspective: Big Data Needs Approximate Computing
"Neural Acceleration for General-Purpose Approximate Programs" demonstrates the significant advantages in cost, power, and latency through approximate computing...
From Communications of the ACM
Neural Acceleration For General-Purpose Approximate Programs
This paper describes a new approach that uses machine learning-based transformations to accelerate approximation-tolerant programs.
From Communications of the ACM
Technical Perspective: Rethinking Caches For Throughput Processors
As GPUs have become mainstream parallel processing engines, many applications targeting GPUs now have data locality more amenable to traditional caching. The...
From Communications of the ACM
Learning Your Limit: Managing Massively Multithreaded Caches Through Scheduling
This paper studies the effect of accelerating highly parallel workloads with significant locality on a massively multithreaded GPU.
From Communications of the ACM
Technical Perspective: Attacking a Problem from the Middle
"Dissection: A New Paradigm for Solving Bicomposite Search Problems," by Itai Dinur, Orr Dunkelman, Nathan Keller, and Adi Shamir, presents an elegant new algorithm...
From Communications of the ACM
Dissection: A New Paradigm For Solving Bicomposite Search Problems
In this paper, we introduce the new notion of bicomposite search problems, and show that they can be solved with improved combinations of time and space complexities...
From Communications of the ACM