Douglas M. Carmean

A 0.18 /spl mu/m CMOS IA32 microprocessor with a 4 GHz integer execution unit

The processor has an execution unit with high bandwidth capability and low average instruction latency. The processor pipeline includes an Execution Trace Cache, Renamer, Scheduler, register file and execution unit. IA32 instructions are decoded when they are fetched from the L2 cache after a miss in the Execution Trace Cache. Serving as the primary instruction cache, the Execution Trace cache stores decoded instructions to remove the long delay for decoding IA32 instructions from this path, reducing the branch missprediction loop. Instruction traces follow the predicted execution path, not sequential instruction addresses. While this pipeline supplies the high bandwidth work stream, the length of this pipe contributes to instruction latency only when there is a branch miss-prediction (roughly once in 100 instructions).

Power modeling and thermal management techniques for manycores

The rising number of cores in manycore architectures, along with technology scaling, results in high power densities and thermal issues on the die. To explore innovative thermal management techniques in such processors, we need an accurate online estimate of the power consumption. In this paper, we present the first ever power model for Intel many integrated core processors, which we use to show the benefit of a novel manycore specific thermal management technique called workload intermixing. Our proposed model leverages performance monitoring events and accounts for operating voltage and clock frequency. We validate our model for Intel Knights Ferry (KNF) design and show that we have an average of 4.73% prediction error vs. measurements. We provide the breakdown of total power into three main components: compute, memory, and interconnect and use it as an input for thermal management. Our simulation results show that our proposed intermixing of workloads in KNF architecture can reduce the total number of thermal emergency situations by 58% with energy savings of 14% on average.

PIXEE: pictures, interaction and emotional expression

An interactive system, PIXEE, was developed to promote greater emotional expression in image-based social media. Images shared on social media were projected onto a large interactive display at public events. A multimodal interface displayed the sentiment analysis of images and invited viewers to express their emotional responses. Viewers could adjust the emotional classification and thereby change the color and sound associated with a picture, and experiment with emotion-based composition. An interdisciplinary team deployed this system around the world to explore new ways for technology to catalyze emotional connectedness. This paper describes the system, design iterations, and observations about how people used it for self-expression and connection.

Demonstrating PIXEE: pictures, interaction and emotional expression

An interactive system, PIXEE, was developed to promote greater emotional expression in image-based social media. An interdisciplinary team developed this system and has deployed it as a cultural probe around the world to explore ways that technology can foster emotional connectedness. In this system, images that participants share on social media are projected onto a large interactive display. A multimodal interface displays the sentiment analysis of image captions and invites viewers to adjust this classification in order to express their emotional response to the images. Viewers can adjust the emotional classification and thereby change the colors and sound associated with a picture, and compose musical scores by touching a series of images. CHI participants will be able to share their own content and their emotional reaction to other images throughout the conference. If CHI attendees share feedback about presentations through this system, an affective map of the conference may emerge.