Search

Power delivery has become a key element for the mobile platform to improve battery life and meet the energy star requirements. In this abstract, we will show innovative modeling and design techniques to improve the energy efficiency of the mobile platform power delivery subsystem. These techniques includes better modeling of the on die voltage distribution and optimizing the power delivery subsystem by optimizing for the light load. We will also show measurements of typical systems demonstrating the need for drastic improvements in the notebook power delivery in the next few years.

According to expansion of the market for portable products such as mobile phone, digital camera and PMP, adoption of WLP for various devices was increasing. Recently, WLP was being applied to higher pin count device as well as lower pin count device. In case that WLP is applied to higher pin count device, various reliability issues are expected from large die size and fine solder ball pitch. The more die size increase and contact area decrease, the more solder joint is fragile, especially at the outermost. To apply the WLP to mobile application, the reliability of various types of WLP in terms of die size, UBM (Under Bump Metallurgy) type, solder material and solder ball pitch have been evaluated at package and board level. First, the reliability for WLP of 3 mm ? 3 mm size and 0.5 mm solder ball pitch using solder ball with Sn3.0Ag0.5Cu solder composition on electroplated Cu/Ni/Au UBM had been evaluated by employing TC (Thermal Cycling) and HTS (High Temperature Storage) at package level and passed TC 2000cycles and HTS 1000hrs. Second, the reliability for WLP of 3.7 mm ? 3.9 mm size and 0.5 mm solder ball pitch using solder ball with Sn3.0Ag0.5Cu solder composition on electroplated Cu/Ni/Au UBM had been evaluated at package level and passed TC 500cycles, HTS 1000hrs and PCT 168hrs. Third, the reliability for WLP of 4 mm ? 4 mm size and 0.4 mm solder ball pitch using Sn3.0Ag0.5Cu on electroplated Ni UBM and ENIG (Electroless Nickel Immersion Gold) finished substrate has been evaluated by employing TC, THT (Temperature Humidity Test), 4-point bending and Drop test at board level. All items electrically passed except the one TC sample without underfill. As a result of cross-section analysis of TC samples without underfill, some solder crack on PCB side were found. For the 4-point bending samples, separation between Cu pad and TiW were found at die side of outermost row and crack between IMC and Ni Pad was found at PCB side of outermost. In case of drop test, there was no issue except the PCB resin crack under the outermost bonding pad. Finally, WLP of 6 mm ? 6 mm size and 0.4 mm solder ball pitch has fabricated and the reliability of TC, HTS, PCT, drop and electro-migration are being evaluated.

In this paper, we present several design and simulation innovations that resulted in significant improvements in the notebook energy efficiency. At the component level, we show innovations in the 45 nm technology & processor simulation and design methodologies. At the platform level, we did several measurements to understand where the energy is spent and proposed architectural improvements to several subsystems with special emphasis on the power delivery subsystem.

In this paper we show the measured power consumption of the different subsystems of the notebook with special emphasis on the power delivery subsystem. We prove by measurements that the energy efficiency of many of these subsystems can be significantly improved with a workload adaptive architecture without sacrificing performance at the peak load.

We present a practical and systematic strategy for measuring machine (robot) intelligence. A lot of research related to intelligent control has been carried out, but the subjects of definition and measurement of machine intelligence are not clearly formulated yet. We propose a human-oriented definition of machine intelligence and an intelligence task graph (ITG) as a modeling and analysis tool. By using an ITG, the machine contribution of human-machine cooperative systems is easily separated from the human contribution and directly described as numerical equations. Therefore we conclude that the ITG is very useful for estimating the machine intelligence quotient. This research will help engineers design intelligent robots which support human-friendly interfaces and perform environment controls with high performance.