Here is the abstract you requested from the SysPack_2011 technical program page. This is the original abstract submitted by the author. Any changes to the technical content of the final manuscript published by IMAPS or the presentation that is given during the event is done by the author, not IMAPS.
|High-Temperature Computing in Enterprise Environments: Challenges and Opportunities|
|Keywords: High temperature computing, TCO, reliability|
|The total power consumption in data centers has risen above 30 GW and many studies over the last decade have focused on making them more energy efficient. A significant percentage (approximately 25 – 35 %) of this energy is required in cooling IT equipment. Recently, dynamic thermal management of the IT equipment based on ‘when-and-where required' avoiding over-provisioning of cooling infrastructure has shown significant savings. To further curtail the cooling cost, industry has explored the opportunities of increasing the data center temperature. ASHRAE has supported this by widening its recommended air temperature envelope from 20-25 °C in 2004 to 15-32 °C in 2011. The elevated air temperature is expected to increase the cooling infrastructure heat removal efficiency and thereby, decrease the cooling cost. However, the impact of the elevated air temperature needs to be analyzed in a holistic manner by considering energy consumption at every layer in heat transport path from chip-to-ambient. The thermal requirements of the electronics devices can lead to increased server fan power at higher air temperature and hence, counteracting the gained savings in the chiller plant. Moreover, the increased leakage in the semiconductor leads to wasting a high percentage of the system power. Another important constituent in high-temperature computing challenge is IT equipment failure rate which aggravates with increasing temperature. The higher failure rate results in higher capital cost required to replace IT equipment more frequently. We will review a holistic end-to-end total cost of ownership (TCO) model which will be utilized to highlight the key challenges in achieving savings by using high temperature computing.|
|Niru Kumari, Research Scientist
Palo Alto, CA