Application-aware Performance Optimization for Software Managed Manycore Architectures

157100-Thumbnail Image.png
Description
One of the main goals of computer architecture design is to improve performance without much increase in the power consumption. It cannot be achieved by adding increasingly complex intelligent schemes in the hardware, since they will become increasingly less power-efficient.

One of the main goals of computer architecture design is to improve performance without much increase in the power consumption. It cannot be achieved by adding increasingly complex intelligent schemes in the hardware, since they will become increasingly less power-efficient. Therefore, parallelism comes up as the solution. In fact, the irrevocable trend of computer design in near future is still to keep increasing the number of cores while reducing the operating frequency. However, it is not easy to scale number of cores. One important challenge is that existing cores consume too much power. Another challenge is that cache-based memory hierarchy poses a serious limitation due to the rapidly increasing demand of area and power for coherence maintenance.

In this dissertation, opportunities to resolve the aforementioned issues were explored in two aspects.

Firstly, the possibility of removing hardware cache altogether, and replacing it with scratchpad memory with software management was explored. Scratchpad memory consumes much less power than caches. However, as data management logic is completely shifted to Software, how to reduce software overhead is challenging. This thesis presents techniques to manage scratchpad memory judiciously by exploiting application semantics and knowledge of data access patterns, thereby enabling optimization of data movement across the memory hierarchy. Experimental results show that the optimization was able to reduce stack data management overhead by 13X, produce better code mapping in more than 80% of the case, and improve performance by 83% in heap management.

Secondly, the possibility of using software branch hinting to replace hardware branch prediction to completely eliminate power consumption on corresponding hardware components was explored. As branch predictor is removed from hardware, software logic is responsible for reducing branch penalty. Techniques to minimize the branch penalty by optimizing branch hint placement were proposed, which can reduce branch penalty by 35.4% over the state-of-the-art.
Date Created
2019
Agent

Evaluation of compound semiconductors for infrared photo-detection applications

155448-Thumbnail Image.png
Description
In this dissertation research, conventional and aberration-corrected (AC) transmission electron microscopy (TEM) techniques were used to evaluate the structural and compositional properties of thin-film semiconductor compounds/alloys grown by molecular beam epitaxy for infrared photo-detection. Imaging, diffraction and spectroscopy techniques were

In this dissertation research, conventional and aberration-corrected (AC) transmission electron microscopy (TEM) techniques were used to evaluate the structural and compositional properties of thin-film semiconductor compounds/alloys grown by molecular beam epitaxy for infrared photo-detection. Imaging, diffraction and spectroscopy techniques were applied to TEM specimens in cross-section geometry to extract information about extended structural defects, chemical homogeneity and interface abruptness. The materials investigated included InAs1-xBix alloys grown on GaSb (001) substrates, InAs/InAs1-xSbx type-II superlattices grown on GaSb (001) substrates, and CdTe-based thin-film structures grown on InSb (001) substrates.

The InAsBi dilute-bismide epitaxial films were grown on GaSb (001) substrates at relatively low growth temperatures. The films were mostly free of extended defects, as observed in diffraction-contrast images, but the incorporation of bismuth was not homogeneous, as manifested by the lateral Bi-composition modulation and Bi-rich surface droplets. Successful Bi incorporation into the InAs matrix was confirmed using lattice expansion measurements obtained from misfit strain analysis of high-resolution TEM (HREM) images.

Analysis of averaged intensity line profiles in HREM and scanning TEM (STEM) images of the Ga-free InAs/InAs1-xSbx type-II strained superlattices indicated slight variations in layer thickness across the superlattice stack. The interface abruptness was evaluated using misfit strain analysis of AC-STEM images, electron energy-loss spectroscopy and 002 dark-field imaging. The compositional profiles of antimony across the superlattices were fitted to a segregation model and revealed a strong antimony segregation probability.

The CdTe/MgxCd1-xTe double-heterostructures were grown with Cd overflux in a dual-chamber molecular beam epitaxy with an ultra-high vacuum transfer loadlock. Diffraction-contrast images showed that the growth temperature had a strong impact on the structural quality of the epilayers. Very abrupt CdTe/InSb interfaces were obtained for epilayers grown at the optimum temperature of 265 °C, and high-resolution imaging using AC-STEM revealed an interfacial transition region with a width of a few monolayers and smaller lattice spacing than either CdTe or InSb.
Date Created
2017
Agent

Molecular Beam Epitaxy Using Bismuth as a Constituent in InAs and a Surfactant in InAs/InAsSb Superlattices

130244-Thumbnail Image.png
Date Created
2014
Agent