Undergraduate Research Scholars Capstone (2006–present)https://hdl.handle.net/1969.1/33672024-03-29T08:01:13Z2024-03-29T08:01:13ZStreamlining TNS Data Collection for ML-Based RTL QoR Predictionhttps://hdl.handle.net/1969.1/2009252024-03-20T14:44:57ZStreamlining TNS Data Collection for ML-Based RTL QoR Prediction
Chip designs must meet several requirements before they are ready for fabrication. One of these requirements is achieving convergence on timing (frequency). Meeting this requirement is a time-consuming task for chip designers in the industry for two reasons. First, the standard approach to procuring this metric involves running logic synthesis and placement, both of which can take hours to weeks on larger RTL designs. Second, since the timing requirement is rarely met after one design iteration, these processes need to be rerun multiple times to recalculate the metric to ultimately converge on the design’s requirements. A critical measure of timing convergence is the total negative slack, commonly referred to by its acronym TNS. It indicates the sum of timing margins of all ‘negative slack’ paths that fail to meet the target clock cycle time. To expedite design convergence, our research team previously presented a machine learning-based approach to estimate the TNS values for chip designs expressed in Verilog hardware description language. This technique was orders of magnitude faster than running logic synthesis and placement on those same chips. In this work, we build on the previous approach by improving the initial data generation process. Getting “true” TNS values for training the machine learning models involves running logic synthesis and placement with hundreds of synthesis recipes for each design, resulting in tens of thousands of synthesis and placement runs. Driven by the need to create a rich training data set, since new designs will be continuously added to the RTL developer’s set of training designs, it behooves to reduce the number of synthesis and placement runs necessary to generate machine learning (ML) training data. By taking advantage of similarities in the distributions of TNS values across chip designs, the number of required synthesis and placement runs for n Verilog RTL designs and m unique synthesis recipes can be reduced from O(nm) to O(n+m) without meaningfully compromising the integrity of the training data and the accuracy of ML predictions. We present two methods for achieving this, both of which involve finding the common TNS distribution, then normalizing and computing missing values in the data set. The discoveries made by our research team have the potential to drastically reduce the time to market for a variety of semiconductor computing products, including but not limited to graphics processors, motherboards, and flash memory.
On a Series Involving Euler's Functionhttps://hdl.handle.net/1969.1/2009242024-03-19T18:50:46ZOn a Series Involving Euler's Function
The goal of this thesis is to provide an in-depth analysis and discussion of an equivalence to the Riemann Hypothesis (RH) proven by Jean-Louis Nicolas. Nicolas’ proof relates RH to an inequality of Euler’s totient function φ, and establishes a number-theoretic equivalence to RH. If Nicolas’ criterion holds for all primorial numbers, then RH is true. If not, then RH is false. This proof is given an original translation into English from French and annotated, with small corrections to computations and commentary when deemed necessary. His work is then extended by relating the equivalence to the convergence of an infinite series which is shown to converge to 1/2. Using this series and the related partial sum, consequences of the truth or falsehood of RH are explored in the context of Nicolas’ criterion. We assume both the truth and falsehood of RH, and in doing so underscore the extreme difficulty of this problem as well as the delicacy of the inequalities involved. Also provided are multiple programs which computationally verify expectations regarding different quantities from the analytic results section. Optimization of these programs are discussed as well as difficulties. These programs produce plots of the behavior of consequential arithmetic-valued functions, which are included in Chapter 4. The research results were limited by the nature of the problem. None of the analysis on the convergence criterion yielded a contradiction to an established result or conjecture, assuming either RH true or false. However, RH is known to be one of the most difficult problems in modern mathematics and significant progress was largely outside the scope of this thesis. The hope is that this research renews interest into Nicolas’ criterion specifically and arithmetic inequalities equivalent to RH in general.
Hydromedusae Blooms and Seasonal Biodiversity Changes in Galveston Bayhttps://hdl.handle.net/1969.1/2006662023-12-13T21:43:10Z2017-04-27T00:00:00ZHydromedusae Blooms and Seasonal Biodiversity Changes in Galveston Bay
Jellyfish of the class Hydrozoa (phylum Cnidaria) in the Gulf of Mexico are greatly understudied despite the fact that they are top predator and may have a significant ecological impact on fisheries and marine plankton in general. Medusae of the class Hydrozoa were collected every other day from October 2016 to February 2017 at the boat basin at the Texas A&M University at Galveston campus. Hydromedusae were isolated and examined for morphological characters. Each medusa was photographed, and DNA was extracted from every collected medusa. The mitochondrial 16S gene was amplified and sequenced, and the sequences were analyzed and compared with available sequences in a public repository, such as GenBank. Results illustrated that the abundance of Hydromedusae was not significantly correlated to water temperature, but was significantly correlated to salinity. Species diversity was varied throughout the sampling period, exhibiting the greatest amount of diversity in the Fall. The goal of this project is to contribute to long-term monitoring to assess the diversity and temporal fluctuations of the Hydromedusae population in Galveston Bay, and will be continued to obtain further information about the frequency and intensity of Hydromedusae blooms.
2017-04-27T00:00:00ZTowards Carbon Neutral Industrial Parkshttps://hdl.handle.net/1969.1/2006652023-12-13T21:42:00Z2020-04-23T00:00:00ZTowards Carbon Neutral Industrial Parks
CO2 emissions from industrial processes adversely affect the environment, with significant contributions to climate change. Therefore, there is a global need to reduce CO2 emissions into the atmosphere. Through this work, a tool that optimizes energy reuse while reducing emissions and maximizing profit was applied to an industrial cluster made up of several plants and/or processes producing several different products. There has been a recent focus on carbon capture utilization and storage solutions that integrate natural gas, energy, and other key materials like CO2. This work introduces an integration approach to design a carbon neutral industrial park from resources such as natural gas, water, air, emissions, and energy as heat and power, to produce value-added products. The approach applies a Linear Program (LP) that can be applied to various combinations of plants, to find the optimum configuration for a set target. An illustrative example that explores different target scenarios and combinations was investigated to verify the approach.
2020-04-23T00:00:00Z