Urama Ifeoma Helen
Masters program in Intelligent Embedded Systems
()
Author, Title
Iain Bate and Ralf Reutemann,Worst Case Execution Time Analysis For Dynamic Branch Predictors.
Organisation
The paper organisation is good, with relevant sections presented in a comprehensible and well arranged manner.
Title
The title does not match perfectly with the contentsince it says little about the actual content of the work presented in the paper, which is "improving upon an existing branch prediction analysis".
Abstract
The abstract is good and well written, stating clearly the problem and authors contributions in solving the problem and also the manner the result was achieved.
Introduction
This part of the research paper introduced properly all the necessary points required in this section, starting with background information to understanding branch-prediction technique, with citation or references done where necessary. This was followed by statement of the problem and their contributions as well as summary.Related work in the research area were explained by comparing the present state of art and the need for the new approach which is planned to introduced. So, I would say that the section is good.
Main Section
The section is composedof two main parts namely; binomial and global history branch prediction schemes. The first part explained the approach used. It was mentioned that easy-to-predict and hard-to-predict were criteria for analysis classification. However nothing was mentioned again in subsequent sectionsregarding this and how they relatewith their approach.Explanation of basic terminologies and assumptions used in this paper makesthe paper easy to comprehend the analysis done for simple-loop construct and conditional construct.
Conclusions
The paper concluded well by summarizing how the approach was able to extend a more complex branch prediction-scheme example in an existing work. Furthermore, it showed how their approach positively affected the result of an existing example by providing more insights into the example. The future work that could be done in this area to extend the analysis to a more general WCET analysis of other components in the field of study were also highlighted.
References
The references were well cited within the paper and presented correctly in the reference section.
Language
The language used in this paper is good, though technical and requiring sufficient knowledge in subject matter in order to aid understanding.
General
In general, the work is organised and presented in an understandable way. Relevant sections were presented in well-structured sentences which make it easy to understand the authors ideas and contributions to the work.
------
Paper 2 Review
Author, Title
Norman P. Jouppi, Improving Direct-Mapped Cache Performance by the Addition of a Small Fully-Associative Cache and Prefetch Buffers.
Organisation
The paper is organized because the objectives as highlighted in Abstract and Introduction sections were clear and sequentially discussed in succeeding sections of the paper.
Title
The title sounds emphatic, implying that direct relationship exists. However, the content of the paper is narrative, providing sequence of events on how caches and prefetch buffers affect certain parameters of direct-mapped cache which in turn results in its improvement.
Abstract
It discussed hardware techniques required to improve performance of direct-mapped caches. Caching techniques were also discussed and their strengths and weaknesses highlighted. However, it contained too many details which were repeated within the body of paper.
Introduction
The introduction was brief and did not explain much about the subject of interest. It briefly mentioned of increasing interest on improving caches performance and the need for it. One would have expected a little review of past-related work to be discussed as well in order to ascertain originality and present status. Furthermore, Table 1.1 was taken from a published work and its source was not referenced.
Main Section(s)
In section 2, the author presented the baseline design. The modules that compose of the design and guidelines used were presented. Parametric analysis of the performance of the system and lost in memory hierarchy were also presented. The paper did not make a comparison of the baseline design results with those available today in the market. Section 3 discussed how misses in caches can be reduced. It also assessed miss caching and victim caching organisations. However, it failed to reference the source of charts shown in figures 3.4 and 3.5 respectively. In section 4, the paper examined different types of prefetch techniques and how they can be used to reduce capacity and compulsory misses.
Conclusions
It provided a recap of the discourse. The paper focussed on the main issues that concern the subject; that is, how direct-mapped cache performance can be improved by caches and prefetch buffers. Also, the limitations and delimitations of the study, together with future work required were presented. However, the elaborate discussion on the performance of the original base system and its accompanying chart (figure 5.1) should have been done within the body of the paper so that conclusion focuses on what have been achieved, etc.
References
The referencing was good. References made within the body of the paper match well with those made in the reference section. Also, the style and format used in writing the references are good and acceptable. The only issue is that the fifth point in reference section was poorly presented. There are guidelines such as APA, IEEE, etc, used to reference private conversations.
Language
The language used is clear and satisfying even though there were some sort of colloquial such as "have our cake and eat it too".
General
In general, the paper is clear and easy to understand by those with basic understanding of subject matter. Typographical errors were minimal. One critical issue is the absence of detailed literature review on the subject matter. It is imperative to see what others have done. Consequently, it became difficult to ascertain the originality of the paper and the current status of the subject matter.