Paper: 1
Author: Iain Bate and Ralf Reutemann
Title: Worst-Case Execution Time Analysis for Dynamic Branch Predictors
Is the paper well organized?
Yes, paper is well organized. All the relevant sections are included in a sequential manner
Title: It is self descriptive and complete. It gives a good amount of description to the audience about the Worst-Case Execution Time Analysis for Dynamic Branch Predictors.
Abstract:
o The authors had covered all the important points in the abstract which make the topic interesting to the audience.
o Here the authors provided the summary of the problem in a very difficult language as the use of long sentences can be avoided.
o The authors give a good description of the work done in the paper about the improvement in the existing branch prediction techniques.
o The authors can also specify the use of these techniques in a more correct way rather than specifying the reference where the techniques are applicable.
Introduction:
o This section is good and interesting for the audience to read as the authors gives a good description of the branch prediction techniques.
o The authors is also successful in explaining the purpose of the paper which states about tight but safe upper bounds on the number of branch mispredictions that can be expected for control statements.
o The authors provide a good summary on the organization of the rest of the paper.
Main section(s):
o The authors explain dynamic branch prediction techniques successfully. Use of the state transitions diagram and the table adds more to the understanding of the audience.
o The authors provide a good overview of the relevant work done by other The authors in this field. Provides good information on the work done in the field of improving processor performances.
o In this section, the use of the term static timing analysis is confusing as in previous section of the documents the authors uses timing analysis. As the terms used in the document must be consistent.
o All the theorems specified in the paper are justified.
o Use of the term T and N in this section used for representing static branch execution pattern needs description as what does it means is not specified in the document.
o Use of the diagram that represents the correlation between the branch history register (BHR) and pattern history table (PHT) adds more to the understanding of the concept. However, the way of explanation of the concept of the Global History Predictor is highly informative.
o The authors makes a good use of mathematical constructs tables to explain the upper bound on the number of mispredictions for the repeated execution of a simple nested loop using Global History Predictor and repeated execution of a repeated loop statement and repeated conditional statement within the loop using Bimolar predictor.
o Author’s gives provide explanations of the work done by him using the simulator which state practicality of the analysis. Overall gives good impression of the work done.
Conclusions:
o The authors had concluded paper very well provides information on the upper bound on the number of branch mispredictions has been derived for loop constructs and conditional statement
o The authors also provides limitations to the work done by them,
o The authors also give good description of the future work extension.
Reference:
o The authors had specified references in the paper wherever it was required and the format for specifying reference is also as per the standards. This further helps audience in getting more information on this topic.
Comment on the language used in the paper:
In this paper, the authors had explained every topic with the use of most feasible language except one which is used for describing summary of the problem. All the paragraphs are easy to read and understand.
General comments to the paper:
The paper is good and understandable by the audience. Subsequent use of Mathematical formulas, Graphs and some examples makes paper easy to understand for audience. However use of the some elements used in the paper needs some more explanations as the description of the table 1 can be specified by some mathematical formula or equation so that it will add more to the understanding of the audience.
Paper: 2
Author: Norman P. Jouppi
Title:
Improving Direct-Mapped Cache Performance by the Addition of a Small Fully Associative Cache and Prefetch Buffers
Is the paper well organized?
Yes, Paper is well organized .Author has covered various section in the paper. All the section describes are complete and structuring of the paper is also good which helps audience to understand and navigate paper.
Title:
It is complete within itself gives clear understanding to the audience about the improvement in the performance of the Direct-Mapped Cache.
Abstract:
· In this section author gives good description of the problem which states about the improvement needed in the caching techniques rather than improving the performance of CPU make the topic more interesting for the audience.
· Use of MIPS can to be replaced with its actual meaning as it has two meanings which is used to measure CPU performance as Million Instructions per seconds and Microprocessor without Interlocked Pipeline Stages.
· Author gives good description of the cache improvement techniques.
· Author also gives good description on the usage of cache improvement techniques.
Introduction:
· Author has provided good background information on how cache performance affects the processor performance.
· Use of Examples on machines gives more understanding of the problem definition.
· Author has also presented good summary of paper which help audience to navigate paper easily.
Main section(s):
· Author had explained various techniques used to improve Conflict Misses, and reducing Capacity and Compulsory Misses.
· Author has also explained the improvement in the miss rate of the cache on applying these hardware techniques on various hardware and has simulated results by use of various programs.
· Use of graph to show the test results gives clear understanding of the work done by the author in this field.
Conclusion:
· Author has concluded the paper very well by giving appropriate description of the work done for the performance improvement in the memory hierarchy
· Author provides limitation to the techniques presented in very elucidative way
· Author also gives good description of future work done that can be done in this field.
References:
Author has used correct reference format for giving reference to the terms and concepts used in the paper.
Comment on the language used in the paper:
The language used in the paper is simple and readable .It makes audience paper easy to understand. Use of the figures and graphs make paper more understanding. Examples which are presented by the author for explanation of the techniques gives audience good amount of understanding of the topic in the practical way.
General comments to the paper:
The paper is good and can be easily understood by the audience. Use of good examples explains the improvement techniques specified. Author is successful in justifying the problem and solution of the problem. The entire figures are explained neatly by the author; however author can include the related work done in the section of the cache improvement techniques which adds more understandability to the topic.
.