Approximate Reciprocal-based Divider

 The paper can be downloaded from the following link: https://eclectx.org/Publications/C70.pdf

First, it summarizes the state of the art in approximate computing with a focus on the division. It is easily understood without prior knowledge of the state of the art. 

As for the proposed algorithm, the designed divisor is only 8/4 bits, and I am not convinced it is possible to easily extend it to a larger number of bits. The idea is that the transform the division into multiplication by an approximated reciprocal value. More specifically, the approximate 1/B by the K/2^n. With this approximation, the division becomes the multiplication by K followed by the shift by n.

The issue is the selection of the number K. In the paper, the authors divide only by 4-bit numbers, while the K itself is an 8-bit number. As such, the K is represented by 8 independent logical functions. Each of the functions has 4 inputs. 

For using this approach with larger numbers, the K generation needs to be revisited. 

An overview of lossless digital image compression techniques by Ming Yang, Nikolaos Bourbakis

The paper describes basic algorithms for the lossless image compression together with the references to other papers for the description of the specific algorithms. The length of the paper is unfortunately limited to four pages and therefore each algorithm has at most one paragraph of description. The paper seems to be good start for further study.

The paper formulates that every algorithm for lossless compression has two main stages: DECORRELATION and ENTROPY CODING.

Decorrelation phase removes spatial correlation between two samples (pixels) and ENTROPY CODING is responsible for the most effective encoding of the information.  Paper stated that Huffman coding was preferred over arithmetic coding due to the computation complexity and Intellectual property issues. The algorithms mantioned in the paper are LOCO-I, LOCO-A, CALIC, SCAN-based and DPCM. FELICS and CREW are briefly mentioned.

I will definitely revisit this paper if I will work on any image compression in the future.

Link: http://ieeexplore.ieee.org/document/1594297/

Comparative Study on Different Lossless Data Compression Methods by K.A. Ramya , M.Pushpa

Short paper mentioning  LZW, Shanon-Fano coding and Huffman coding algorithms. Paper provide basic description of the algorithm and provides their comparison on the ONE string.  I am still not sure, If the paper was written by human or generated by AI. If interested in these algorithms, I recommend to look on wiki instead.



Link: http://ijseas.com/volume2/v2i1/ijseas20160129.pdf

SPDP: A n Automatically Synthesized Lossless Compression Algorithm for Floating - Point Data by Steven Claggett , Sahar Azimi , and Martin Burtscher

The paper describe the automatically generated compression algorithm for dataset of real numbers. Authors describes 48 basic operation on datasets of the real numbers. Described operations work on the byte or word levels. Therefore, we do not see Huffman or arithmetic coding as one of them. Then they use they framework CRUSHER to try every possible combination of the operations of the given length (4 operations) and evaluate the resulting compression on the datasets. The best algorithm is called SPDP and the paper reports its compression ratio on several datasets. The achieved compression ration is between 1-2 for the most cases but the absolute best dataset achieves compression ration 34. 

Operations used by the compression algorithms are sufficiently described and I believe it should be possible to replicate  the work. The paper do not describe the CRUSHER framework but reference several additional paper which both use CRUSHER or should describe it.

Authors mentioned the possibility of the use  of CRUSHER to generate compression algorithm for specific dataset but the did not provide any time  measurement of the generation process.

It would be beneficial to read more papers about CRUSHER and possibly try the described approach on my data. 

Link: https://pdfs.semanticscholar.org/204a/f81b89d7b586d7b20025fd3b7f1120446e6b.pdf

Overview on different Lossless Data Compression Techniques by Shravana Karantha P

The paper very briefly describes several simple and basic compression methods. Every mentioned method is described in one paragraph and the original paper is referenced. After description of methods comparison is provided. However the methodology used for comparing different algorithms is not described and compressed data remain undisclosed. Moreover, the described compression methods are around 20 years old. 

I read this paper in hopes to discover link to the new state of the art approaches. I will look for the referenced articles but I do not expect to return to this paper in the future.


Link: https://www.ermt.net/docs/papers/Special_Issue/2017/ICETSE/120P.pdf

WILL SAFETY CRITICAL DESIGN PRACTICES IMPROVE FIRST SILICON SUCCESS? by HARRY FOSTER

Interesting evening read analysing quality of the verification techniques used on the current project. The paper does discuss statistics over last few years with focus on the first time success. It turns out, that most projects requires respin (in case of ASIC) or new design release (in case of FPGA).

For me as an engineer, the paper does not provide much because it do not recommend any techniques. The figure 8 can be used to compare the several verification techniques but the I would need more detailed discussion to prefer one technique over other.

According to the Figure 8, the constrained random has lowest percentage of the first silicon success which is 60 percent. The figure 6 states, that from all projects, about 33 percent achieves the first silicon success. These two numbers presents gap about 30 percent large. After reading the paper I am unsure what is the origin of the gap. What other project uses for verification? How successful are project without verification at all?

The second conclusion from the paper is that the application of the approaches and techniques for the safety critical systems do not increase change of the first silicon success.

I believe that the paper can be the source of the interesting information for the discussion about verification with management. As it nicely demonstrates, that others are late as well and that the correct application of the verification techniques increases chance of the first silicon success. It would be nice to know, how the application of the verification methodologies affect the timeline of the project. I would expect that the correct application of the verification through all phases of all projects will increase re-usability of the components from previous projects and reduce debugging time since most of the bugs will be catch during the simulation in verification tools instead during the system level testing.

Link to the paper: http://s3.mentor.com/public_documents/whitepaper/resources/mentorpaper_102839.pdf