Learn about Perceptron Trick and its tricks

🚀 A crucially important thing for Understanding Hex Addressing in 32-bit Processors while using the datasheets which often asked to set specific bit or bits🖥️ While working with 32-bit processors, addressing bits using hex values is a critical skill! Each hex digit represents 4 bits, making it a handy way to visualize and set specific bits within a large address space. 🔍 For example, setting the 13th bit requires: 1. Identifying the hex nibble where the bit exists. 2. Drawing a table to map out the bit locations in their respective nibbles. 3. Writing down the previous value and updating the necessary bit, converting to hex again. 💡 Using the example shown: To set the 13th bit (starting from 0), I identified its place in the third nibble (0x000E000A). After setting the bit, the updated value becomes 0x000E200A. This method simplifies complex bit manipulations by visualizing them clearly! 📜 Rule of Thumb: Every hex digit corresponds to 4 bits, so you can systematically find which nibble contains the bit you want to set. Make changes at the nibble level to ensure accurate manipulation. #EmbeddedSystems #HexAddressing #32bitProcessor #BitManipulation #TechTips

When you focus on synthesizable-only Verilog like I do, it's easy to forget that you can do arbitrary computations during elaboration to compute constants for synthesis. Here's an example: a helper function to compute the Greatest Common Divisor of two integers, which is a useful building block for larger calculations. It needs the absolute value helper function, but that's trivial. https://lnkd.in/gvq7a6pG

21/06/2024 "Road to GATE" --------- Week 4 Update-------- * Further continued on the subject and explored new concepts * Learnt about Chip Select ,DRAM Refresh and Associative Memory. * Solved questions on Memory Organisation * Learnt about Locality of Reference and started with types Cache Mapping * Cache will be completed by this week and will start with Another key concepts of Pipelining Will start Operating System by next week as I will approach the end of this subject(70-75%). Bye and Will share more insights next week.

This is the code for my super-fast #DSP algorithm to apply FIR to a PDM bit stream. Instead of just conditionally adding values in a loop, it actually recreates the exact same operation but sideways, adding one FIR bit plane at a time by exploiting properties of bitwise AND. For a 128-bit kernel, it's about 6 times as fast as a naive loop. I haven't figured out if anyone else has invented the same thing. It seems like the field of optimizing fixed-point math for DSP is concentrated on PCM algorithms.

So, finally constructed the LCP algorithm.  There were three mistakes, during the process: 1. Collision Pair Matrix: Initially, I tried to use the adjacent collision matrix directly. However, this didn't work. Therefore, I reconstructed the adjacent collision matrix into pair matrices such as the column is each objects and rows are all of the pair from the given objects. 2. Unmovable Objects: At first, I didn't set unmovable object values to zero since trying to use constraint as equal to zero, which didn't work.  After setting them to zero directly, the algorithm functioned correctly. 3. Numerical Precision: In the beginning, I created fraction code using two-dimensional fractions with numerators and denominators in int value.  Thought it would be more precise than floating-point numbers. However, I later encountered problems with fractions like 1486618624/1796867625. These numbers couldn't be reduced since their GCD was one, and multiplying them with other fractions caused errors due to integer range overflow.  After switching to floating-point numbers, everything worked properly.

For detecting edges by calculating gradients in an image, the Sobel kernel is a great method. It can be applied in two directions: horizontal (x-direction) and vertical (y-direction). Why Sobel? The Sobel kernel incorporates a smoothing effect (weights in the middle are larger) to reduce noise sensitivity. It emphasizes the most significant changes in intensity, making it effective for edge detection. Get the Code example here https://lnkd.in/gRZCRaYV #Image_Processing #DeepLearing #ML

Algorithm (DV, advanced DV, LS)

Building a state machine to implement a serial 2-bit sequential adder. 💻 hashtag #StateMachine #SerialAdder #DigitalDesign #Digital_Electronics

Structure padding in C: It is another important concept which we misunderstood sometimes. Here is the addition of some empty bytes of memory in the structure to naturally align the data members in the memory. It is done to minimize the CPU read cycles to retrieve different data members in the structure. After alignment of data type in structure, there may be an unnecessary void in the structure memory, this unnecessary void need to be filled with null. This process of filling null is called padding. It can be seen that structure alignment and padding lead to an increase in size but this can be solved using a preprocessor directive called hashtag #pragma pack or by simply arranging the data elements in structure with an order of increasing or decreasing.

Building a state machine to implement a serial 2-bit sequential adder. 💻 #StateMachine #SerialAdder #DigitalDesign #Digital_Electronics https://j2geek.com/?p=143