Posts By Category

CodeStudy

[CodeStudy] Python Multiprocessing with Return Values Using Pool Mar 05, 2021
This post introduces multiprocessing in Python with return values from the child processing using Pool class.
[CodeStudy] Python Performance Analysis Mar 03, 2021
This blog introduces Python memory and execution time analysis tools Memory Profiler and cProfile.
[CodeStudy] Python Matplotlib Jan 22, 2021
Using Python Matplotlib to draw graph and plot it in a function.
[CodeStudy] HPWE and Its Interfaces between Hardware and Software Dec 26, 2020
This article introduces the MMIO register files of HWPE in Pulp SoC with its related c codes for simulation. It also gives hints of custom modifying the codes to use more registers or more events. I think it can also help you to understand the interaction between hardware and software.
[CodeStudy] Scala Excel Read: POI XSSF Nov 11, 2020
In this article, I introduced how to read a workbook, a sheet, a row and a special cell. The methods to obtain the row number and column number are also given. One way to filter empty cells is introduced too.
[CodeStudy] RocketChip Optional Bundle Oct 08, 2020
Learned some tips of Chisel via RocketChip. This introduces how to make the bundles be optional.
[CodeStudy] RocketChip MultiWidthFIFO Sep 25, 2020
Learned some tips of Chisel via RocketChip. This includes the Imp of Multi-Width-FIFO.
[CodeStudy] Some Chisel details in the project RocketChip Sep 24, 2020
Learned some tips of Chisel via RocketChip. Here includes come implicit classes, and one implementation of a gray counter.
[CodeStudy] Using C++ Lib Chrono to record execution time Aug 26, 2020
Chrono is a flexible collection of types that track time with varying degrees of precision.
[CodeStudy] Undefined reference to one function in CPP Aug 04, 2020
This blog shows a error caused by the separation of the declaration and implementation in template class.
[CodeStudy] The Implementation of TLM Simple Bus in SystemC Jul 01, 2020
Study the code: the implementation of timed TLM in SystemC
[CodeStudy] RocketChip Fuzzer Mar 29, 2020
Study the code: fuzzer in rocketchip. Including how to generate source id, how to send requirement via TileLink.

Emulate

[Emulate] SystemC Communication Ports Jul 18, 2020
SystemC Communication ports. Including chapter 11 and 12 of the book SystemC from the Ground Up. It introduces the communication ports in SystemC, port array, exports and shows all connectivity possibilities in SystemC as a handy reference.
[Emulate] Syntaxes of SystemC Jul 17, 2020
This blog introduces the different syntaxes of SystemC module and constructors.
[Emulate] Structure Design Hierarchy Jul 17, 2020
This post introduces six approaches of design hierarchy. It's the chapter 10 of the book From the Ground Up.
[Emulate] Overview of SystemC Components Jul 17, 2020
Overview of SystemC Components, the second chapter of from the ground up. It introduces the module and hierarchy as well as the three stages of SystemC simulation.
[Emulate] Typical Modeling Patterns with TLM API Jul 06, 2020
This blog introduces the typical modeling patterns with TML API, such as router, arbiter, pipeline. And also show how to explore the architecture with hub and spoke as well as cross bar switch.
[Emulate] TLM API 1.0 in SystemC Jul 03, 2020
TLM API 1.0 in SystemC. Including core interfaces and standard channels.
[Emulate] Transaction Level Modeling in SystemC Jun 29, 2020
This post descripts the timed TLM, including the three sets of master interfaces, two sets of slave interfaces
[Emulate] Refinement of Computation and Communication Jun 19, 2020
This post introduces Refinement of Computation and Communication in SystemC. Including different kinds of communication refinement, such as channel refinement, module refinement, hw-hw refinement, sw-sw refinement, hw-sw refinement. It also introduces the steps in communication refinement.
[Emulate] Interface and Channel Design Jun 16, 2020
This post introduces Interface and Channel Design in SystemC. Including primitive and hierarchical channels.
[Emulate] Untimed TLM in SystemC Jun 14, 2020
This post introduces Untimed TLM in SystemC
[Emulate] SystemC and Its Simulation Kernel Jun 12, 2020
This post introduces the components of SystemC, including Modules, Interfaces, Ports, Channels, Process and Events.
[Emulate] Different Abstraction Models Jun 11, 2020
Six different abstraction models by nctu. The models includes specification model, component assembly model, bus arbitration model, cycle accurate computation and RTL model.

Glean

[Glean] Backend Design Flow: SDC and Timing Constraints Apr 22, 2023
[Glean] Library Formats: CCS, ECSM, and NLDM Apr 03, 2023
This article provides an overview of three common library formats used in the design and analysis of digital circuits: Composite Current Source (CCS), Effective Current Source Model (ECSM), and Non-Linear Delay Model (NLDM), which is generated by ChatGPT4.
[Glean] Cadence Genus Synthesis Check List Mar 06, 2023
Here lists several messages that should be checked from the Genus synthesis log file to make sure there is no error and mismatch between the simulation and synthesis results.
[Glean] Latex includegraphics decodearray Oct 23, 2022
The use of the decodearray option to includegraphics allows the rendering colors to be changed.
[Glean] Two terms for timing analysis: WNS and TNS Oct 04, 2022
WNS (worst negative slack) and TNS (total negative slack), including a summary table from ChatGPT4.
[Glean] Package hyperref: Token not allowed May 15, 2022
Package hyperref Warning: Token not allowed in a PDF string (PDFDocEncoding):(hyperref). Using texorpdfstring to solve this.
[Glean] A better way to apply subfloat May 15, 2022
Simply wrapper the includegraphics with makebox to adjust the width of the caption and image separately.
[Glean] Logic Synthesis, Physical Synthesis Nov 24, 2021
The difference between Logic Synthesis and Physical Synthesis.
[Glean] GoF Design Pattern Overview Jun 14, 2021
The Overview of GoF Design Patterns: Creational Patterns, Structural Patterns, Behaviour Patterns and J2EE Patterns.
[Glean] Grouped Convolution May 27, 2021
Grouped convolution is a variant of convolution where the channels of the input feature map are grouped and convolution is performed independently for each grouped channels. There are also visualised graphs to show both spatial and channel domain of convolution, grouped convolution and other convolutions.
[Glean] What Is Memory-Hard Apr 19, 2021
In cryptography, a memory-hard function (MHF) is a function that costs significant amount of memory to evaluate. I also show the solution from Linzhi.
[Glean] Five Steps to Make an ASIC for Algorithm X Apr 19, 2021
Five Steps to Make an ASIC for Algorithm X: Math first, Optimization Target, Hardware-Software Boundary, Building Blocks, Physical Implementation
[Glean] ResNet-50 Architecture and # MACs Apr 03, 2021
This posts shows the basic architecture of the ResNet-50 and the number of weights as well as the MAC operations.
[Glean] 2.5D and 3D Interposer Mar 23, 2021
Interposers are wide, extremely fast electrical signal conduits used between die in a 2.5D configuration.
[Glean] Terms in the Intel Xeon E5-2600 V3 “Haswell-EP” Workstation Die Configurations Feb 06, 2021
Some terms in Intel Xeon E5-2600 V3 Haswell-EP Workstation Die Configurations, including ACA, LLC, Cbo, SAD, QPI, IIO.
[Glean] VC Formal Apps Feb 02, 2021
This post introduces the Apps of VC formal, including AEP, FCA, CC, SEQ, FRV, FXP, FPV, FTA, FSV, DPV, RMA, AIP and FuSa.
[Glean] Static Sign-Off, Formal & Simulation Feb 01, 2021
This post introduces the differences of Static Sign-Off, Formal and Simulation by three key functional verification metrics. analysis always finishes, all the violations flagged by the analysis, 100% of the failures are found.
[Glean] Formal Signoff Jan 29, 2021
This post introduces VC Formal Apps, the reason and goal of formal signoff. Later seven steps of formal signoff based on Synopsys are listed.
[Glean] Branch Strategy Jan 29, 2021
Version control strategies, like TBD, Git-Flow, GitLab-Flow.
[Glean] Turning Tax Jan 24, 2021
Turning Tax is a term taught in the advanced computer architecture by Paul H J Kelly at IC London. It describes the overhead (performance, cost, or energy) of the universality of the universal computing devices. It can be caused by instructions, data routing, register access and configurable ALU, where we can reduce the Turning Tax.
[Glean] Tomasulo Algorithm Jan 24, 2021
Tomasulo Algorithm eliminate three kinds of hazard RAW, WAR and WAW hazards by forwarding and renaming. The three stages of this algorithm are issue, execute and write back.
[Glean] Computer Architectures for Next Generation Applications Jan 18, 2021
This post is mainly translated from one zhihu answer by Bao Yungang. It introduces three laws: Moore_s law, Makimoto_s wave, Bell_s law and design methods and optimizations for performance and power as well as fragmented requirements in AIoT aging. These methods including reducing data movements, reducing data precious, improve parallelism and agile hardware development.
[Glean] Posit: A Potential Replacement for IEEE 754 Jan 18, 2021
Introduce the type III Unum, Posit. Including its four parts, the computation of the real value and recommend exponent bits.
[Glean] Functional Verification Cycle and Challenges Jan 14, 2021
This post introduces the four phases in the functional verification cycle and its four challenges to reduce time and improve robustness at each stage. The corresponding solutions are mentioned as well, which can be seen as the suitable situations for different verification methodology.
[Glean] State Explosion Problem and Formal Verification Jan 13, 2021
[Glean] Remove Empty File Folder Jan 11, 2021
Introduces two Linux command find and xargs. By combining this two command, you can easily remove empty directories and finish more jobs.
[Glean] Debugging Git Using Trace Jan 08, 2021
Debugging Git Using GIT_TRACE and restart the gpg-agent to solve the gpg failed to sign the data.
[Glean] Algorithm DG (Directed Graph) Dec 14, 2020
Usually, an algorithm is graphically represented as a DG to illustrate the data dependencies among the algorithm tasks.
[Glean] Classifying Algorithms Based On Task Dependences Dec 14, 2020
Algorithms can be broadly classified based on task dependences: Serial algorithms/Parallel algorithms/Serial–parallel algorithms (SPAs)/Nonserial–parallel algorithms (NSPAs)/Regular iterative algorithms (RIAs).
[Glean] IoU and NMS Dec 03, 2020
Intersection over Union (IoU) is an evaluation metric used to measure the accuracy of an object detector on a particular dataset. Non-maximum suppression (NMS) is a technique to remove duplicates and false positives in object detection.
[Glean] Design Compiler ddc file Nov 24, 2020
In general, it is binary file which contains both verilog gate level description and design constrains.
[Glean] Impact Map Sep 17, 2020
The introduction of impact map and how to create an impact map.
[Glean] Hardware/Software Codesign Aug 29, 2020
As the name implies, Hardware/Software Codesign (HSCD) denotes design methodologies for electronic systems that exploit the trade-offs and the synergy of Hardware (HW) and Software (SW).
[Glean] Content-addressable memory Aug 28, 2020
Content-addressable memories (CAMs) are hardware search engines that are much faster than algorithmic approaches for search-intensive applications. CAMs are composed of conventional semiconductor memory (usually SRAM) with added comparison circuitry that enable a search operation to complete in a single clock cycle. The two most common search-intensive tasks that use CAMs are packet forwarding and packet classification in Internet routers. I introduce CAM architecture and circuits by first describing the application of address lookup in Internet routers. Then we describe how to implement this lookup function with CAM.
[Glean] Code bloat Aug 25, 2020
In computer programming, code bloat is the production of program code (source code or machine code) that is perceived as unnecessarily long, slow, or otherwise wasteful of resources. Code bloat can be caused by inadequacies in the programming language in which the code is written, the compiler used to compile it, or the programmer writing it. Thus, while code bloat generally refers to source code size (as produced by the programmer), it can be used to refer instead to the generated code size or even the binary file size.
[Glean] ANSI Escape Codes Aug 06, 2020
ANSI escape sequences are a standard for in-band signaling to control the cursor location, color, and other options on video text terminals and terminal emulators.
[Glean] Tightly Coupled Memory Jul 20, 2020
The concept of Tightly Coupled Memory (TCM) and the difference between TCM and Cache.
[Glean] Network Topology Jul 10, 2020
Network topology is the arrangement of the elements of a communication network. Including point to point, bus, star, ring or circular, mesh, tree, hybrid, or daisy chain.
[Glean] Operator Fusion Jun 28, 2020
There are many opportunities, where fused operators—in terms of fused chains of basic operators—can significantly improve performance.
[Glean] All-Reduce Operations Jun 28, 2020
The all reduce operations are one kind of collective operations in NCCL and MPI lib.
[Glean] Unified Power Format Jun 25, 2020
The Unified Power Format (UPF) is a published IEEE standard. It is intended to ease the job of specifying, simulating and verifying IC designs that have a number of power states and power islands.
[Glean] Round-Robin Arbitration Jun 25, 2020
Round robin arbitration is a scheduling scheme which gives to each requestor its share of using a common resource for a limited time or data elements.
[Glean] Entropy Coding and Tunstall Coding Jun 14, 2020
This post introduces the concepts of entropy coding and Tunstall coding.
[Glean] 3D Convolution: kernel will traverse in 3-D Jun 12, 2020
This post introduces the 3D convolution.

ReadPaper

[Read Paper] Fast Algorithms for Convolutional Neural Networks Aug 19, 2021
Winograd’ s minimal filtering algorithms compute minimal complexity convolution over small tiles, which makes them fast with small filters and small batch sizes. However, this paper introduces only stride 1.
[Read Paper] Minimizing Computation in Convolutional Neural Networks Aug 18, 2021
Strassen algorithm can compute 2x2 Matrix Mult using only 7 multiplications.
[Read Paper] Attention Is All You Need Jan 07, 2021
This blog is the combination of two blogs which introduces the paper Attention is All You Need. Shortages and one improvement is shown, too.
[Read Paper] MaxpoolNMS: Getting Rid of NMS Bottlenecks in Two-Stage Object Detectors Dec 04, 2020
MaxpoolNMS, a parallelizable alternative to the NMS algorithm, which is based on max-pooling classification score maps.
[Read Paper] Domain-Specific Hardware Sep 24, 2020
Some tricks for design domain specific accelerators.
[Read Paper] Code compression for embedded VLIW processors using variable-to-fixed coding Aug 26, 2020
In this paper, it introduces a compress method which uses variable-to-fixed coding schemes based on either Tunstall coding or arithmetic coding to overcome the communication bottleneck between memory and CPU, especially for RISC or VLIW processors, which have a code size bloating problem compare to CISC processors.
[Read Paper] C-LSTM Enabling Efficient LSTM using Structured Compression Techniques on FPGAs Aug 11, 2020
This paper introduce some methods to accelerate the LSTM targeted on FPGA. It main utilize two methods: Block-Circulant Matrix; smaller coarse-grained pipelines with double-buffers
[Read Paper] In-Datacenter Performance Analysis of a Tensor Processing Unit Jul 31, 2020
This paper is the first generation of TPU. It introduces the goal and architecture of TPU and shows the performance comparation.
[Read Paper] Very Long Instruction Word Architectures and the ELI-512 Jul 20, 2020
This paper introduces three problems that VLIW meets and the corresponding possible solutions. I also posts the pros and cons of VLIW.
[Read Paper] The N3XT Approach to Energy-Efficient Abundant-Data Computing Jul 14, 2020
This paper introduces the framework of N3XT as well as its evaluation methodology. It also mentions the RRAM endurance resiliency.
[Read Paper] Deep Learning Hardware: Past, Present, and Future Jul 05, 2020
This paper introduces the following aspects: 1) identifies trends in deep learning research that will influence hardware architectures and software platforms of the future; 2) Five DL use cases with different hardware requirements; 3) Present and Future Deep-Learning Architectures; 4) Requirements for Future DL Hardware and Software
[Read Paper] Systolic Arrays for VLSI Jul 04, 2020
This paper proposes new multiprocessor structures and parallel algorithms for processing some basic matrix computations which are capable of pipelining matrix computations with optimal speed-up.
[Read Paper] TEA-DNN: the Quest for Time-Energy-Accuracy Co-optimized Deep Neural Networks Jun 28, 2020
This paper introduces a method to optimize DNN with Pareto-optimal models and Bayesian optimization.
[Read Paper] A Domain-Specific Supercomputer for Training Deep Neural Networks Jun 28, 2020
This paper introduces the TPU v2 and v3.
[Read Paper] An efficient kernel transformation architecture for binary- and ternary-weight neural network inference Mar 08, 2020
The initial kernels are transformed into much fewer and sparser ones, and the output feature maps are rebuilt from the immediate results. Overall, the number of total operations in convolution is reduced.
[Read Paper] A Survey of Accelerator Architectures for Deep Neural Networks Mar 01, 2020
A Survey of Accelerator Architectures for Deep Neural Networks.
[Survey] HPCforML and MLforHPC Feb 23, 2020
This survey contains two papers 1) Understanding ML driven HPC: Applications and Infrastructure; 2) Learning Everywhere: A Taxonomy for the Integration of Machine Learning and Simulations.
[Survey] MLforHPC Benchmarks Feb 16, 2020
I attached my recent survey on ML4HPC benchmarks, including three papers 1) A Modular Benchmarking Infrastructure for High-Performance and Reproducible Deep Learning; 2) HPC AI500: A Benchmark Suite for HPC AI Systems; 3) A Modular Benchmarking Infrastructure for High-Performance and Reproducible Deep Learning; and some other presentation slides.
[Read Paper] FPGA/DNN Co-Design: An Efficient Design Methodology for IoT Intelligence on the Edge Feb 09, 2020
Develop DNNs and the corresponding FPGA accelerators simultaneously. DNN designs should be FPGA-architecture driven, and FPGA accelerators should be DNN-aware.
[Read Paper] BitFlow: Exploiting Vector Parallelism for Binary Neural Networks on CPU Feb 09, 2020
A gemm operator network three-level optimization framework for fully exploiting the computing power of BNNs on CPU.
[Read Paper] Eyeriss v2: A Flexible Accelerator for Emerging Deep Neural Networks on Mobile Devices Jan 19, 2020
Sze Vivienne's Paper. This article contains more details of `Eyeriss V2`.
[Read Paper] ProxylessNAS: Direct Neural Architecture Search on Target Task and Hardware Jan 12, 2020
Han Song's Paper: ProxylessNAS. ProxylessNAS that can directly learn neural network architectures on the target task and target hardware without any proxy.
[Read Paper] Learning to Design Circuits Jan 12, 2020
Han Song's Paper: Learning to Design Circuits. Using ML to design analogue circuits.
[Read Paper] EIE: Efficient Inference Engine on Compressed Deep Neural Network Jan 12, 2020
Han Song's paper: EIE using CSC data format
[Read Paper] Design Automation for Efficient Deep Learning Computing Jan 12, 2020
Han Song's Paper, Design Automation, NAS, contains automatically designing specialized fast models, auto channel pruning, and auto mixed-precision quantization.
[Read Paper] A Survey on Methods and Theories of Quantized Neural Networks Jan 05, 2020
There is no excerpt to show~
[Read Paper] Eyeriss v2: A Flexible and High-Performance Accelerator for Emerging Deep Neural Networks Dec 29, 2019
Sze Vivienne's Paper. This article describes a performance analysis framework named `Eyexam` with roofline models and a DNN accelerator named `Eyeriss v2` which uses a hierarchical NoC.
[Read Paper] Eyeriss: An Energy-Efficient Reconfigurable Accelerator for Deep Convolutional Neural Networks Dec 29, 2019
Sze Vivienne's Paper. This article introduces the way to reduce the cost of data movement by exploiting data reuse in a multilevel memory hierarchy. Also, techniques such as compression and data-adaptive processing are introduced to save both memory bandwidth and processing power.
[Read Paper] Eyeriss: A Spatial Architecture for Energy-Efficient Dataflow for Convolutional Neural Networks Dec 29, 2019
Sze Vivienne's Paper. This article proposed RS dataflow which can adapt to different CNN shape configurations and reduces all types of data movement through maximally utilizing the processing engine (PE) local storage, direct inter-PE communication and spatial parallelism.
[Read Paper] Efficient Processing of Deep Neural Networks A Tutorial and Survey Dec 29, 2019
This survey focuses on: processing of DNN inference, addressing the efficiency of the CONV layers.

Survey

[Survey] GEMM, Strassen and Winograd Fast Convolution Algorithms Aug 21, 2021
This blog surveys the papers that optimise the convolution by GEMM, Strassen and Winograd algorithms.
[Survey] Current Verification Methods And Their Limited Situations Jan 11, 2021
This post introduces the current verification methods, steps and their limitations, including formal verification, constrained random verification (CRV) and hardware-software co-verification using virtual platform with hardware emulation and acceleration.
[Survey] HPCforML and MLforHPC Feb 23, 2020
This survey contains two papers 1) Understanding ML driven HPC: Applications and Infrastructure; 2) Learning Everywhere: A Taxonomy for the Integration of Machine Learning and Simulations.
[Survey] MLforHPC Benchmarks Feb 16, 2020
I attached my recent survey on ML4HPC benchmarks, including three papers 1) A Modular Benchmarking Infrastructure for High-Performance and Reproducible Deep Learning; 2) HPC AI500: A Benchmark Suite for HPC AI Systems; 3) A Modular Benchmarking Infrastructure for High-Performance and Reproducible Deep Learning; and some other presentation slides.

Tutorial

[Tutorial] Obtain GPT Prompts. Nov 12, 2023
This article introduces how to obtain the system prompts of the GPT, such as ChatGPT.
[Tutorial] Develop Chisel with Dev Container in Idea. Oct 19, 2023
This article introduces how to develop Chisel with Dev Container in IntelliJ Idea.
[Tutorial] Chisel simulation with the hierarchical BlackBox Module Jul 23, 2023
This article introduces how to do Chisel simulation with the hierarchical BlackBox Module.
[Tutorial] Scrum Master Guide Apr 07, 2023
Generated by ChatGPT4 -- This Scrum Master Guide provides an overview of the Scrum Master's role in five key Scrum meetings -- Sprint Planning, Daily Stand-up, Backlog Refinement, Sprint Review, and Sprint Retrospective. It also discusses the process of assisting in task breakdown and the importance of having a clear and concise Definition of Done. The guide is designed to be a helpful resource for Scrum Masters to facilitate team communication, collaboration, and continuous improvement.
[Tutorial] Obtain Objects in the collection in Genus Using Tcl Mar 06, 2023
collection is an extension provided by EDA vendors like Synopsys to support a list of objects in their Tcl API. Usually, most database query operations in Cadence and Synopsys would return a collection object. Complex query operations with filters may be slow in large design. Pre-store the query results might reduce runtime when it will be used in multiple places.
[Tutorial] Background Execution of Reporting Commands in Cadence Genus Mar 06, 2023
Cadence Genus supports doing report in parallel and running them in the background. This tutorial introduces how to conditional enable this feature using Tcl syntax.
[Tutorial] Quick Debug and Run Test on Chisel Repos based on CI Flow Files Feb 28, 2023
This tutorial introduces the quick way to debug the code of Chisel environment, such as Chisel3, playground, Rocket Chip, et al. The method introduced in this tutorial can also be used for other repos.
[Tutorial] Config CentOS 8 Server in 2022 Nov 02, 2022
Config CentOS 8 Server in 2022
[Tutorial] GTKWave and Verdi Enum tcl Commands Nov 26, 2021
In this tutorial, the tcl commands of GTKWave and Verdi for displaying enum are introduced.
[Tutorial] Python Hardware Behaviour Model Jan 31, 2021
In this tutorial, I will try to help you understand how to write a Python hardware behaviour model.
[Tutorial] Linux aliases Jan 30, 2021
Some useful Linux aliases.
[Tutorial] Configure Sublime Text for Verilog Dec 20, 2020
Configure Sublime Text for Verilog and SystemVerilog
[Tutorial] Configure and Run PULPissimo Dec 20, 2020
From install gcc and SDK to Run simple runtime examples for PULPissimo
[Tutorial] Install PDK Nov 23, 2020
How to install the PDK and obtain the db files
[Tutorial] Layout Two Graphs at One Main Graph Oct 17, 2020
Using column width to fit the image width
[Tutorial] Build SystemC Environment Jun 18, 2020
How to Build SystemC Environment in Windows and Linux Ubuntu.
[Tutorial] Suggest Using private Before val Apr 03, 2020
This tutorial suggests to use private as a prefix of val when create a wire or register and mentions one possible problem when using private.
[Tutorial] TileLink Spec Mar 21, 2020
The study of SiFive TileLink. Including TileLink buses, nodes and its chisel codes in chipyard.
[Tutorial] TileLink RegMap Mar 20, 2020
The study of TileLink TLRegMap
[Tutorial] Establish Linux Environment for Chisel and Chipyard Developments Jan 02, 2020
This tutorial will help you to establish a Linux environment for Chisel and Chipyard development quickly with little error.

WeeklyReview

[Weekly Review] 2022/01/03-2022/01/09 Jan 03, 2022
The weekly review 2022/01/03-2022/01/09
[Weekly Review] 2021/12/13-2021/12/19 Dec 19, 2021
The weekly review 2021/12/13-2021/12/19
[Weekly Review] 2021/12/06-2021/12/12 Dec 12, 2021
The weekly review 2021/12/06-2021/12/12
[Weekly Review] 2021/11/29-2021/12/05 Dec 05, 2021
The weekly review 2021/11/29-2021/12/05
[Weekly Review] 2021/11/15-2021/11/21 Nov 21, 2021
The weekly review 2021/11/15-2021/11/21
[Weekly Review] 2021/10/11-2021/10/17 Oct 17, 2021
The weekly review 2021/10/11-2021/10/17
[Weekly Review] 2021/09/27-2021/10/03 Oct 03, 2021
The weekly review 2021/09/27-2021/10/03
[Weekly Review] 2021/09/20-2021/09/26 Sep 26, 2021
The weekly review 2021/09/20-2021/09/26
[Weekly Review] 2021/09/13-2021/09/19 Sep 19, 2021
The weekly review 2021/09/13-2021/09/19
[Weekly Review] 2021/08/30-2021/09/06 Sep 06, 2021
The weekly review 2021/08/30-2021/09/06
[Weekly Review] 2021/07/26-2021/08/01 Aug 01, 2021
The weekly review 2021/07/26-2021/08/01
[Weekly Review] 2021/07/19-2021/07/25 Jul 25, 2021
The weekly review 2021/07/19-2021/07/25
[Weekly Review] 2021/05/24-2021/05/30 May 30, 2021
The weekly review 2021/05/24-2021/05/30
[Weekly Review] 2021/05/17-2021/05/23 May 23, 2021
The weekly review 2021/05/17-2021/05/23
[Weekly Review] 2021/05/10-2021/05/16 May 16, 2021
The weekly review 2021/05/10-2021/05/16
[Weekly Review] 2021/05/03-2021/05/09 May 09, 2021
The weekly review 2021/05/03-2021/05/09
[Weekly Review] 2021/04/26-2021/05/02 May 02, 2021
The weekly review 2021/04/26-2021/05/02
[Weekly Review] 2021/04/19-2021/04/25 Apr 25, 2021
The weekly review 2021/04/19-2021/04/25
[Weekly Review] 2021/04/12-2021/04/18 Apr 18, 2021
The weekly review 2021/04/12-2021/04/18
[Weekly Review] 2021/04/05-2021/04/11 Apr 11, 2021
The weekly review 2021/04/05-2021/04/11
[Weekly Review] 2021/03/29-2021/04/04 Apr 04, 2021
The weekly review 2021/03/29-2021/04/04
[Weekly Review] 2021/03/22-2021/03/28 Mar 28, 2021
The weekly review 2021/03/22-2021/03/28
[Weekly Review] 2021/03/15-2021/03/21 Mar 21, 2021
The weekly review 2021/03/15-2021/03/21
[Weekly Review] 2021/03/08-2021/03/14 Mar 14, 2021
The weekly review 2021/03/08-2021/03/14
[Weekly Review] 2021/03/01-2021/03/07 Mar 07, 2021
The weekly review 2021/03/01-2021/03/07
[Weekly Review] 2021/02/22-2021/02/28 Feb 28, 2021
The weekly review 2021/02/22-2021/02/28
[Weekly Review] 2021/02/15-2021/02/21 Feb 21, 2021
The weekly review 2021/02/15-2021/02/21
[Weekly Review] 2021/02/08-2021/02/14 Feb 14, 2021
The weekly review 2021/02/08-2021/02/14
[Weekly Review] 2021/02/01-2021/02/07 Feb 07, 2021
The weekly review 2021/02/01-2021/02/07
[Weekly Review] 2021/01/25-2021/01/31 Jan 31, 2021
The weekly review 2021/01/25-2021/01/31
[Weekly Review] 2021/01/18-2021/01/24 Jan 24, 2021
The weekly review 2021/01/18-2021/01/24
[Weekly Review] 2021/01/11-2021/01/17 Jan 17, 2021
The weekly review 2021/01/11-2021/01/17
[Weekly Review] 2020/12/21-2021/01/10 Jan 10, 2021
The weekly review 2020/12/21-2021/01/10
[Weekly Review] 2020/12/14-2020/12/20 Dec 20, 2020
The weekly review 2020/12/14-2020/12/20
[Weekly Review] 2020/12/07-2020/12/13 Dec 13, 2020
The weekly review 2020/12/07-2020/12/13
[Weekly Review] 2020/11/30-12/06 Dec 06, 2020
The weekly review 2020/11/30-12/06
[Weekly Review] 2020/11/02-11/15 Nov 15, 2020
The weekly review from 2020/11/02 to 11/15
[Weekly Review] 2020/10/26-11/01 Nov 01, 2020
The weekly review from 2020/10/26 to 11/01
[Weekly Review] 2020/10/19-10/25 Oct 25, 2020
The weekly review from 2020/10/19 to 10/25
[Weekly Review] 2020/10/12-10/18 Oct 18, 2020
The weekly review from 2020/10/12 to 10/18
[Weekly Review] 2020/10/05-10/11 Oct 11, 2020
The weekly review from 2020/10/05 to 10/11
[Weekly Review] 2020/09/21-10/04 Oct 04, 2020
The weekly review from 2020/09/21 to 10/04
[Weekly Review] 2020/09/14-20 Sep 20, 2020
Last week, I finished one implementation of one circuit in Chisel. I also learned Impact Map and User Story Map during the week.
[Weekly Review] 2020/08/31-09/13 Sep 13, 2020
Last week, I used Python to quickly verify my algorithm and its quite easy, especially reading from Excels and txt files. This week, I picked up Chisel and finished one submodule of the project. As these are heavily engineering works, I do think I need to learn some agile methods to boost my efficient and make my progress in the time table.
[Weekly Review] 2020/08/24-30 Aug 30, 2020
This week, the accelerator model successfully simulated the first layer of MobileNet V2. I also took almost two days to write my research proposal.
[Weekly Review] 2020/08/17-23 Aug 23, 2020
This week, I still worked on the Loosely Timed TLM.
[Weekly Review] 2020/08/10-16 Aug 16, 2020
This week, I still worked on the Loosely Timed TLM. I'm a little knowledge the concept of memory cell and memory structure. I spent a lot of time on optimizing the memory structure. I also learned a little about the SystemC TLM quantum keeper, but didn't use it in my modelling as I didn't think I need it to sync the time.
[Weekly Review] 2020/08/03-09 Aug 09, 2020
This week, I still worked on the Loosely Timed TLM. This post contains some thinking about implementation and modelling, Chisel and SystemC.
[Weekly Review] 2020/07/27-08/02 Aug 02, 2020
This week, I focusing on the TLM of Arch v2 in SystemC. I only studied the the Doulos tutorial and example. I post one paper note I read several weeks ago.
[Weekly Review] 2020/07/20-26 Jul 26, 2020
This week, I read the paper related to VLIW and learned the TLM tutorial video by accellera.org. I watched one tinyML workshop video related to the optimization of computation and memory.
[Weekly Review] 2020/07/13-19 Jul 19, 2020
This week, I read the paper related to N3XT technology and read the textbook SystemC From the Ground Up
[Weekly Review] 2020/07/06-12 Jul 12, 2020
This week, I read several papers related to Systolic Array.
[Weekly Review] 2020/06/29-07/05 Jul 04, 2020
This week, I read one paper and read the SystemC example project the Simple Bus.
[Weekly Review] 2020/06/22-28 Jun 28, 2020
This week, I read two and a half papers, including two TPU papers. I also learned several terms like round-robin arbitration, unified power format, all-reduce operations.
[Weekly Review] 2020/06/15-21 Jun 21, 2020
This week, I read two papers, one is Deep Learning Hardware: Past, Present, and Future.
[Weekly Review] 2020/06/08-14 Jun 14, 2020
This week, I began to learn the methodology of SystemC and its syntaxes. Also, I read some articles related to 3D convolution.
[Weekly Review] 2020/05/25-31 May 25, 2020
There is no excerpt to show~
[Weekly Review] 2020/05/18-24 May 24, 2020
This weekly review contains some backgrounds related to hardware-software co-design and ESL.
[Weekly Review] 2020/05/11-17 May 17, 2020
This weekly review includes cache atomic and false sharing.
[Weekly Review] 2020/05/04-10 May 10, 2020
This weekly review includes some knowledge related to cache indexed and tagged methods, TLB, coherence between Cache and DMA, coherence between iCache and dCache, coherence between multiple processors.
[Weekly Review] 2020/04/27-05/03 May 03, 2020
This weekly review contains spanning tree, A*, Primi's algorithm, Kruskal's algorithm, MST, dynamic programming and LIS. Also introduce some basic terms of Cache, such as offset, cache line, way, cache thrashing, et.
[Weekly Review] 2020/04/20-26 Apr 26, 2020
This weekly review includes the introduction of SystemC, modeling, JVM Memory, Rocket Chip's interruption PLIC and CLINT. Also, including CS61B's Graph.
[Weekly Review] 2020/04/13-19 Apr 19, 2020
Contains CS61B binary search tree, red-black trees, and hashing, heap; Three methods to dump vcd files (waveform) in Chisel Testers2; The first two generations verification and the coming third generation verification, plus the defination of simulator, emulation and formal verification.
[Weekly Review] 2020/04/06-12 Apr 12, 2020
Contains cuset retiming, zen of Python, and some knowledge of CS61B.
[Weekly Review] 2020/03/30-04/05 Apr 05, 2020
Contains BDD, TDD, CI and CS61B. Plus the ICR in Chisel.
[Weekly Review] 2020/03/23-29 Mar 29, 2020
This weekly review contains Scala intersection, union and complement, as well as ScalaDoc tags. Also, introduce using console to print colorful log. An error occurred while I using `RegInit` without giving the width to UInt.
[Weekly Review] 2020/03/16-22 Mar 22, 2020
include Scala high-order function, Scala Regex, Chisel forkwithRegion. Also, the definition of `base address` and `offset`
[Weekly Review] 2020/03/09-15 Mar 15, 2020
git commit types. chisel `withRegion`, Scala `collect`, et.
[Weekly Review] 2020/03/02-08 Mar 08, 2020
This weekly review contains the usage of Linux `tree` and Chisel `<>` as well as `:=`. Also, DecoupledDriver.
[Weekly Review] 2020/02/24-03/01 Mar 01, 2020
This week I read a deep learning accelerator survey named 'A Survey of Accelerator Architectures for Deep Neural Networks'. Also, I tried to use a Scala library named `Breeze`.
[Weekly Review] 2020/02/17-23 Feb 23, 2020
This week, I continued on the survey of ML4HPC and found several papers of Indiana University, which described the definitions of ML4HPC and its subcategories. Also, I finished the draft implementation of GLB cluster with some test.
[Weekly Review] 2020/02/10-16 Feb 16, 2020
This review contains one way to think matrix multiply, one Chisel class named DataMirror which can monitor the details of ports, and a discussing of how can RoCC accelerator communicate with L2 cache. Also, I continued my survey at AI for HPC.
[Weekly Review] 2020/02/03-09 Feb 09, 2020
This review contains the usage of general data type in Chisel, the basic architecture of NN and the introductions of BNN and the BitFlow algorithm. Also, some materials related to HPC+ML.
[Weekly Review] 2020/01/27-02/02 Feb 02, 2020
This review contains some hotchip19's slides and materials of HPC
[Weekly Review] 2020/01/20-26 Jan 26, 2020
This weekly review contains the usage of `grep` as well as Scala Patton Match
[Weekly Review] 2020/01/13-19 Jan 19, 2020
This weekly review contains Chisel syntax such as mem, Vec and mem test.
[Weekly Review] 2020/01/06-12 Jan 12, 2020
This review contains some Chisel and Scala syntaxes studying notes.
[Weekly Review] 2019/12/30-2020/01/05 Jan 05, 2020
This weekly review contains two terms: `Stackelberg Game` and `Fog Computation`.
[Weekly Review] 2019/12/23-29 Dec 29, 2019
This review contains some troubles I met while setting up the Chisel develop environment.
[Weekly Review] 2019/12/16-22 Dec 22, 2019
This review contains some basic knowledge of Scala, and the tutorial of deep learning accelerator designs named 'Efficient Processing of Deep Neural Network: from Algorithms to Hardware Architectures'.
[Weekly Review] 2019/12/09-15 Dec 15, 2019
This review contains come basic knowledge related to git, RISC-V, Chipyard, RoCC interface, SHA3 and cache.

Workshop

[Workshop] Using UPF for Low Power Design and Verification Nov 27, 2021
This workshop describes the detailed information related to UPF. Including its definition, terminology, some Tcl commands, etc.
[Workshop] ESWEEK 2021 QuantumFlow Oct 09, 2021
This workshop introduces the basic of Quantum Circuit and its applications on Quantum Deep Learning Neural Networks.
[Workshop] Hot Chips 2020 Marvell Details ThunderX3 CPUs Sep 04, 2020
Hot Chips 2020, Marvell Details ThunderX3 CPUs
[Workshop] Hot Chips 2020 Google TPUv2 and TPUv3 Sep 04, 2020
Hot Chips 2020, Google TPUv2 and TPUv3
[Workshop] Hot Chips 2020 Cerebras WSE Programming Sep 04, 2020
Hot Chips 2020, Cerebras WSE Programming
[Workshop] Hot Chips 2020 Alibabas Hanguang 800 NPU Sep 04, 2020
Hot Chips 2020, Alibabas Hanguang 800 NPU from Anandtech
[Workshop] Hot Chips 2020 Highlights Aug 31, 2020
Some highlights of Hot Chips 2020
[Workshop] tinyML Talks: AIML SoC for Ultra-Low-Power Mobile and IoT devices Jul 22, 2020
This workshop introduces two computation optimization methods and three memory optimization methods. Address Generation HW Unit and pipeline architecture are helpful to computation optimization. Dequantization, entropy compression and pooling on the fly are benefit to memory optimization.
[Workshop] ScaledML: Moore's Law in the age of AI Chips Jul 12, 2020
The presentation by Jim Keller, Intel. Introduces the Moore's law, complexity limits and technology optimism for AI chips.
[Workshop] tinyML Talks: Saving 95% of Your Edge Power with Sparsity Jun 28, 2020
It will explain these types of sparsity (time, space, connectivity, activation) in terms of edge processes, and how they affect computation on a practical level.
[Workshop] tinyML Talks: Low-Power Computer Vision Jun 28, 2020
By utilizing hierarchical neural network, we can separate the big neural network into much small ones, hence reduce the training time and inference power consumption. However, it might increase the latency.