A Collaborative CPU Vector Offloader: Putting Idle Vector Resources to Work on Commodity Processors

Most modern processors contain a vector accelerator or internal vector units for the fast computation of large target workloads. However, accelerating applications using vector units is difficult because the underlying data parallelism should be uncovered explicitly using vector-specific instructions. Therefore, vector units are often underutilized or remain idle because of the challenges faced in vector code generation. To solve this underutilization problem of existing vector units, we propose the Vector Offloader for executing scalar programs, which considers the vector unit as a scalar operation unit. By using vector masking, an appropriate partition of the vector unit can be utilized to support scalar instructions. To efficiently utilize all execution units, including the vector unit, the Vector Offloader suggests running the target applications concurrently in both the central processing unit (CPU) and the decoupled vector units, by offloading some parts of the program to the vector unit. Furthermore, a profile-guided optimization technique is employed to determine the optimal offloading ratio for balancing the load between the CPU and the vector unit. We implemented the Vector Offloader on a RISC-V infrastructure with a Hwacha vector unit, and evaluated its performance using a Polybench benchmark set. Experimental results showed that the proposed technique achieved performance improvements up to 1.31× better than the simple, CPU-only execution on a field programmable gate array (FPGA)-level evaluation.

Effectiveness Between Daily and After-Each-Case Room Disinfection of the Endoscopy Unit

Background: To evaluate the effectiveness between daily and after-each-case room disinfection in the endoscopy unit.Methods: This study was conducted in an endoscopy unit of the First Affiliation of Zhejiang Chinese Medical University. We cultured samples from the surface of endoscopy unit items, including operation unit air, isolation gown of an endoscopist, control panel buttons, workstation mouse, and the bed head of the patient. All the samples were divided into daily and after-each-case room disinfection groups. In addition, each group was subdivided into sedation and nonsedation gastroscopy with and without ventilation room groups.Results: The qualified rate of bed head samples of the patient were lower in the daily room disinfection group (76.67%) compared with the after-each-case group (100%). The isolation gown, mouse at the workstation, and the bed head of the patient demonstrated the lowest bacterial and fungal load in the after-each-case room disinfection group compared with the daily room disinfection group (p < 0.05). In the subgroup analysis, a higher microbial load was observed for the isolation gown of the endoscopist used during nonsedation gastroscopy in an unventilated room under the after-each-case room disinfection pattern (p < 0.05); a higher microbial load was observed for the control panel buttons used during nonsedation gastroscopy under the after-each-case room disinfection pattern (p < 0.05).Conclusions: For risk-free or low-risk patients, daily room disinfection provides the basic health requirements of the endoscopy procedure. However, it is better to adopt the after-each-case room disinfection for the isolation gown of the endoscopist and bed head of the patient. For the patients with high risk, the after-each-case room disinfection is more suitable for every endoscopy unit (www.ClinicalTrials.gov, NCT04399005).

Efficient Implementation of Multiply Accumulate Operation Unit Using an Interlaced Partition Multiplier

A new strategy for quick multiplication of two numbers is introduced. Inputs are separated into segments, and one segment is replaced by two with zeros interlocking in each alternative segments. With zero carries between segments the product are computed, within the time needed to multiply the short partitions and add the partial sums. The multiplication of two numbers generated and adds that product to an accumulator by multiply accumulate operation (MAC unit). This operation is performed within the MAC unit. MAC is an advanced co-processor that plays a vital role in FFT, DFT, etc. The MAC unit is utilized for additional execution and its input is given to the proposed multiplier that provides a trivial speed increment over the array multiplier designs. This paper is utilized to design speed enhanced multiply Accumulate Unit by an Interlaced Partition Multiplier. This new multiplier design simulation is optimized with existing method.

Design of Efficient Floating-Point Convolution Module for Embedded System

The convolutional neural network (CNN) has made great success in many fields, and is gradually being applied in edge-computing systems. Taking the limited budget of the resources in the systems into consideration, the implementation of CNNs on embedded devices is preferred. However, accompanying the increasingly complex CNNs is the huge cost of memory, which constrains its implementation on embedded devices. In this paper, we propose an efficient, pipelined convolution module based on a Brain Floating-Point (BF16) to solve this problem, which is composed of a quantization unit, a serial-to-matrix conversion unit, and a convolution operation unit. The mean error of the convolution module based on BF16 is only 0.1538%, which hardly affects the CNN inference. Additionally, when synthesized at 400 MHz, the area of the BF16 convolution module is 21.23% and 18.54% smaller than that of the INT16 and FP16 convolution modules, respectively. Furthermore, our module using the TSMC 90 nm library can run at 1 GHz by optimizing the critical path. Finally, our module was implemented on the Xilinx PYNQ-Z2 board to evaluate the performance. The experimental results show that at the frequency of 100 MHz, our module is, separately, 783.94 times and 579.35 times faster than the Cortex-M4 with FPU and Hummingbird E203, while maintaining an extremely low error rate.

Layout design of new production shop based on SLP

It has become the main way for heavy industry enterprises to decrease the cost to excavate “logistics resources” according to the Systematic Layout Planning (SLP). Taking the new tool shaft machining workshop of Luoyang yituo zhongcheng machinery co., LTD. as an example, through the method of field investigation and visit, we get the data of the process flow, material handling, the relationship between each process and the general situation of the new workshop. The SLP method is used to divide the work units in the workshop, analyze the relationship between logistics and non-logistics among each work unit, determine the comprehensive relationship between each operation unit, and make the position correlation diagram. Besides, we also combine the current situation of the workshop to sort out the layout scheme to reduce the production costs of the workshop.

Digitalisierung als Herausforderung für das Rechtsregime des Handwerks

Digitization is changing work tools and processes, customer acquisition opportunities and business models in the skilled crafts sector. This study examines the effects of this change on the applicability of the Crafts Code and the adequacy of this legal framework. The focus is on the demarcation of skilled crafts from industrial production, the demarcation of the stationary trade from the itinerant trade and the identification of the operation unit that is relevant for the applicability of the Crafts Code in different platform models. This study examines the aspects concerned and points out options for a future-oriented interpretation of the current law as well as for its further development.

PRINCIPLES OF FORMATION AND DESIGNING EQUIPMENT FOR FINISHING PARTS ON SCREW ROTOR BASIS

There are presented technologies and equipment based on screw rotors where technological and transportation operations are combined. The methods and ways for the speed performance comparative analysis of screw rotors are shown, for that there are carried out experimental investigations to determine a motion rate of parts with different weights from rotation frequency of screw rotors of 1-4 types. The results are accumulated in alignment charts which allowed creating a procedure for an engineering calculation of screw rotors. The original designs of such operation units have Russian priority and they are presented in fifty one patents for invention of the Russian Federation. In the paper there are offered methods and ways for the comparative analysis of screw rotor speed performance, the procedure for the engineering calculation of screw rotors. There is presented a machine for finishing parts with replaceable screw rotors of 1-4 types. The productivity and reliability increase of operation in such equipment is achieved at the expense of uninterrupted supply, dosage and reliable delivery of abrasive grains and parts into the operation unit made in the form of a screw rotor, worked parts separation from abrasive grains and working wastes, damage decrease of parts worked and their marketable appearance improvement, and also labor conditions improvement for the staff. There are shown inventions developed by the authors and for which they obtained patents of the RF and the information of which may be useful at the creation of equipment for OZO parts based on screw rotors.

Microfluidic Chamber Design for Controlled Droplet Expansion and Coalescence

The defined formation and expansion of droplets are essential operations for droplet-based screening assays. The volumetric expansion of droplets causes a dilution of the ingredients. Dilution is required for the generation of concentration graduation which is mandatory for many different assay protocols. Here, we describe the design of a microfluidic operation unit based on a bypassed chamber and its operation modes. The different operation modes enable the defined formation of sub-µL droplets on the one hand and the expansion of low nL to sub-µL droplets by controlled coalescence on the other. In this way the chamber acts as fluidic interface between two fluidic network parts dimensioned for different droplet volumes. Hence, channel confined droplets of about 30–40 nL from the first network part were expanded to cannel confined droplets of about 500 to about 2500 nL in the second network part. Four different operation modes were realized: (a) flow rate independent droplet formation in a self-controlled way caused by the bypassed chamber design, (b) single droplet expansion mode, (c) multiple droplet expansion mode, and (d) multiple droplet coalescence mode. The last mode was used for the automated coalescence of 12 droplets of about 40 nL volume to produce a highly ordered output sequence with individual droplet volumes of about 500 nL volume. The experimental investigation confirmed a high tolerance of the developed chamber against the variation of key parameters of the dispersed-phase like salt content, pH value and fluid viscosity. The presented fluidic chamber provides a solution for the problem of bridging different droplet volumes in a fluidic network.

Investigation and invention in carbapenem-resistant Klebsiella pneumoniae infection cases associated with Endoscopic retrograde cholangiopancreatography operation

Background: Carbapenem-resistant Klebsiella pneumoniae (CRKP) is associated with nosocomial infections that poses a serious threat to public health. According to a previous study, endoscopic retrograde cholangiopancreatography (ERCP) is considered a risk factor of CRKP transmission in the hospital. Methods: In this study, two cases infected with CRKP after ERCP were investigated. The origin of CRKP was determined by collecting the isolates from patients and screening the environment for ERCP units and the specific endoscopes. The antimicrobials susceptibility testing and molecular typing were performed for these CRKP. After post-ERCP infection happened, the procedure of endoscope disinfection was changed and hydrogen dioxide disinfection of ERCP unit was performed.Results: A total of five CRKP isolates were obtained from patients and screening the environment for ERCP units and the specific endoscopes, including three from the patients and two from the ERCP operating room. The CRKP from the patients and environment were both ST11, and the pulsed-field gel electrophoresis results showed that they shared identical bands, which indicated that the contaminated environment was associated with the nosocomial CRKP infections. After the control measures of endoscopes and hydrogen dioxide disinfection, post-ERCP infection decreased in the next six months.Conclusion: Early warning and response system should be established to control the spreading of CRKP in ERCP operation unit.

Airport Surface Modelling and Simulation Based on Timed Coloured Petri Net

In order to satisfy the requirements of International Civil Aviation Organization (ICAO) for aircraft taxi route planning in Advanced Surface Movement Guidance and Control System (A-SMGCS), an airport surface operation modelling and simulation approach based on timed and coloured Petri net is presented. According to the layout of the airport surface and the features of surface operation units, a static Petri net model of the airport surface is established. On this basis, in line with the requirements on the aircraft taxiing velocity in ICAO DOC 9830, the dynamic Petri net model of the airport surface operation is established by adding the time attribute to the static model. Additionally, the method of defining the capacity of airport operation unit place is proposed and the constraints of the airport surface operation are incorporated using Petri net elements. Unlike other papers in the field, the airport surface Petri net model established in this paper can simulate conflict-free taxiing using a Petri net simulator without relying on other model-independent algorithms. Based on the CPN Tools software, taking Toulouse Airport as an example, the validity of the model has been verified by comparing the model running data with real flight data.