Hasil Pencarian  ::  Simpan CSV :: Kembali

"Heterogeneous integration uses packaging technology to integrate dissimilar chips, LED, MEMS, VCSEL, etc. from different fabless houses and with different functions and wafer sizes into a single system or subsystem. How are these dissimilar chips and optical components supposed to talk to each other? The answer is redistribution layers (RDLs). This book addresses the fabrication of RDLs for heterogeneous integrations, and especially focuses on RDLs on: A) organic substrates, B) silicon substrates (through-silicon via (TSV)-interposers), C) silicon substrates (bridges), D) fan-out substrates, and E) ASIC, memory, LED, MEMS, and VCSEL systems. The book offers a valuable asset for researchers, engineers, and graduate students in the fields of semiconductor packaging, materials sciences, mechanical engineering, electronic engineering, telecommunications, networking, etc."

"Dalam HPC, pemanfaatan GPU untuk kapabilitas pemrosesan paralelnya dapat mempercepat komputasi secara masif, terutama untuk masalah yang embarrassingly parallel. Namun, saat memilih model pemrograman GPU, portabilitas model pemrograman dan sistem vendor GPU harus dipertimbangkan. Untuk memahaminya dengan lebih baik, makalah ini menganalisis waktu eksekusi baseline CUDA, HIP, dan SYCL pada GPU NVIDIA dan AMD. Program UVaFTLE, sebuah program yang digunakan untuk menentukan Lagrangian Coherent Structures melalui ekstraksi Finite-Time Lagrangian Exponents (FTLE), digunakan untuk mengukur waktu eksekusi. Eksperimen ini menunjukkan kinerja CUDA dan SYCL pada kedua platform GPU, yang secara konsisten mengalahkan HIP dalam waktu eksekusi. Upaya untuk mengoptimalkan waktu eksekusi fungsi kernel GPU di seluruh platform juga dilakukan, secara drastis memangkas waktu eksekusi kernel preproc hingga lebih dari 90%. Setelah optimasi, SYCL tetap menjadi yang terbaik, sementara CUDA berada di posisi kedua, dan HIP yang terlihat jelas paling lambat. Makalah ini juga membahas tantangan development yang dihadapi. CUDA dan SYCL membanggakan dokumentasi dan dukungan komunitas yang sangat baik, sementara dokumentasi HIP tertinggal dan tidak memberikan pengalaman development yang positif seperti kedua model lainnya.

---

In HPC, leveraging GPUs for their parallel processing capabilities can massively accelerate computation, especially for embarrassingly parallel problems. However, when choosing a GPU programming model, one must take into consideration the portability of the programming model and the GPU vendor of their system. To understand them better, this paper analyzes the baseline execution time of CUDA, HIP, and SYCL on both NVIDIA and AMD GPUs. The UVaFTLE program, a program used to determine Lagrangian Coherent Structures through the extraction of Finite-Time Lagrangian Exponents (FTLE), is used to benchmark execution time. The experiment showcases the performance of CUDA and SYCL on both GPU platforms, which consistently beat HIP in execution time. An effort to optimize the execution time of the GPU kernel functions across is made, drastically cutting the execution time of the preproc kernel by over 90%. After the optimizations, SYCL remains as the champion, while CUDA comes second, and HIP is clearly the slowest. This paper also discusses the development challenges encountered. CUDA and SYCL boast excellent documentation and community support, while HIP’s documentation falls behind and does not provide a developer experience as positive as the other two."

"Dalam HPC, pemanfaatan GPU untuk kapabilitas pemrosesan paralelnya dapat mempercepat komputasi secara masif, terutama untuk masalah yang embarrassingly parallel. Namun, saat memilih model pemrograman GPU, portabilitas model pemrograman dan sistem vendor GPU harus dipertimbangkan. Untuk memahaminya dengan lebih baik, makalah ini menganalisis waktu eksekusi baseline CUDA, HIP, dan SYCL pada GPU NVIDIA dan AMD. Program UVaFTLE, sebuah program yang digunakan untuk menentukan Lagrangian Coherent Structures melalui ekstraksi Finite-Time Lagrangian Exponents (FTLE), digunakan untuk mengukur waktu eksekusi. Eksperimen ini menunjukkan kinerja CUDA dan SYCL pada kedua platform GPU, yang secara konsisten mengalahkan HIP dalam waktu eksekusi. Upaya untuk mengoptimalkan waktu eksekusi fungsi kernel GPU di seluruh platform juga dilakukan, secara drastis memangkas waktu eksekusi kernel preproc hingga lebih dari 90%. Setelah optimasi, SYCL tetap menjadi yang terbaik, sementara CUDA berada di posisi kedua, dan HIP yang terlihat jelas paling lambat. Makalah ini juga membahas tantangan development yang dihadapi. CUDA dan SYCL membanggakan dokumentasi dan dukungan komunitas yang sangat baik, sementara dokumentasi HIP tertinggal dan tidak memberikan pengalaman development yang positif seperti kedua model lainnya.

---

In HPC, leveraging GPUs for their parallel processing capabilities can massively accelerate computation, especially for embarrassingly parallel problems. However, when choosing a GPU programming model, one must take into consideration the portability of the programming model and the GPU vendor of their system. To understand them better, this paper analyzes the baseline execution time of CUDA, HIP, and SYCL on both NVIDIA and AMD GPUs. The UVaFTLE program, an program used to determine Lagrangian Coherent Structures through extraction of Finite-Time Lagrangian Exponents (FTLE), is used to benchmark execution time. The experiment showcases the performance of CUDA and SYCL on both GPU platforms, which consistently beat HIP in execution time. An effort to optimize the execution time of the GPU kernel functions across is made, drastically cutting the execution time of the preproc kernel by over 90%. After the optimizations, SYCL remains as the champion, while CUDA comes second and HIP is clearly the slowest. This paper also discusses the development challenges encountered. CUDA and SYCL boast excellent documentation and community support, while HIP’s documentation falls behind and does not provide adeveloper experience as positive as the other two."

"Dalam HPC, pemanfaatan GPU untuk kapabilitas pemrosesan paralelnya dapat mempercepat komputasi secara masif, terutama untuk masalah yang embarrassingly parallel. Namun, saat memilih model pemrograman GPU, portabilitas model pemrograman dan sistem vendor GPU harus dipertimbangkan. Untuk memahaminya dengan lebih baik, makalah ini menganalisis waktu eksekusi baseline CUDA, HIP, dan SYCL pada GPU NVIDIA dan AMD. Program UVaFTLE, sebuah program yang digunakan untuk menentukan Lagrangian Coherent Structures melalui ekstraksi Finite-Time Lagrangian Exponents (FTLE), digunakan untuk mengukur waktu eksekusi. Eksperimen ini menunjukkan kinerja CUDA dan SYCL pada kedua platform GPU, yang secara konsisten mengalahkan HIP dalam waktu eksekusi. Upaya untuk mengoptimalkan waktu eksekusi fungsi kernel GPU di seluruh platform juga dilakukan, secara drastis memangkas waktu eksekusi kernel preproc hingga lebih dari 90%. Setelah optimasi, SYCL tetap menjadi yang terbaik, sementara CUDA berada di posisi kedua, dan HIP yang terlihat jelas paling lambat. Makalah ini juga membahas tantangan development yang dihadapi. CUDA dan SYCL membanggakan dokumentasi dan dukungan komunitas yang sangat baik, sementara dokumentasi HIP tertinggal dan tidak memberikan pengalaman development yang positif seperti kedua model lainnya.

---

In HPC, leveraging GPUs for their parallel processing capabilities can massively accelerate computation, especially for embarrassingly parallel problems. However, when choosing a GPU programming model, one must take into consideration the portability of the programming model and the GPU vendor of their system. To understand them better, this paper analyzes the baseline execution time of CUDA, HIP, and SYCL on both NVIDIA and AMD GPUs. The UVaFTLE program, a program used to determine Lagrangian Coherent Structures through the extraction of Finite-Time Lagrangian Exponents (FTLE), is used to benchmark execution time. The experiment showcases the performance of CUDA and SYCL on both GPU platforms, which consistently beat HIP in execution time. An effort to optimize the execution time of the GPU kernel functions across is made, drastically cutting the execution time of the preproc kernel by over 90%. After the optimizations, SYCL remains as the champion, while CUDA comes second, and HIP is clearly the slowest. This paper also discusses the development challenges encountered. CUDA and SYCL boast excellent documentation and community support, while HIP’s documentation falls behind and does not provide a developer experience as positive as the other two.

"

"Heterogeneous embedded systems, signal representations and different abstraction levels, architectures and components based on hardware and software, in all the main phases of design (modeling, validation with multiple models of computation, synthesis and optimization). It concentrates on the specific issues at the interfaces, and is divided into two main parts. The first part examines mainly theoretical issues and focuses on the modeling, validation and design techniques themselves. The second part illustrates the use of these methods in various design contexts at the forefront of new technology and architectural developments."

"This book defines the heterogeneous multicore architecture and explains in detail several embedded processor cores including CPU cores and special-purpose processor cores that achieve highly arithmetic-level parallelism. The authors developed three multicore chips (called RP-1, RP-2, and RP-X) according to the defined architecture with the introduced processor cores. The chip implementations, software environments, and applications running on the chips are also explained in the book.Provides readers an overview and practical discussion of heterogeneous multicore technologies from both a hardware and software point of view. Discusses a new, high-performance and energy efficient approach to designing SoCs for digitally converged, embedded systems. Covers hardware issues such as architecture and chip implementation, as well as software issues such as compilers, operating systems, and application programs. Describes three chips developed according to the defined heterogeneous multicore architecture, including chip implementations, software environments, and working applications."