X-Git-Url: https://bilbo.iut-bm.univ-fcomte.fr/and/gitweb/book_gpu.git/blobdiff_plain/a60840091cd318f2c8e530740dd7940fc393e898..a2aa3f0f91a668ee6e799bad0f4de90b7b2be452:/BookGPU/Chapters/chapter16/ch16.tex diff --git a/BookGPU/Chapters/chapter16/ch16.tex b/BookGPU/Chapters/chapter16/ch16.tex index e7ba3ba..ac52b32 100644 --- a/BookGPU/Chapters/chapter16/ch16.tex +++ b/BookGPU/Chapters/chapter16/ch16.tex @@ -52,21 +52,22 @@ \input{Chapters/chapter16/gpu.tex} \input{Chapters/chapter16/exp.tex} +\clearpage \section{Summary} \label{sec:summary} -In this chapter, we present a new envelope-following method for +In this chapter, we have presented a new envelope-following method for transient analysis of switching power converters. First, the -computationally expensive step, the solving of Newton update equation, +computationally expensive step, the solving of the Newton update equation, has been parallelized on CUDA-enabled GPU platforms with iterative GMRES solver to boost performance of the analysis method. To further -speed up the GMRES solving for Newton update equation, we have +speed up the GMRES solving for the Newton update equation, we have employed the matrix-free Krylov basis generation technique. The proposed method also applies the more robust Gear-2 integration to compute the sensitivity matrix. Experimental results from several integrated on-chip power converters have shown that the proposed GPU envelope-following algorithm can lead to about 10$\times$ speedup compared to its CPU counterpart, and 100$\times$ faster than the -traditional envelope-following methods while still keeps the similar +traditional envelope-following methods while still keep the similar accuracy. @@ -74,7 +75,7 @@ accuracy. \begin{Glossary} \item[Envelope-Following] In transient simulation of switching power circuits, nodal voltage waveforms in neighboring high frequency clock cycles are similar, -but not exactly the duplicates. Envelope-following technique approximates +but not exactly duplicates. Envelope-following technique approximates the slowly changing transient trend over a lot of clock cycles without calculating waveforms in all cycles. \end{Glossary}