a verilog programmed Multiply unit algorithm
標(biāo)簽: programmed algorithm Multiply verilog
上傳時(shí)間: 2017-07-19
上傳用戶(hù):zhangyigenius
verilog Multiply algorithm
標(biāo)簽: algorithm Multiply verilog
上傳時(shí)間: 2017-07-19
上傳用戶(hù):xiaoyunyun
紋理映射在計(jì)算機(jī)圖形計(jì)算中屬于光柵化階段,處理的是像素,主要的特點(diǎn)是數(shù)據(jù)的吞吐量大,對(duì)實(shí)時(shí)系統(tǒng)來(lái)說(shuō)轉(zhuǎn)換的速度是一個(gè)關(guān)鍵的因素,人們尋求各種加速算法來(lái)提高運(yùn)算速度。傳統(tǒng)的方法是用更快的處理器,并行算法或?qū)S糜布kS著數(shù)字技術(shù)的發(fā)展,尤其是可編程邏輯門(mén)陣列(FPGAs)的發(fā)展,提供了一種新的加速方法。FPGAs在密度和性能上都有突破性的發(fā)展,當(dāng)前的FPGA芯片已經(jīng)能夠運(yùn)算各種圖形算法,而在速度上與專(zhuān)用的圖形卡硬件相同。因此,F(xiàn)PGA芯片非常適合這項(xiàng)工作。 本文主要工作包括以下幾個(gè)方面: 1、本文提出了一種MIPmapping紋理映射優(yōu)化方法,改進(jìn)了MIPmapping映射細(xì)化層次算法及紋理圖像的存儲(chǔ)方式,減少紋理尋址的計(jì)算量,提高紋理存儲(chǔ)的相關(guān)性。詳細(xì)內(nèi)容請(qǐng)閱讀第三章。 2、提出了一種MIPmapping紋理映射優(yōu)化方法的硬件實(shí)現(xiàn)方案,該方案針對(duì)移動(dòng)設(shè)備對(duì)功耗和面積的要求,以及分辨率不高的特點(diǎn),在參數(shù)空間到紋理地址的計(jì)算中用定點(diǎn)數(shù)來(lái)實(shí)現(xiàn)。詳細(xì)內(nèi)容請(qǐng)閱讀第四章。 3、實(shí)現(xiàn)了紋理映射流水線(xiàn)單元紋理地址產(chǎn)生電路,及紋理濾波電路的FPGA設(shè)計(jì),并給出設(shè)計(jì)的綜合和仿真結(jié)果。詳細(xì)內(nèi)容請(qǐng)閱讀第五章4、實(shí)現(xiàn)了符合IEEE 754單精度標(biāo)準(zhǔn)的乘法、乘累加及除法運(yùn)算器電路。乘法器采用改進(jìn)型Booth編碼電路以減少部分積數(shù)量,用Wallace對(duì)部分積進(jìn)行壓縮;乘累加器采用Multiply-add fused算法,對(duì)關(guān)鍵路徑進(jìn)行了優(yōu)化;除法器為基于改進(jìn)型泰勒級(jí)數(shù)展開(kāi)的查找表結(jié)構(gòu)實(shí)現(xiàn),查找表尺寸只有208字節(jié),電路為固定時(shí)延,在電路尺寸、延時(shí)及復(fù)雜度方面進(jìn)行了較好的平衡。
上傳時(shí)間: 2013-04-24
上傳用戶(hù):yxvideo
本章將介紹μ’nSP™系列單片機(jī)的應(yīng)用領(lǐng)域,具體講述SPCE061A單片機(jī)在通訊、語(yǔ)音領(lǐng)域里的應(yīng)用,并詳細(xì)給出了有關(guān)系統(tǒng)的電路原理圖、程序流程圖以及程序代碼,供讀者參考。 μ’nSP™家族產(chǎn)品具有電源電壓范圍和工作速率范圍較寬、集成度高、性能價(jià)格比高以及功耗低等特點(diǎn),故其有非常廣泛的應(yīng)用領(lǐng)域。μ’nSP™家族系列產(chǎn)品,涵蓋了非常廣泛的應(yīng)用。包括:發(fā)音與語(yǔ)音識(shí)別的微控制器(SPCE系列)、通信來(lái)電辯識(shí)應(yīng)用的微控制器(SPT660x系列)、以及通用型微控制器等等,主要體現(xiàn)在以下幾個(gè)方面: 用于數(shù)字信號(hào)處理 用于開(kāi)發(fā)研制便攜式移動(dòng)終端 用于開(kāi)發(fā)嵌入式計(jì)算機(jī)應(yīng)用系統(tǒng) 用于數(shù)字信號(hào)處理1. 數(shù)字濾波器 (Digital Filter)數(shù)字濾波器是一種計(jì)算處理或算法。借助于此,可以將輸入的一種數(shù)字信號(hào)或序列變換為另一種序列輸出。數(shù)字濾波器已被廣泛地應(yīng)用于數(shù)字語(yǔ)音、數(shù)字圖像處理以及模式識(shí)別和頻譜分析。數(shù)字信號(hào)處理器(DSP,Digital Signal Processor)的作用是通過(guò)一系列數(shù)字來(lái)表示信號(hào)及其信息,并借助數(shù)字計(jì)算方法變換和處理這些信號(hào)。為了構(gòu)成DSP,必須有一種部件能夠快速地完成兩個(gè)數(shù)值的乘法運(yùn)算并將乘積累加于寄存器。“快速”意味著乘和累加(MAC,Multiply & ACcumulate)較高的運(yùn)算速度。若以16位數(shù)值進(jìn)行乘和累加,其結(jié)果應(yīng)為32位。顯然,μ’nSP™的硬件結(jié)構(gòu)與其指令系統(tǒng)的結(jié)合足以構(gòu)成DSP應(yīng)用的硬件MAC單元,因而很適用于一些DSP方面的應(yīng)用。
標(biāo)簽: nSP 單片機(jī)應(yīng)用 開(kāi)發(fā)技術(shù)
上傳時(shí)間: 2014-01-26
上傳用戶(hù):qb1993225
一種基于二維鏈表的稀疏矩陣模半板類(lèi)設(shè)計(jì) A template Class of sparse matrix. Key technology: bin,2-m linked matrix. constructors: 1.normal constuctor 2.copy constuctor. 3.assignment constructor. Basic operator: 1. addition(sub) of two matrix 2. inverse of a matrix. 3. Multiply of two matrix. etc.
標(biāo)簽: matrix technology template linked
上傳時(shí)間: 2013-12-13
上傳用戶(hù):lwwhust
Matrix Transposition and Multiplication It is a MIPS assembly program that does the following: given two matrices, M1 and M2, first transpose M2 to obtain M2tran. Then Multiply M1 and M2tran.
標(biāo)簽: Multiplication Transposition following assembly
上傳時(shí)間: 2016-05-03
上傳用戶(hù):kernaling
(1) 設(shè)計(jì)和編寫(xiě)代表矩陣的Matrix類(lèi)。該類(lèi)包括矩陣行列數(shù)變量int rows和int cols,矩陣數(shù)據(jù)數(shù)組double data[][],構(gòu)造方法Matrix()、Matrix(int rows,int cols)、Matrix(int rows,int cols,double data[][]),獲取某元素值的方法getData(int row,int col),設(shè)置某元素值的方法setData(int row,int col,double value),計(jì)算兩個(gè)矩陣的乘積的方法Multiply(Matrix m)以及toString()等內(nèi)容。
標(biāo)簽: int Matrix cols rows
上傳時(shí)間: 2016-08-19
上傳用戶(hù):qiao8960
cordic methods describe essentially the same algorithm that with suitably chosen inputs can be used to calculate a whole range of scientific functions including sin, cos, tan, arctan, arcsin, arccos, sinh, cosh, tanh, arctanh, log, exp, square root and even Multiply and divide. the method dates back to volder [1959], and due to its versatility and compactness, it made possible the microcoding of the hp35 pocket scientific calculator in 1972. here is some code to illustrate the techniques. ive split the methods into three parts linear, circular and hyperbolic. in the hp35 microcode these would be unified into one function (for space reasons). because the linear mode can perform Multiply and divide, you only need add/subtract and shift to complete the implementation. you can select in the code whether to do the multiples and divides also by cordic means. other multiplies and divides are all powers of 2 (these dont count). to eliminate these too, would involve ieee hackery.
標(biāo)簽: essentially algorithm describe suitably
上傳時(shí)間: 2017-03-02
上傳用戶(hù):litianchu
karatsuba multiplication using vectors. O(n^l.6). Base 10. To change to higher bases like 10000 you have to change the Multiply order.
標(biāo)簽: multiplication karatsuba vectors change
上傳時(shí)間: 2014-01-25
上傳用戶(hù):han_zh
Fast Fourier Transform power point The rectangular window introduces broadening of any frequency components [`smearing鈥? and sidelobesthat may overlap with other frequency components [`leakage鈥?. 鈥he effect improves as Nincreases 鈥owever, the rectangle window has poor properties and better choices of wncan lead to better spectral properties [less leakage, in particular] 鈥搃.e. instead of just truncating the summation, we can pre-Multiply by a suitable window function wnthat has better frequency domain properties. 鈥ore on window design in the filter design section of the course
標(biāo)簽: rectangular introduces broadening Transform
上傳時(shí)間: 2017-03-25
上傳用戶(hù):change0329
蟲(chóng)蟲(chóng)下載站版權(quán)所有 京ICP備2021023401號(hào)-1