Browsing by Author "Wu, PC"

Now showing 1 - 10 of 10

Case studies on cache performance and optimization of programs with unit strides
(1997-02-01)
Wu, PC
;
Huang, KC
;
交大名義發表
;
資訊工程學系
;
National Chiao Tung University
Cache performance in modern computers is important for program efficiency. A cache is thrashing if a significant amount of time is spent moving data between the memory and the cache. This paper presents two cache thrashing examples, one in scientific computing and one in image processing, both of which involve several one-dimensional arrays that are accessed sequentially, i.e., with unit strides. Accessing arrays in unit strides was considered very efficient on cache-based computer systems. However, the existence of cache thrashing is demonstrated by significant increases in computing speed in the equivalent programs tuned for cache locality. This shows that accessing several arrays sequentially may cause cache thrashing. Thus, to improve cache performance, it is important that the compiler or the programmer takes all arrays inside a loop into consideration. (C) 1997 by John Wiley & Sons, Ltd.
Highly efficient red phosphorescent osmium(II) complexes for OLED applications
(2004-07-19)
Tung, YL
;
Wu, PC
;
Liu, CS
;
Chi, Y
;
Yu, JK
The nonionic, red-emitting complexes [Os(fppz)(2)L-2] (L = PPh2Me (1), PPhMe2 (2)) and [Os(bptz)(2)L-2] (L = PPh2Me (3)) were synthesized, showing highly intense red phosphorescent emission in CH2Cl2 solution at lambda(max) 617, 632, and 649 nm, respectively. The electroluminescent properties of these compounds on OLEDs showed promising device efficiencies required for future OLED applications.
Interplay between intra- and interligand charge transfer with variation of the axial N-heterocyclic ligand in osmium(II) pyridylpyrazolate complexes: Extensive color tuning by phosphorescent solvatochromism
(2005-10-21)
Li, SW
;
Cheng, YM
;
Yeh, YS
;
Hsu, CC
;
Chou, PT
The rational design and syntheses of a new series of Os-II complexes with formula [Os(fppz)(2)(CO)(L)] (1: L=4-dimethylaminopyridine; 2: L=pyridine; 3: L = 4,4'bipyridine; 4: L=pyridazine; 5: L = 4-cyanopyridine), bearing two (2-pyridyl)pyrazolate ligands (fppz) together with one carbonyl and one N-heterocyclic ligand at the axial positions are reported. Single-crystal X-ray diffraction studies of, for example, 2 reveal a distorted octahedral geometry in which both fppz ligands reside in the equatorial plane with a trans configuration and adopt a bent arrangement at the metal center with a dihedral angle of similar to 23 degrees, while the carbonyl and pyridine ligands are located at the axial positions. Variation of the axial N-heterocyclic ligand leads to remarkable changes in the photophysical properties as the energy gap and hence the phosphorescence peak wavelength can be tuned. For complexes 1 and 2 the solvent-polarity-independent phosphorescence originates from a combination of intraligand (3)pi-pi* ((ILCT)-I-3) and metal-to-ligand charge transfer transitions ((MLCT)-M-3). In sharp contrast, as supported by cyclic voltammetry measurements and theoretical calculations, complexes 3-5 exhibit mainly ligand-to-lilgand charge transfer (LLCT) transitions, resulting in a large dipolar change. The phosphorescence of complexes 3-5 thus exhibits a strong dependence on the polarity of the solvent, being shifted for example, from 560 (in C6H12) to 665 nm (in CH3CN) and from 603 (in C6H12) to 710 nm (in CH3CN) for complexes 3 and 5, respectively. The results clearly demonstrate that a simple, straightforward derivatization of the axial N-heterocyclic ligand drastically alters the excitation properties per se from intraligand charge transfer (ILCT) to LLCT transitions. The latter exhibit remarkable LLCT phosphorescence solvatochromism so that a broad range of color tunability can be achieved.
Iterators: Taxonomy, strength, weakness, and efficiency
(2003-09-01)
Wu, PC
;
Wang, FJ
;
資訊工程學系
;
Department of Computer Science
Iteration is an operation that traverses the objects of a collection object. Iterators are operators/methods widely used for iteration on a collection of objects. However, iterators were considered to be ad hoc and to be a sign of weakness in object-oriented programming languages. There is a need to formalize iterators and to study their strength and weaknesses. In this paper we first propose a taxonomy of iterators based on the number of data states and the number of control points. The taxonomy identifies four categories of iterators. The simplest one can be implemented straightforwardly in an object-oriented language. Iterators of other categories need transformations and are more difficult to design and implement. Finally, we compare the efficiency of iterators with other implementations of iteration. The results show that all iteration techniques tested, including iterators, have been well optimized and are almost as fast as directly coded loops using recent C++ compilers.
Locating free positions in LR(k) grammars
(2002-05-01)
Ouyang, ST
;
Wu, PC
;
Wang, FJ
;
資訊工程學系
;
Department of Computer Science
LR(k) is the most general category of linear-time parsing. Before a symbol is recognized in LR parsing, it is difficult to invoke the semantic action associated with the symbol. Adding semantic actions to an LR(k) grammar may result in a non-LR(k) grammar. There are two straightforward approaches adopted by practitioners of parser generators. The first approach is to delay all semantic actions until the whole parse tree is constructed. The second is to add semantic actions to the grammar by chance. This paper presents an efficient algorithm for finding positions (called free positions) that can freely put semantic actions into an LR(k) grammar. The speedups of our method range from 2.23 to 15.50 times for the eight tested grammars.
Multiplicative, congruential random-number generators with multiplier +/-2(k1)+/-2(k2) and modulus 2(p)-1
(1997-06-01)
Wu, PC
;
交大名義發表
;
National Chiao Tung University
The demand for random numbers in scientific applications is increasing. However, the most widely used multiplicative, congruential random-number generators with modulus 2(31) - 1 have a cycle length of about 2.1 X 10(9). Moreover, developing portable and efficient generators with a larger modulus such as 2(61) - 1 is more difficult than those with modulus 2(31) - 1. This article presents the development of multiplicative, congruential generators with modulus m = 2(p) - 1 and four forms of multipliers: 2(k1) - 2(k2), 2(k1) + 2(k2), m - 2(k1) + 2(k2), and m - 2(k1) - 2(k2), k1 > k2. The multipliers for modulus 2(31) - 1 and 2(61) - 1 are measured by spectral tests, and the best ones are presented. The generators with these multipliers are portable and very fast. They have also passed several empirical tests, including the frequency test, the run test, and the maximum-of-t test.
On efficiency and optimization of C++ programs
(1996-04-01)
Wu, PC
;
Wang, FJ
;
交大名義發表
;
資訊工程學系
;
National Chiao Tung University
The efficiency of object-oriented programs has become a point of great interest, One necessary factor for program efficiency is the optimization techniques involved. This paper presents the performance of several variations of a given C++ program and compares them with a version that uses no object-oriented features, Our result indicates that some object-oriented features in C++ are not well optimized in current C++ compilers. We thus discuss some code optimization techniques that can improve the efficiency based on the given C++ program.
Parallelizing a level 3 BLAS library for LAN-connected workstations
(1996-10-10)
Huang, KC
;
Wang, FJ
;
Wu, PC
;
交大名義發表
;
資訊工程學系
LAN-connected workstations are a heterogeneous environment, where each workstation provides time-varying computing power, and thus dynamic load balancing mechanisms are necessary for parallel applications to run efficiently. Parallel basic linear algebra subprograms (BLAS) have recently shown promise as a means of taking advantage of parallel computing in solving scientific problems. Most existing parallel algorithms of BLAS are designed for conventional parallel computers; they do not take the particular characteristics of LAN-connected workstations into consideration. This paper presents a parallelizing method of Level 3 BLAS for LAN-connected workstations. The parallelizing method makes dynamic load balancing through column-blocking data distribution. The experiment results indicate that this dynamic load balancing mechanism really leads to a more efficient parallel level 3 BLAS for LAN-connected workstations. (C) 1996 Academic Press, Inc.
Parallelizing a level 3 BLAS library for LAN-connected workstations
(1996-10-10)
Huang, KC
;
Wang, FJ
;
Wu, PC
;
資訊工程學系
;
Department of Computer Science
LAN-connected workstations are a heterogeneous environment, where each workstation provides time-varying computing power, and thus dynamic load balancing mechanisms are necessary for parallel applications to run efficiently. Parallel basic linear algebra subprograms (BLAS) have recently shown promise as a means of taking advantage of parallel computing in solving scientific problems. Most existing parallel algorithms of BLAS are designed for conventional parallel computers; they do not take the particular characteristics of LAN-connected workstations into consideration. This paper presents a parallelizing method of Level 3 BLAS for LAN-connected workstations. The parallelizing method makes dynamic load balancing through column-blocking data distribution. The experiment results indicate that this dynamic load balancing mechanism really leads to a more efficient parallel level 3 BLAS for LAN-connected workstations. (C) 1996 Academic Press, Inc.
Phosphorescence of red Os(fptz)(2)(PPh2Me)(2) doped organic light-emitting devices with n and p hosts
(2006-02-06)
Liu, TH
;
Hsu, SF
;
Ho, MH
;
Liao, CH
;
Wu, YS
We have applied a stericly-hindered red phosphorescent dopant Os(fptz)(2)(PPh2Me)(2) [fptz=3-trifluoromethyl-5-(2-pyridyl)-1,2,4-triazole, PPh2Me=phosphine ligand] in p-type 4,4(')-N,N-'-dicarbazole-biphenyl or n-type bis(2-methyl-8-quinolinolato)(p-phenylphenolato) aluminum host and found that the latter produced higher luminance efficiency at lower doping concentration. We present a model to rationalize this phenomenon in which the n-type host impedes hole transport, which leads to narrower recombination zone near the hole transport layer/emission layer interface than the p-type host, hence, more effective recombination.