### Table III. Comparison of implementation strategies

### Table 4: Translation of Fortran77 BLAS routines to Fortran90 BLAS routines Algorithms BLAS Fortran90 facilities

1992

"... In PAGE 30: ... The Fortran77 BLAS routines have to be substituted by the appropriate Fortran90 vector operations and/or standard functions. Table4 shows the translation of BLAS routines to correspond Fortran90 routines. For DM-MIMD machines, the implementation is not straightforward and pipelining is introduced to get a better performance.... ..."

Cited by 29

### Table 4: Translation of Fortran77 BLAS routines to Fortran90 BLAS routines Algorithms BLAS Fortran90 facilities

"... In PAGE 22: ... The Fortran77 BLAS routines have to be substituted by the appropriate Fortran90 vector operations and/or standard functions. Table4 shows the translation of BLAS routines to correspond Fortran90 routines. For DM-MIMD machines, the implementation is not straightforward and pipelining is introduced to get a better performance.... ..."

### Table 4: Translation of Fortran77 BLAS routines to Fortran90 BLAS routines Algorithms BLAS Fortran90 facilities

"... In PAGE 29: ... The Fortran77 BLAS routines have to be substituted by the appropriate Fortran90 vector operations and/or standard functions. Table4 shows the translation of BLAS routines to correspond Fortran90 routines. For DM-MIMD machines, the implementation is not straightforward and pipelining is introduced to get a better performance.... ..."

### Table 4: Translation of Fortran77 BLAS routines to Fortran90 BLAS routines

"... In PAGE 22: ... The Fortran77 BLAS routines have to be substituted by the appropriate Fortran90 vector operations and/or standard functions. Table4 shows the translation of BLAS routines to correspond Fortran90 routines. For DM-MIMD machines, the implementation is not straightforward and pipelining is introduced to get a better performance.... ..."

### Table 5: Implementation of Fortran 90 Features not in Subset.

"... In PAGE 38: ...7.3 Support for Fortran 90 not in Subset HPF Table5 details some desirable facets of Fortran 90 that were excluded from the Subset. Features were excluded from the Subset to ensure timely appearance of products in the market place, however, some of these features are highly desirable.... In PAGE 75: ...91 74.6 Table5 0: Execution of IS with BLOCK distribution over linear chain of 1; 2; 4; 8 and 16 processors on a SPARCCentre 2000 and a Power Challenge XL. Number of Processors 1 2 4 8 APR xHPF PGI PVM 69.... In PAGE 75: ...8 363.3 Table5 1: Execution of IS with CYCLIC distribution over linear chain of 1; 2; 4; 8 and 16 processors on a SPARCCentre 2000. buckets(i) = buckets(i) + buckets(i-1) ENDDO ! .... ..."

### Table 4. FORTRAN Parameters

2004

"... In PAGE 11: ... In the FORTRAN 77 reference implementation of the PDNET main program, available in le fdriver.f, we initially set the maximum problem and workspace sizes by specifying the FORTRAN compile-time parameters in Table4 . As pre- sented in Tables 3 and 5, the external arrays passed to the PDNET core routines must be statically dimensioned using the FORTRAN parameters for the maximum number of nodes and arcs.... ..."

Cited by 1

### Table 2. Comparison of GPUs

2006

Cited by 1

### Table 1. Examples of GPU implementations Problem Type Authors year Processing engine Implementation GPUvsCPU

"... In PAGE 1: ... However these rates should not be compared directly because of the different architectures and instruction sets of the CPU and GPU. Table1 gives some small sample of the problems solved on GPUs and performance comparisons between GPU and CPU. Second, the GPU acts as a co-processor, therefore it can free up the CPU to perform other tasks [3].... ..."

### TABLE 1. Comparison between Matlab and Fortran based implementations.

1999

"... In PAGE 35: ...0. TABLE1 5: Contents of SLICOT: user-callable routines. Routine Function AB01MD Orthogonal controllability form for single-input system.... In PAGE 36: ... TABLE1 5: (continued) Routine Function AB06MD Minimal block Hessenberg realization for a state-space rep- resentation. AB07MD Dual of a given state-space representation.... In PAGE 37: ... TABLE1 5: (continued) Routine Function MB03MD Computation of an upper bound using a bisection method such that a bidiagonal matrix has precisely ` singular values greater than or equal to plus a given tolerance. MB03ND Computation of the number of singular values of a bidiago- nal matrix which are smaller than or equal to a given value .... In PAGE 38: ... TABLE1 5: (continued) Routine Function MC01ND Computation of the value of the real polynomial P(x) at a given complex point x = x0 using Horner apos;s algorithm. MC01OD Computation of the coe cients of a complex polynomial from its zeros.... In PAGE 39: ... TABLE1 5: (continued) Routine Function SB02OD Solution of either the continuous-time algebraic Riccati equation Q+AT X+XA?XBR?1BT X = 0 or the discrete- time algebraic Riccati equation X = AT XA ? AT XB(R + BT XB)?1BT XA + Q using the method of de ating sub- spaces, where Q = CT C, R = DT D and CT D = 0. SB03MD Solution of either the continuous-time Lyapunov equation AT X + XA = C or the discrete-time Lyapunov equation AT XA ? X = C using Bartels/Stewart or Barraud apos;s meth- ods, respectively.... In PAGE 40: ... TABLE1 5: (continued) Routine Function TB01ND Upper/lower observer Hessenberg form. TB01QD Transfer matrix of a state-space representation.... ..."

Cited by 53