Numerical Aerospace Simulation Facility. NAS parallel benchmarks, 1997. http://www.nas.nasa.gov/Software/NPB/.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....for small reductions, but the cost of the reduction itself becomes overwhelming for a large number of nodes. 4.3. Scalability of Conjugate Gradient The matrix sizes we have used are based upon the problem sizes associated with the distribution of the NAS Conjugate Gradient (CG) benchmark [19]. We have used the Class W (n = 7,000) A (14,000) and B (75,000) problem sizes for our experiments. The total numbers of non zeroes in the matrices are 508,402, 1,853,104, and 13,708,072, respectively. The scalability of the class W problem on the 4 different EARTH configurations is shown in ....

Numerical Aerospace Simulation Facility. NAS parallel benchmarks, 1997. In http:// www.nas.nasa.gov/Software/NPB/.

....is more than 3 years old, and current Alpha processors have clock speeds around 600MHz. However Alpha based workstations tend to command a premium price, and are 5 generally not competitive with Pentium PCs in terms of price performance. 5. 2 NAS Parallel Benchmarks The NAS Parallel Benchmarks [18] are a well known suite of several application kernels that are indicative of some of the large scale scientific and engineering computations done at NASA. They are commonly used to benchmark the performance of parallel high performance computers. The eight kernels are written in Fortran or C ....

Numerical Aerospace Simulation Facility, "NAS Parallel Benchmarks", http://science.nas.nasa.gov/Software/NPB/.

....of the most recent offerings from Intel, and still significantly in front of the PowerPC. However Alpha based workstations tend to command a premium price, and are generally not competitive with Pentium PCs in terms of price performance. 5. 2 NAS Parallel Benchmarks The NAS Parallel Benchmarks [26] are a well known suite of several application kernels that are indicative of some of the large scale scientific and engineering computations done at NASA. They are commonly used to benchmark the performance of parallel high performance computers. The eight kernels are written in MPI Fortran and ....

Numerical Aerospace Simulation Facility, "NAS Parallel Benchmarks", http://science.nas.nasa.gov/Software/NPB/.

....what features would be worth optimising. To this end, we have developed some new benchmarks to augment the Java Grande Forum Benchmark Suite [6] with some additional low level benchmarks and several scientific application kernels, including Java implementations of some of the NAS benchmarks [15] that are commonly used for benchmarking high performance computers for scientific applications. We have also undertaken a systematic analysis of these new benchmarks, as well as the existing Java Grande benchmarks, on serial platforms including Pentium II (Linux, Solaris and NT) Celeron (NT) ....

....generates pseudo random numbers with a Gaussian probability distribution. ffl NAS IS (integer sorting) benchmark does key sorting. ffl NAS FT (Fast Fourier Transform) is a commonly used scientific computation that requires floating point arithmetic and data shuffling. The NAS Parallel Benchmarks [15] are a well known suite of several application kernels that are indicative of some of the large scale scientific and engineering computations done at NASA. They have mainly been used to benchmark the performance of parallel high performance computers for these kinds of applications. It is quite an ....

Numerical Aerospace Simulation Facility, NAS Parallel Benchmarks, http://science.nas.nasa.gov/Software/NPB/.

....matrices which are multiplied. mm kes shows both DAG and loop parallelism. mm is a standard triply nested matrix multiply, which parallelizes well. The argument is the length of a side of the square matrices which are multiplied. nas multigrid is the 3D multigrid benchmark from the NAS suite [59], which parallelizes well. The first argument is the number of levels that the multigrid uses, and the second argument is the number of iterations of performed. The size of the grid at maximum resolution is 64 Theta 64 Theta 64. poly mult performs a polynomial multiplication, where the ....

Numerical Aerospace Simulation Facility. NAS Parallel Benchmarks. Web Site. http://science.nas.nasa.gov/Software/NPB/.

No context found.

Numerical Aerospace Simulation Facility. NAS parallel benchmarks, 1997. http://www.nas.nasa.gov/Software/NPB/.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC