Results 1 - 10
of
26,306
Table 4: Local memory of B after CHOLESKY
"... In PAGE 12: ... Thus the local memorys of the processes contain L in the same columnwise wrapped fashion as previously B. This is demonstrated in Table4 for our exemplarily treated matrices and four processes. The technique of receiving and immediately sending the column `i in connection with the wrapped mapping of the matrix leads to a good load balancing and thus we get a satisfying e ciency as shown in Table 5.... ..."
TABLE II Memory access bandwidth and processing power for the local memory solution
Table 4: Synthesis results for the nine subtask system with local memory.
"... In PAGE 6: ... The generated designs have the same cost and performance as those pre- sented in [18] since the B amp;B algorithm assures that the optimal solution is generated. Table4 presents synthesis results of the same task data- flow graph but with the cost function extended with the local memory cost as defined in [19]. Our approach can han- dle different heterogeneous constraints, such as subtask execution times on different processors, communication times constraints and finally local memory cost, and still provide optimal solutions in a reasonable time.... ..."
Table 3. The bandwidth and powerreductions obtained by our Local Memory Customization Algorithm.
2001
"... In PAGE 6: ... For madd, the hit ratio is substantially improved, due to the fact that it contains a large number of stream accesses, and can benefit from the inclusion of the spatial cache to han- dle them, while keeping variables without spatial locality in the temporal cache. Table3 presents the bandwidth and power decrease gener- ated by our approach. The second and third columns show the bandwidth and memory/interconnect power for the traditional cache architecture.... ..."
Cited by 10
Table 1. Summary of local memories selected for the different benchmarks.
2006
Cited by 2
Table 1: Local-memory performance. Feature SPP1000
"... In PAGE 4: ...Figure 5: Cache misses in the transition region (C lt; w lt; C + C=A) for array size w, and an A-way associative cache of size C. All the information shown in Table1 is derived only from these simple kernels and resulting graphs. The cache load and store transfer rates are calculated from the min- imum call time in the hit region.... ..."
Table 5: Effect of locality in shared memory version.
1989
"... In PAGE 8: ....3.2 Locality in the Shared Memory Approach This section examines the sensitivity of the shared memory approach to locality. Table5 shows the solution quality and amount of traffic generated, as a function of the amount of locality exploited in the application. Table 5: Effect of locality in shared memory version.... ..."
Cited by 21
Table 5: Effect of locality in shared memory version.
1989
"... In PAGE 8: ....3.2 Locality in the Shared Memory Approach This section examines the sensitivity of the shared memory approach to locality. Table5 shows the solution quality and amount of traffic generated, as a function of the amount of locality exploited in the application. Table 5: Effect of locality in shared memory version.... ..."
Cited by 21
Results 1 - 10
of
26,306