Optimal synthesis of algorithm-specific lower-dimensional processor arrays

Abstract

Unclassified 1b. RESTRICTIVE MARKINGS None______ ______ 2a. SECURITY CLASSIFICATION AUTHORITY 2b. DECLASSIFICATION /DOWNGRADING SCHEDULE 3. DISTRIBUTION/AVAILABILITY OF REPORT Approved for public release; distribution unlimited 4. PERFORMING ORGANIZATION REPORT NUMBER(S) U ILU-EN G-93-2^c_93_23 5. MONITORING ORGANIZATION REPORT NUMBER(S) 6a. NAME OF PERFORMING ORGANIZATION Coordinated Science Lab University of Illinois 6b. OFFICE SYMBOL (I f applicable) N/A 7a. NAME OF MONITORING ORGANIZATION National Science Foundation-MIPS Division Office of Naval Kggg 6c ADDRESS (G'ty, S taff, an d ZIP Code

An approach to checking link conflicts in the mapping of uniform dependence algorithms into lower dimensional processor arrays

“…There have been several attempts on the problem of checking the link conflicts [13], [14], [16], [17], [18], [19], [20]. One of these methods [13], [14] has to examine the whole index set of size yx Pn , where x is the problem size.…”

“…One of these methods [13], [14] has to examine the whole index set of size yx Pn , where x is the problem size. Other methods [16], [17], [18], [19], [20] use the I/O spaces concept to check link conflicts. Each I/O space is associated with a data dependence vector d. An I/O space can be an input space or an output space and is defined as the set fi À d j i P t nd i À d T P tg and fi j i P t nd i d T P tg, respectively, where t is the index set.…”

An approach to designing modular extensible linear arrays for regular algorithms

“…The first one [2], [3], [4], [5] designed linear arrays for different problems case by case. The second category [6], [7], [8], [9], [10] proposed procedures to design systematically linear arrays for SUREs. On the design of modular extensible linear arrays, four types of conflict-free conditions should be satisfied.…”

A reindexing based approach towards mapping of DAG with affine schedules onto parallel embedded systems

“…As stated earlier, Wong and Delosme [47], Shang and Fortes [40], Ganapathy and Wah [21], have developed an approach based on integer programming to get the best projection direction possible. But for large problem instances the time spent to solve such programs cannot be neglected [9].…”

“…The main drawback of this allocation technique is the low PE utilization occurring in the resulting array [8]. Wong and Delosme [47], Shang and Fortes [40], Ganapathy and Wah [21] use integer programming to get the best linear allocation possible. However, such a linear allocation may not correspond to an optimal allocation.…”