Davis, M., personal communication.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....Once we were awarded the business, it took about 2 weeks to get a signed contract. This was the start of a great relationship with this customer. We delivered 30 days early with 0 defects. The main reason the customer went with us was because they felt comfortable with us every step of the way. [Davis] A remote site was experiencing database problems that were interfering with providing certain services to the customer. All attempts to recreate the problem using our interactive simulation tool or the prototype failed. We had an entire week s database changes shipped to us. Using the same load ....

Davis, M., personal communication.

....of the I O activity goes to the least active 75 of the file system [21] Others found that 66 of user files are not accessed for a month [42] which implies that all user activity was focused on at most 34 of the file system. Further evidence that most files are dormant is provided by Alsoft [30]. Their disk reorganization utility for the Apple MacIntosh, DiskExpress II, reorganizes the disk based on file activity statistics collected during normal system use. Alsoft found that roughly 80 of disk space is inactive even in a full file system. 2.1.2 Sequential Access Most file accesses ....

....activity. The negative speedups might be due to their method of reorganization, which separates logical blocks that the file system perceives as adjacent, so UNIX s block allocation strategy allocates adjacent logical blocks for a file but the physical blocks are widely separated. Disk Express II [30] monitors disk use on a per file basis and reorganizes the disk nightly based on the access history. It places the most active data close to cylinder 0 (the innermost track of the disk) and dormant data near the outer most cylinder, leaving all the free space in the middle. It places the active ....

L. Davis. Personal communication, Oct. 1990.

....in an image database [23] Scaling is one of several important issues in template matching. The others are local errors and rotation. However, the scaling problem is important even devoid of rotation as, for example, in the problem of reading a text with different sized letters (e.g. a newspaper) [15]. In reality, the scaling problem is intimately connected with approximation. Therefore, to be truly practical, such an algorithm should be able to cope with non discrete scales as well. However, there is no known fast algorithm even for scaled string matching with general scales. The problem of ....

Larry S. Davis. Personal Communication, 1989.

....RCC theory has shown that many interesting predicates can be defined once one takes the notion of a convex hull of a region (or equivalently, a predicate to test convexity) For example one can distinguish all the different kinds of shape displayed in figure 7. In fact, it has recently been shown [24], that this system essentially is equivalent to an affine geometry: any two compact planar shapes not related by an affine transformation can be distinguished. Fig. 7. All these shapes and many more can be distinguished using a convex hull primitive. Various different notions of the inside of a ....

E Davis. Personal communication., June 1996.

....applications. The revival of interest in self timed circuits can be attributed to the growing significance being attached to the above issues and the successful demonstration of asynchronous design and verification techniques by researchers like Sutherland [56] Martin [32] Dill [18] and Davis [17] among others. 1.1.2.3 Lower Power Dissipation As clock speed and system size increase, the portion of the system s power budget dedicated to distributing the clock increases dramatically. Self timed systems do not incur the power overhead of distributing a clock across the entire system. ....

....are indispensable in practical hardware design and are best implemented using an asynchronous design methodology. A solution based on synchronous hardware for these applications is awkward to design and implement and always runs the risk of metastability. Work by Nowick [43] and Davis et. al [17] has demonstrated that competitive asynchronous designs for some of these applications are possible. 5 1.2 Integrated CAD Framework Though asynchronous circuits have emerged as a promising area for research in the past few years, most of the work has been restricted to the development of ....

Davis, A., 1991. Personal communication.

....were they not batched. Since we are only using the Winstone benchmark to impose load, we exploit this timecompression to collect data for a shorter period of time. Data as to how long a typical user would take to execute the Business Winstone 97 benchmark input sequence are unavailable [23], but we can derive a conservative lower bound to the compression ratio under very weak assumptions. To do this we assume that a typing speed of 120 5 character words per minute (or about 1 character every 100 milliseconds) is the upper limit of sustainable human input speed. Based on the default ....

Ziff-Davis Corp., Winstone webmaster, personal communication, 1998.

....multiplications, and therefore to better exploit memory hierarchies. They perform symbolic factorization and generate supernodes on the fly as the factorization proceeds. UMFPACK is another competitive sequential code for this problem and neither SuperLU nor UMFPACK is always better than the other [3, 4, 7]. MA41 is a code for sparse matrices with symmetric patterns. All of them are regarded as of high quality and deliver excellent megaflop performance. In this paper we focus on the performance analysis and comparison with SuperLU code since the structure of our code is closer to that of SuperLU. In ....

T. Davis. Personal Communication, 1997.

....state. The majority classification task has been the subject of extensive study (Mitchell, Hraber, and Crutchfield 1993; Das, Mitchell, and Crutchfield 1994; Mitchell, Crutchfield, and Hraber 1994; Crutchfield and Mitchell 1995; Das, Crutchfield, Mitchell, and Hanson 1995; Mitchell 1996) Lawrence Davis (1995) cleverly modified the GKL rule and created a rule that achieved slightly better accuracy than the GKL rule. His rule has an accuracy of about 81.8 . In 1993, Rajarshi Das (1995) created another rule that achieved an accuracy of about 82.178 . Since several of these rules have not been previously ....

....Pentium type computer. Table 1 The succession of best cellular automata rules for the majority classification task. Rule State Transitions GKL 1978 00000000 01011111 00000000 01011111 00000000 01011111 00000000 01011111 00000000 01011111 11111111 01011111 00000000 01011111 11111111 01011111 Davis 1995 00000000 00101111 00000011 01011111 00000000 00011111 11001111 00011111 00000000 00101111 11111100 01011111 00000000 00011111 11111111 00011111 Das 1995 00000111 00000000 00000111 11111111 00001111 00000000 00001111 11111111 00001111 00000000 00000111 11111111 00001111 00110001 00001111 ....

[Article contains additional citation context not shown here]

Davis, Lawrence. 1995. Personal communication.

....functions to capture common subexpressions. Table 1 Out of sample performance of various rules for the majority classification task. Rule Accuracy Test cases Best rule evolved by bit string genetic algorithm (Das, Mitchell, and Crutchfield 1994) 76.9 10 6 GKL 1978 human written 81.6 10 6 Davis 1995 human written 81.8 10 6 Das 1995 human written 82.178 10 7 Best rule evolved by genetic programming (in this paper) 82.326 10 The two genetically evolved consensus motifs detect the two families either as well as, or slightly better than, the comparable human written motifs found in the ....

.... majority classification task has been the subject of extensive study (Mitchell, Hraber, and Crutchfield 1993; Das, Mitchell, and Crutchfield 1994; Mitchell, Crutchfield, and Hraber 1994; Crutchfield and Mitchell 1995; Das, Crutchfield, Mitchell, and Hanson 1995; Mitchell 1996) In 1993, Lawrence Davis (1995) cleverly modified the GKL rule and created a rule that achieved an accuracy of about 81.8 . Rajarshi Das(1995) created another clever rule that achieved slightly better accuracy than the GKL rule and the Davis rule. The Das rule has an accuracy of about 82.178 . Table 2 shows the GKL rule, the ....

[Article contains additional citation context not shown here]

Davis, Lawrence. 1995. Personal communication.

....other remedies have also been proposed. 2. Reordering code: Optimizing and parallelizing compilers use many techniques to improve performance, and some of these could be used to generate code variations. For example, a) Basic blocks: Rearrange the basic blocks of compiled code in random order [1]. This would cause instructions to be stored in different locations but would not affect the order in which they are executed, although there would likely be some long jumps executed instead of short jumps. This method could potentially disrupt some viruses. However, most file infector viruses ....

A. L. Davis, 1996. personal communication.

....and therefore to better exploit memory hierarchies. SuperLU performs symbolic factorization and generates supernodes on the fly as the factorization proceeds. UMFPACK is another competitive sequential code for this problem and neither SuperLU nor UMFPACK is always better than the other [3, 4, 7]. MA41 is a code for sparse matrices with symmetric patterns. All of them are regarded as of high quality and deliver excellent megaflop performance. In this paper we focus on the performance analysis and comparison with SuperLU code since the structure of our code is closer to that of SuperLU. In ....

T. Davis. Personal Communication, 1997.

....An earlier version of the following outline proof of the RCC axiom set s undecidability was constructed shortly before the author s discovery of (Grzegorczyk 1951) This contains a proof of undecideability for a number of geometric theories. One of these, as Ernest Davis has pointed out (Davis 1996), can readily be adapted to give a proof of RCC s undecidability. It is nevertheless worth including the following outline proof, as the spherical model of the RCC axioms is particularly suitable for the purpose, and much of the necessary definitional work has been done in preceding sections. To ....

Davis, E.: 1996, Personal communication.

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