"... In PAGE 15: ...75 MHz and 2V. A detailed breakdown of the energy estimates at the algorithm and architecture levels is given in Table5 . As described in section 3.... In PAGE 16: ... Note that the analysis of this effect can only be performed using the DBT-based architectural power analyzer - an estimator based on the uniform white noise model would not be able to make this distinction. The results of assignment for locality on the cascade are also shown in Table5 . In this case, the overall energy is reduced an additional 22% with no penalty in area or performance.... ..."

"... In PAGE 15: ...75 MHz and 2V. A detailed breakdown of the energy estimates at the algorithm and architecture levels is given in Table5 . As described in section 3.... In PAGE 16: ... Note that the analysis of this effect can only be performed using the DBT-based architectural power analyzer - an estimator based on the uniform white noise model would not be able to make this distinction. The results of assignment for locality on the cascade are also shown in Table5 . In this case, the overall energy is reduced an additional 22% with no penalty in area or performance.... ..."

"... In PAGE 77: ... The energy comparisons are then based on combining these energy per access values with the table activity. Relevant data structures, their config- uration, and the energy per port access are presented in Table5 . One important thing to note is the energy effi- ciency of the filter cache and trace cache compared to the hybrid value predictor tables.... In PAGE 77: ... Without embedded filtering data, the trace cache is only written when a unique instruction trace is constructed. With filter data, the trace cache entries are larger and the Table5 : Energy Per Port Access Structure Entries Ports Assoc Data Energy LastVP/StrideVP 8192 4 1 8B 5.80 nJ Context VP 1024 4 1 8B 2.... ..."

"... In PAGE 1: ...able 1. Truth table of 1-qudit ternary reversible gates. Input Output +1 +2 12 01 02 0 1 2 1 2 0 2 0 1 0 2 1 1 0 2 2 1 0 If two 1-qudit gates x and y in cascade have the resultant effect that the input signal to the gate x is restored at the output of the gate y, then the gate y is said to be the inverse gate of the gate x. Table2 shows the inverse gates of all the 1-qudit gates. Table 2.... ..."

"... In PAGE 8: ... At last, RA is performed on all the survivals for rescoring. Table1 lists the eight fllters and the flnal rescorer used in the system. The idea of the cascade fllters comes from Viola [14] who used about 20 level fllters... In PAGE 8: ... The idea of the cascade fllters comes from Viola [14] who used about 20 level fllters to detect human faces. In fact, each fllter in Table1... ..."

"... In PAGE 9: ... A full sensitivity study is beyond the scope of this paper but, for comparison, an inversion was performed for case waa using observed data from the array closest to the sound source ( 1 km) using both the full band of frequencies (FB) and using the low frequencies only (LF) 25-200 Hz in 1 Hz increments. In Table4 the di erence is given as a percent of the ground truth value for the 5kmarrayand 1kmarray for both the full band and low frequencies only. As the present environment is truly range independent it is expected that most infor- mation, and thus best results, about the bottom parameters is obtained at low frequencies using receivers close to the source, as more energy penetrates into the bottom 22 .... In PAGE 11: ... The agreementisvery good. results in Table4 imply that for both objective functions using only the low frequencies and array position at 1 km gives the best results in terms of percentage error. Individual param- eters may be estimated better using other con gurations of array position and frequency content.... In PAGE 13: ... The results in both Fig. 4 and Table4 indicate that for the benchmark cases, one receiver is su cient if a coherent processor in frequency is used. The broadband single hydrophone inversion results demonstrate that a few receivers may be su cient to determine ocean bottom properties.... ..."

"... In PAGE 3: ... Wu and S. Ishikawa, using inversion potentials, are given in Table2 . This calculation follows their recent article [7].... ..."