"... In PAGE 8: ... Thus, they suffer frequent squashes in line-based protocols. The number of depen- dences in Bdna and Track is shown in Table4 of [5]. That table counts dependences to the same word and to different words of a line assuming per-word disambiguation support.... In PAGE 8: ... In multi-version line-based protocols, all these RAWs and WAWs (both same-word and false) cause squashes if they occur out of order. In fact, some of the false WAWs in Table4 of [5] will also appear as RAW violations due to the inability to disambiguate at word level. 5 Evaluation 5.... In PAGE 8: ... The results are shown in Table 4. Based on Table4 , we can place each application on a qualitative 3-D chart (Figure 6). Each dimension of the chart shows whether or not an optimization is expected to be beneficial.... ..."

"... In PAGE 8: ... Thus, they suffer frequent squashes in line-based protocols. The number of depen- dences in Bdna and Track is shown in Table4 of [5]. That table counts dependences to the same word and to different words of a line assuming per-word disambiguation support.... In PAGE 8: ... In multi-version line-based protocols, all these RAWs and WAWs (both same-word and false) cause squashes if they occur out of order. In fact, some of the false WAWs in Table4 of [5] will also appear as RAW violations due to the inability to disambiguate at word level. 5 Evaluation 5.... In PAGE 8: ... The results are shown in Table 4. Based on Table4 , we can place each application on a qualitative 3-D chart (Figure 6). Each dimension of the chart shows whether or not an optimization is expected to be beneficial.... ..."

"... In PAGE 8: ... Thus, they suffer frequent squashes in line-based protocols. The number of depen- dences in Bdna and Track is shown in Table4 of [5]. That table counts dependences to the same word and to different words of a line assuming per-word disambiguation support.... In PAGE 8: ... In multi-version line-based protocols, all these RAWs and WAWs (both same-word and false) cause squashes if they occur out of order. In fact, some of the false WAWs in Table4 of [5] will also appear as RAW violations due to the inability to disambiguate at word level. 5 Evaluation 5.... In PAGE 8: ... The results are shown in Table 4. Based on Table4 , we can place each application on a qualitative 3-D chart (Figure 6). Each dimension of the chart shows whether or not an optimization is expected to be beneficial.... ..."

"... In PAGE 8: ... Thus, they suffer frequent squashes in line-based protocols. The number of depen- dences in Bdna and Track is shown in Table4 of [5]. That table counts dependences to the same word and to different words of a line assuming per-word disambiguation support.... In PAGE 8: ... In multi-version line-based protocols, all these RAWs and WAWs (both same-word and false) cause squashes if they occur out of order. In fact, some of the false WAWs in Table4 of [5] will also appear as RAW violations due to the inability to disambiguate at word level. 5 Evaluation 5.... In PAGE 8: ... The results are shown in Table 4. Based on Table4 , we can place each application on a qualitative 3-D chart (Figure 6). Each dimension of the chart shows whether or not an optimization is expected to be beneficial.... ..."

"... In PAGE 2: ... For example, adding an application-specific instruction to a micro- processor/controller by simple changes to the behavioral description, followed by behavioral synthesis, can provide significant benefits. Table1 shows execution cycle-count and code-size reductions provided by two application- specific instructions added to a Motorola HC11 CPU de- sign described in [10]. Assigning a new opcode to a control-intensive application code such as hex-to-binary conversion routine results in 76% reduction in the number of clock cycles to execute.... ..."

"... In PAGE 7: ... This leads us to the next step in the value chain: Publishing and Aggregation . Value Added In Table1 the value added by the different application development companies are shown. ... ..."

"... In PAGE 5: ... The control unit also monitors the recovery time and when it exceeds some threshold, it resorts to program restart solutions. 4 Preliminary Results We have investigated some real, buggy server programs, listed in Table2 . Our analysis shows that these software failures can be dynamically survived by our methods.... ..."

"... In PAGE 5: ... The control unit also monitors the recovery time and when it exceeds some threshold, it resorts to program restart solutions. 4 Preliminary Results We have investigated some real, buggy server programs, listed in Table2 . Our analysis shows that these software failures can be dynamically survived by our methods.... ..."