"... In PAGE 7: ... Table1 . Example study tasks.... ..."

"... In PAGE 4: ... Stability Stress Levelling Sharpening Activeness Passiveness Subtlety Boldness Representation Abstraction Realism Distortion Flatness Depth Table1 . Graphical representations of visual techniques.... ..."

"... In PAGE 9: ... As in the previous example, spatial transforma- tions are a crucial part of how he uses the visualization to resolve the problem. Similarly, Table3 shows an instance of a forecaster needing to compare visual chunks that are not visible on any current display. In this situation, the forecaster is at-... ..."

"... In PAGE 3: ... But since these commands do not rely on the stimuli generated by a certain test bench, they can be used for system exploration as well. Table1 assem- bles a list of visualized high-level debugging commands. Examining commands vlsb Visualize the specified channel and all connected modules.... In PAGE 4: ... Example debug session The following two commands illustrate the provided visualized debugging functionality exemplarily. The vlsb command ( Table1 ) visualizes the specified channel and all connected modules in the cone view of RTLVision. In case of a failure related to a specific sig- nal the user gets a quick overview about all its connec- tions.... In PAGE 4: ... Figure 4 sketches the visualization output after call- ing vlsb with two signals of the RISC-CPU design in order to check their bindings to the right ports: (gdb) vlsb quot;ram_cs quot; (gdb) vlsb quot;next_pc quot; Figure 4. Debug command vlsb The vtrace_at command ( Table1 ) is a typical repre- sentative of the monitoring command type. It traces the given signal or port and records the actual value at the specified simulation time stamp.... In PAGE 5: ... Debug command vtrace_at Table 2 underlines the efficiency of our non-intrusive, patch-free approach using library interposition (Sec- tion 4.1) while illustrating the performance of the vtrace command ( Table1 ). So, the observation of 750000 data sets over 125 signals leads to a slow down of factor 4 com- pared to a trace-free simulation while the tracing of 50 signals increases the simulation time about 80%.... In PAGE 5: ... So, we assume that the defect has to be searched in the controlling of the ALU where the ALU is implemented by the module instance IEU. To get the right control sig- nal the vlsio_rx command ( Table1 ) is applied at first. We suppose that the name of the attached control port includes the string code: (gdb) vlsio_rx quot;IEU quot; quot;code quot; Using the path fragment navigation feature in RTLVi- sion shows subsequently that the only port reported by vlsio_rx is connected to the signal alu_op (Figure 6).... In PAGE 5: ... Consequently, we initiate a monitoring of the two interesting signals using the vtrace command (Ta- ble 1) and continue simulation: (gdb) vtrace quot;program_counter quot; 110000 (gdb) vtrace quot;alu_op quot; 110000 (gdb) c After simulation has stopped we investigate the traced behavior. To focus the error search onto the relevant de- sign parts only, the vlsb command ( Table1 ) is applied... ..."

"... In PAGE 4: ... The frame rate is 30 Hz, as in HDTV (note that PAL uses a 25 Hz rate), and the target bitrate is 38400. Table2 and Table 3 summarize our MPEG-4 visual encoding and decoding experiments, respectively. The numbers for instruction cache and TLB misses are neg- ligible, and are omitted.... In PAGE 6: ... Table 6 and Table 7 present the numbers for three visual objects with two visual object layers each. Table 4, Table 5, Table 6, and Table 7 with Table2 and Table 3 show that cache performance does not change noticeably as the number of VOs and VOLs increases. Figure 3 and Figure 4 visually depict part of this compar- ison, showing L1C and L2C data miss rates on an R10K machine with an L2 cache of 2MB (note the different or- ders of magnitude in the y axis scales).... ..."

"... In PAGE 6: ... Surprisingly, we find this not to be the case. Table4 and Table 5 contain statistics for processing three visual objects. Each of the three objects is encoded and decoded for the same con- figuration as in the single object experiments, with the single-object input becoming a subset of the multiple- object input.... In PAGE 6: ... Table 6 and Table 7 present the numbers for three visual objects with two visual object layers each. Table4 , Table 5, Table 6, and Table 7 with Table 2 and Table 3 show that cache performance does not change noticeably as the number of VOs and VOLs increases. Figure 3 and Figure 4 visually depict part of this compar- ison, showing L1C and L2C data miss rates on an R10K machine with an L2 cache of 2MB (note the different or- ders of magnitude in the y axis scales).... ..."

"... In PAGE 6: ... Average JNDs and PSEs for the three visual conditions in [Ellis et al. 2004] are reported in Table1 . The statistical significance of any differences between visual conditions was investigated by ANalysis Of Variance (ANOVA) [Coolican 1999].... ..."

"... In PAGE 6: ... Each of the three objects is encoded and decoded for the same con- figuration as in the single object experiments, with the single-object input becoming a subset of the multiple- object input. Table6 and Table 7 present the numbers for three visual objects with two visual object layers each. Table 4, Table 5, Table 6, and Table 7 with Table 2 and Table 3 show that cache performance does not change noticeably as the number of VOs and VOLs increases.... In PAGE 6: ... Table 6 and Table 7 present the numbers for three visual objects with two visual object layers each. Table 4, Table 5, Table6 , and Table 7 with Table 2 and Table 3 show that cache performance does not change noticeably as the number of VOs and VOLs increases. Figure 3 and Figure 4 visually depict part of this compar- ison, showing L1C and L2C data miss rates on an R10K machine with an L2 cache of 2MB (note the different or- ders of magnitude in the y axis scales).... ..."