### Table 1. Timing and area costs of digit-level LFSR multi- pliers in F397 for different values of digit-size D D Multiplication # of slices Maximum # of clock

708

"... In PAGE 3: ... Our approach is similar to [12]. In Table1 we show the results of implementing F397 multipliers on a XC2VP20-6FF896 FPGA. In this table the first column is the digit-size D.... ..."

### Table 1. Descriptive characteristics of the samples (n=31)

"... In PAGE 7: ...887. Findings Table1 summarizes descriptive statistics regarding blog types, authors, industry, topics, and blog features. The most popular format of the corporate blogs was a blog run by multiple authors.... ..."

### Table 9 R and D activities (percentages of valid responses)*

2002

"... In PAGE 15: ... The new and changed functions of R and D capacities were also registered in the sample. The most important positive mes- sage in Table9 is that subcontractors and other companies alike did much more R and D in 1999 than in 1996, or at least the fre- quency of such activities increased consid- erably. The figures supported the hypothesis: subcontractors did much less basic and applied research but were deeply involved in product and tech- nology development and in changing pro- duction lines (test production and re- tooling).... ..."

### Table 8: Individually best matching standards of behavior

"... In PAGE 24: ... RAP enjoyed great support for the risk scenario and somewhat less support for the ignorance scenario. Indeed, in Table8 , RAP was roughly on par with EU for the ignorance scenario, while Table 7 conveys the impression that it was signi cantly inferior. Again, it seems that RAP disposed of a partisan group, and met distinctly less sympathy outside this group.... ..."

### Table 3: Field operations in point multiplication

2004

"... In PAGE 4: ... We use the Montgomery algorithm (with projective coordinates) described in [17] to implement point multiplication, which is the fastest method that does not require significant pre-computations and/or storage. We first use gprof to profile the point multiplication operation and examine how it decomposes into the field arithmetic operations ( Table3 ). On average, squaring takes 6.... In PAGE 4: ...ultiplication 87.25%. The time shown as other is the execution overhead, which primarily includes the main control loop (which iterates over the point multiplication function). The time per point multiplication in Table3 (3318 us) differs from Table 1 (3218 us) because execution with profiling slightly degrades performance. 2.... ..."

Cited by 3

### Table 3: Number of multiplications for (univariate) multiple draws

"... In PAGE 15: ...moothing algorithms. To sample from p( jy) we use Algorithm 2 in x2 4. The method JS is di erent in the sense that it rst samples from p( jy) and then computes draws for t using the second equation of (1). Table3 presents the numbers of multiplications required for multiple draws of univariate (p = 1) state space models with di erent state vector dimensions. It is... ..."

### lable has multiple candidates. Such multiple can-

### TABLE I ASYMPTOTIC RELAXATION RECOVERY VOLTAGES

1990

Cited by 3

### Table 1. Matrix multiplication

1991

"... In PAGE 4: ... The performance ofthe conventionalmatrix multiplicationalgorithmon vectormachines is not a smooth function of n, but peaks at points when n is a multiple of the vectorregister length, drops immediately afterwards, and then increases again to the next multiple of the vector register length. For the CRAYY-MPthere is a 14%drop at n = 64, an 8% drop at n = 128,and a 4% drop at n = 256 ( Table1 ).All measurementsin Table 1were made on a CRAYY-MPin multiuser mode.... In PAGE 4: ...All measurementsin Table 1were made on a CRAYY-MPin multiuser mode. The performance in Table1 was obtained by using an assembly coded matrix multiplication subroUtineprovidedby CrayResearch in SCILIB [Cray Research, Inc.... In PAGE 4: ... For the Strassen algorithm, with Q = 64, we expect to see an increase in MFLOPS at each level of recursion due to the reduced number of operations (ignoring n2 terms). At n = 130,however,the Strassen algorithm would require sevenmatrix multiplications with n = 65,and these multiplicationswouldbe performed atthe lowrateof about244 MFLOPS compared with 269 MFLOPS using the conventionalalgorithm for n = 130( Table1 ).The lower performance would cancel out the gain in reduction in the number of operations (Figure 1).... ..."

Cited by 36