### Table 6. Average Throughput Time for Asynchronous Requests in Minutes for the Networking Case

"... In PAGE 7: ...Table 6. Average Throughput Time for Asynchronous Requests in Minutes for the Networking Case The advantage of the separation strategy for presales requests can also be identified for asynchronous requests (see Table6 ). In all analyzed situations the average throughput time for presales requests is lower when the communication centers are completely separated.... ..."

### Table 3: Additional axioms for asynchronous data ow networks

"... In PAGE 14: ... Although the former interpretation seems more close to asynchronous data ow than the latter interpretation, both are found in asynchronous data ow. In Table3 , axioms for the additional constants ^m, m, _m and m are given. We consider the axioms in Table 3 desired axioms for asynchronous data ow networks.... In PAGE 14: ... In Table 3, axioms for the additional constants ^m, m, _m and m are given. We consider the axioms in Table3 desired axioms for asynchronous data ow networks. They are all valid in the process algebra model described below, but not in some other models.... In PAGE 16: ... Analogously, the constants In and mXn in AProc(D) are the instances of the ones de ned on Proc(D) for sd1 1 as wire. For n = 1, the additional constants in AProc(D) are de ned as follows: Name Notation split ^1 2 AProc(D)(1; 2) sink 1 2 AProc(D)(1; 0) merge _1 2 AProc(D)(2; 1) dummy source 1 2 AProc(D)(0; 1) asynchronous copy ^1 2 AProc(D)(1; 2) asynchronous equality test _1 2 AProc(D)(2; 1) De nition ^1 = I1 (1; 2; split1) I2 where split1 = (er1(x) ; (s1(x) + s2(x))) 1 = I1 (1; 0; sink1) where sink1 = (er1(x) ; ) _1 = I2 (2; 1; merge1) I1 where merge1 = ((er1(x) + er2(x)) ; s1(x)) 1 = (0; 1; source1) I1 where source1 = ^1 = I1 (1; 2; acopy1) I2 where acopy1 = (er1(x) ; (s1(x) k s2(x))) _1 = I2 (2; 1; aeq1) I1 where aeq1 = ((er1(x1) k er2(x2)) ; s1(x1) x1 = x2 s1(p )) For n 6 = 1, the constants split, sink, merge and dummy source are de ned by the equations occurring as axioms A12{A19 in Table3 and the constants copy and equality test are de ned in the same way as split and merge, respectively. 2 In order to be fully precise, we have to adapt the de nitions given in Section 4.... ..."

### Table 3. Required cycles for network resource allocation problem asynchronous min-conflict only asynchronous

### Table I. The Main Kernel Functions for Supporting Application-Level Networking Transmit net xmit Asynchronously transmit a packet on a given network interface Demux dpf insert Insert a filter and associate a packet ring with it dpf delete Dereference a filter

2002

Cited by 24

### Table 2 shows the latency result when applying network traffic between pairs of processors with all other PEs idle. Each class of paths had nearly identical results independent of the source and destination PEs. The table contains an example path for that network class. The latency through the network is effectively equal to the number of buffering elements, either routers or elastic buffers. This is as expected because the network is clocked. In an asynchronous version the delays will be much more dependent upon topology and the complexity of the buffering element.

2007

Cited by 2

### Table I gives the qualitative comparison of asynchronous wakeup mechanisms under the symmetric and asymmet- ric communication models. As the symmetric communi- cation model is more in line with the any-to-any commu- nication paradigm in ad hoc networks, in the next two sec- tions, we only focus on designing and evaluating protocols for asynchronous wakeup mechanisms under the symmet- ric communication model.

### Table 2: Scenario yielding stable performance using asynchronous message passing primitive.

1994

"... In PAGE 5: ... On the other hand, the asynchronous communication primitive does nothave this instabilitybecause rollbackcosts arerelativelycon- stant, based on our assumption that invocation of the asynchronous sendprimitive is not time consuming. The same scenariois depicted in Table2 , where it is seen that the instability has been eliminated. Because network transmission delays do not affect the amount of time the sender is blocked when using asynchronous sends, one can expect asynchronousmessage passing primitives will be much more efficient in Time Warp, particularly in distributed computing envi- ronments with substantial transmission delays.... ..."

Cited by 18

### Table 1. (a) Node functions, (b) Genotype segment for one node. At the start of the experiment, each node was assigned a real-valued propa- gation delay, selected uniformly randomly from the range 1.0 to 5.0 nanoseconds, and held to double precision accuracy. These delays were to be the input-output delays of the nodes during the entire experiment, no matter which functions the nodes performed. There were no delays on the interconnections. To commence a simulation of a network apos;s behaviour, all of the outputs were set to logic zero. From that moment onwards, a standard asynchronous event-based logic simu- lation was performed [24], with real-valued time being held to double precision accuracy. An equivalent time-slicing simulation would have had a time-slice of 10?24 seconds, so the underlying synchrony of the simulating computer was only manifest at a time-scale 15 orders of magnitude smaller than the node delays, allowing the asynchronous dynamics of the network to be seen in the simula- tion. A low-pass lter mechanism meant that pulses shorter than 0.5ns never happened anywhere in the network.

1996

"... In PAGE 17: ...boolean functions of Table1 (a) was instantiated by each node, and how the nodes were connected. The nodes were analogous to the recon gurable logic blocks of an FPGA, but an input could be connected to the output of any node without restriction.... In PAGE 17: ... The nodes were analogous to the recon gurable logic blocks of an FPGA, but an input could be connected to the output of any node without restriction. The linear bit-string genotype consisted of 101 segments (numbered 0::100 from left to right), each of which directly coded for the function of a node and the sources of its inputs, as shown in Table1 (b). (Node 0 was a special `ground apos; node, the output of which was always clamped at logic zero.... ..."

Cited by 1

### Table 9: The simulation results for N-bit parity check problem using BPSN with the asynchronous updating rule, K = N.

1992

"... In PAGE 30: ...or all cases, both APSN and BPSN could solve the problem perfectly (i.e. gave correct output for all 2N inputs), in a very short time. Table9 shows the results obtained using the BPSN when the type II learning algorithm with an asynchronous updating rule was used. It is observed that each network is able to map all the patterns correctly.... In PAGE 30: ... It is observed that each network is able to map all the patterns correctly. As seen from Table9 , a smaller learning rate led to slower training for all cases. Divergence in MSE is observed if is more than 0.... In PAGE 30: ...ivergence in MSE is observed if is more than 0.5 for N = 2; 3 and 4. For N = 5, 0:01 caused divergence. It is observed from Table9 that the variances of the results are large. This is also observed for the other two problems.... ..."

Cited by 4