Hardware Reference
In-Depth Information
Tabl e 6. 5
Execution time of bioassays on Chips 1
4
Biochip
Chip 1
Chip2(s)
Chip 3
Chip4(s)
Bioassay
Multiplexed bioassay
63 s
65
NA
62
PCR bioassay
NA
22
NA
23
Protein bioassay
195 s
221
180 s
217
Tabl e 6. 6
Execution time of bioassays on Chips A
D
Biochip
Chip A
Chip B (s)
Chip C
Chip D (s)
Bioassay
Multiplexed bioassay
70 s
77
NA
73
PCR bioassay
NA
20
NA
20
Protein bioassay
264 s
441
220 s
223
in [
9
] are referred to as Chips A
D; and these four biochips derived by ILP-based
pin-count aware design in [
18
] are referred to as Chips E
G.
By simulating multiplexed bioassay, PCR bioassay, and protein dilution bioassay
on Chips 1
4, we can evaluate the flexility of the proposed general-purpose biochip.
The completion times of these bioassays can be found in Table
6.5
.Noteeven
though the Chips 1
4 are general-purpose biochips, their applications are still limited
by their available input/output ports. For example, Chip 1 has only 6 input/output
ports, while the PCR bioassay requires at least 8 input ports (because it has 8
different input reagents). Hence the PCR bioassay cannot be executed on Chip 1,
and the corresponding execution time is not available (written as “NA” in the table).
From the simulation results presented in Table
6.5
, we can find that the proposed
general-purpose biochip can be adapted to different bioassays with only a slight
increase in the bioassay completion times. Therefore, the proposed general-purpose
biochip offers high flexibility when running various bioassays on a dedicated layout.
Although Chips A
D and Chips E
G are bioassay-specific biochip, we still
can “force” them to execute various bioassays by applying the scheduling algorithm
proposed in Sect.
6.6
. The execution times for running multiplexed bioassay, PCR
bioassay, and protein dilution bioassay on Chips A
D can be found in Table
6.6
.
The execution times for running multiplexed bioassay, PCR bioassay, and protein
dilution bioassay on Chips E
G can be found in Table
6.7
.
By comparing the execution time in Tables
6.5
,
6.6
and
6.7
, we can find that
Chips 1
4 can achieve shorter execution time comparing with Chips A
D and
Chips E
G in most scenarios. This is because due to deadlocks, some operations
that are implemented in parallel on Chips 1
4 have to be scheduled for sequential
implementation on Chips A
D and Chips E
G. Therefore, compared with the
bioassay-specific biochip in [
9
], the proposed design has higher flexibility when
running various bioassays on the same layout. Compared with the proposed general-
purpose biochip, the ILP-based design method can reduce the pin-count by 33.3 %
to 69.4 %. We have noted that there is a trade-off between pin-count and flexibility
