Using Standard Alliance Synthesis Flow

gate count               1717
number of cells           535
number of library cells   188
number of used cells       55
max fanin                   3
max input capacitance      35
max internal fanout        39
critical path  0fF       2193
critical path  6fF       2550

In this experiment we check three "standard" Alliance synthesis flows. A standard Alliance synthesis flow uses the same library for both BOOG and LOON, and not a special BOOG library with its own timing. One problem with this is that BOOG will crash with the vsclib library with the message

Library Error: No cell could match  ''1''

1. BOOG and LOON use the library timing without any increment to the input pin capacitance modelling a wireload. Synthesis is one BOOG and one LOON synthesis step. BOOG uses a 0f wireload, and the successive LOON steps use a 0f and a 6f wireload.

   export MBK_TARGET_LIB=../vsclib013_0_nobuf boog -x 0 -m 3 multi8_o export MBK_TARGET_LIB=../vsclib013_0 loon -x 0 -l loon1 multi8_o multi8_0 export MBK_TARGET_LIB=../vsclib013_6 loon -x 0 -l loon1 multi8_o multi8_6

   The loon1.lax file is found to give the best results. The input resistance is set to 1kΩ; the output load to 50f; and the synthesis priority to 1.

   ##loon1.lax #M{1} #I{ x(0):1000; ... y(7):1000; } #C{ r(15):50; ... r(0):50; }

2. BOOG uses the library timing with 6fF added to the input pin capacitance to model a wireload. LOON synthesis then uses the 0f and 6f wireload libraries.

   export MBK_TARGET_LIB=../vsclib013_6_nobuf boog -x 0 -m 3 multi8_o export MBK_TARGET_LIB=../vsclib013_6 loon -x 0 -l loon1 multi8_o multi8_6 export MBK_TARGET_LIB=../vsclib013_0 loon -x 0 -l loon1 multi8_o multi8_0

   For the 6f wireload, LOON synthesis with a priority of 4 gives the best result. For the 0f wireload, a priority of 1 is best.

3. BOOG uses the 0fF wireload library with no buffers. LOON uses both 0fF and 6fF wireload libraries alternately in a sequence to find the fastest speed. This mimics the flow used in the previous experiments. The difference is the starting library used for BOOG.

The three results are compared below with the final result obtained from the previous experiment.

The fastest speed comes from the flow previously developed, but at the cost of a higher gate count. For a simple flow of one LOON synthesis, the best results come when using BOOG with a 0fF wireload library.

The next experiment will be to remove the little used cells and see what impact this has on the performance and area.