FPGA Implementation of a Pipelined Gaussian Calculation for HMM-Based Large Vocabulary Speech Recognition

Table 3

Method B. Device utilization on virtex 5 SX95T FPGA.

	Single core	2 cores	4 cores
Occupied Slices	8,601(58%)	9,117(61%)	9,516(64%)
Slice LUTs	18,618(31%)	19,673(33%)	21,288(36%)
Slice Registers	14,212(24%)	15,597(26%)	16,895(28%)
BlockRam (Kb)	288(3%)	540(6%)	1,044(11%)
DSP48es	8(1%)	14(2%)	26(4%)